Академический Документы
Профессиональный Документы
Культура Документы
Editors
What
Mathematics
Can Do for You
Essays and Tips from Japanese Industry
Leaders
What Mathematics Can Do for You
Yoshikazu Giga • Toshiyuki Kobayashi
Editors
What Mathematics
Can Do for You
Essays and Tips from Japanese
Industry Leaders
123
Editors
Yoshikazu Giga Toshiyuki Kobayashi
Professor Professor
Graduate School of Mathematical Sciences Graduate School of Mathematical Sciences
The University of Tokyo The University of Tokyo
and
Principal Investigator
Kavli IPMU
Japan is a tiny country that occupies only 0.25% of the world’s total land area.
However, this small country is the world’s third largest in economy: Japanese GDP
is roughly equivalent to the sum of any two major countries in Europe as of 2012.
This book is a first attempt to ask business leaders of top Japanese companies
for their thoughts on mathematics. The topics range from mathematical problems
in specific areas (e.g., exploration of natural resources, communication networks,
and finance, etc.) to mathematics literacy that would help a leader who has to
weigh many different issues, and even to mathematical thinking in connection to
quality control and a long-term strategy. We express our deep gratitude to these
business leaders who shared their thoughts on mathematics with us in spite of their
extremely busy schedules. The reader may notice that every article—sometimes
even the choice of vocabulary—reflects the authors’ ways of life and thinking.
This book is an enlarged English edition of the Japanese version, What Mathe-
matics Can Do for You—Essays and Tips from Japanese Industry Leaders. For this
edition we have invited three mathematicians who have been trying to expand and
strengthen the interaction between mathematics and industry.
Mathematics asks nothing in return. The role of mathematics is often invisible
when it is applied effectively and smoothly in science and technology, and math-
ematical strategy is usually hidden when it is used properly and successfully. It
is a pleasant surprise to us that the leaders in global companies appreciate this
invisible feature of mathematics. The editors hope this book will give the reader
an opportunity to notice something hidden but important.
This book owes much to the invaluable support of Mr. Toru Yosano, Senior
Advisor of BNP Paribas Securities (Japan) Limited, and of Springer Japan, from an
earlier stage of the project. We also thank our colleague Professor Shigeo Kusuoka
for his kind suggestions.
v
Contents
vii
viii Contents
Kobayashi (K): I believe that there are a variety of things that are fundamental but
rarely perceived, that are important but too obvious to attract our attention. Do you
think the value of honesty for a company is such a thing? I surmise that a sound
company does not have to think about “the value of honesty” so often because
people there just take it as a matter of course. On the other hand, dishonesty—
needless to say fraud, but also pretending to know everything without performing
investigations, which I also consider a kind of dishonesty—seems to result in a long-
term loss even though it may produce a large short-term gain.
Today I would like to talk with you, by highlighting things that we are usually
unaware of. Of course, there are two opposite cases for these things, namely, because
they are so natural that we take them for granted, or simply because they are not
important.
Mathematics might be the former case. The smoother mathematics is functioning
within society, the more people are unaware of it. Mathematics is thought of as
fundamental in science and technology. Nevertheless it might go well for the time
being even if we ignore mathematics and forget thinking about reason and logic.
But this would not be the right way if we hope to achieve long-term success.
Y. Giga and T. Kobayashi (eds.), What Mathematics Can Do for You: Essays and Tips 1
from Japanese Industry Leaders, DOI 10.1007/978-4-431-54346-6 1,
© Springer Japan 2013
2 F. Cho and T. Kobayashi
Math Is Fun
C: Until I was a sophomore in high school, I was more interested in judo and kendo
than classroom studies. I studied math very hard over the next 2 years. I had been
told by my father that the family’s economic circumstances did not allow me to go to
a private university. So I had to study eight subjects, including Math I and Math II.
But I liked math because math valued logical thinking.
My father also graduated from a law school, and he was a very logical-minded
man. When I was in junior high and high school, he used to talk me down logically
by asking the reason for my argument. When I said something off the top of my
head, I was often scolded, with him saying, “You have to think more logically.”
When I studied math, I realized how my father’s advice was valuable. In the second
year of high school, I decided to major in the humanities in college, but math was
still fun to me. Looking back, the relationship with my father had an influence on
my bent toward mathematics.
C: After graduating from the university, I started working for Toyota. Six years later,
I was told to do “Kaizen” (improvement of production processes), although what I
had majored in was law, not engineering. Kaizen is an engineering person’s job, I
thought. I was embarrassed and tried to persuade my boss to reconsider his decision,
but in vain. I have now been doing Kaizen for 15 years. Looking back on these 15
years, I find I learned many things, mainly two lessons.
Toyota Chairman vs. a Mathematician 3
The first lesson I learned was the fact that production processes consisted of many
numbers. It seems there is something related to what Prof. Kobayashi mentioned
previously with regard to the term “honesty”: If people in a factory do their jobs
based on numbers, there can be little room for cheating and fudging. To be based on
numbers is the same thing as to be based on honesty. Production planning—which
sort and how many vehicles will be produced, how much sales are expected, how
much the cost price will be, etc.—is totally based on numbers. Without numbers,
we can say nothing but subjective remarks like “That vehicle looks cool” or “This
one provides a comfortable ride.” When we have to develop a new car, we cannot
depend on such empty words. What we need is information that is expressed with
numbers, such as bulk weight, noise level, vibration level, etc. By “numbers,” I
am not talking about technical calculations that are conducted in laboratories. I am
talking about simple arithmetic, such as addition, subtraction, multiplication and
division. Numbers and arithmetic are what Toyota’s 300,000 employees share as the
fundamental communication tool.
C: The first lesson I learned from Kaizen is the importance of numbers, which are
objective tools for us rather than abstract ideas. On the other hand, the second lesson
I learned has something more fundamentally mathematical. Doing mathematics and
being logical seem to me to be rather the same thing.
“All things have their causes.” Everyone knows that. However, people in our
company do not stop here, but keep thinking that “All things have their causes, and
all causes also have their real causes.” They should track back from causes to causes
until they find real causes. To do so, they should ask themselves the question “why”
at least five times.
They should go to factories themselves, observe what’s going on there, and
use their own hands to measure and count. They should not make their tasks
self-contained. When there is a good business, there is a sound logic behind it.
Everything derives from logic and numbers, and nothing in the world will happen by
chance. So, I was always tracking back problems until I found effective measures.
I kept asking questions like “Why does the cost price keep rising?” “Why did
this machine break down?” and “Why are there so many defective parts?” After
thoroughly asking myself such questions, I was able to understand the real causes
of a particular problem.
One day in a factory, I encountered a strange case. The blade in a certain machine,
which was supposed to last at least 1 month, was dead in the first week. I guessed
the supplier of the blade might have mixed in the wrong ingredients, but it was not
the case. Then, why? What made the blade break so quickly? That the machine’s
blade broke once a week meant that one could observe the instance of a break if one
kept watching it for a week! We worked in rotation and kept watching it. We noticed
that the blade broke because iron scraps wound around it and exerted a strong force
4 F. Cho and T. Kobayashi
on it. Then, why did the scraps wind around the blade? We found many plausible
causes for this phenomenon. The solution we finally found was to attach a cutting
device to the machine, so that the produced scraps were made too short to wind
around the blade.
These experiences made me develop a habit of seeing things logically. If my
father were still living today, he would be a little proud of me.
K: I am delighted to know your “5-Whys” principle in the Toyota Company! As
a mathematician, I think that such a spirit is crucial also in creating a new theory of
mathematics. As an educator, this is what I am trying to emphasize in classrooms.
Though I usually teach graduate students majoring in mathematics, I like giving
elementary courses for undergraduates, in particular, for students majoring in the
humanities, law, or economics, when my schedule permits it. In these courses, I
advise students not to memorize individual mathematical formulas because what
is crucial is to understand them systematically. I feel it more important to make
an effort to understand the logic and the rationale and to see through something
universal. For this I suggest that they cultivate the habit during my semester course
in the following way: “Do not avoid the things that you do not understand,” “Try to
clarify, capture and pin down by your own language what you do not understand”
and “Keep thinking about it again and again.”
C: That’s a good idea!
K: Toyota’s 5-Whys seems to me to share a quite similar basis. A blind
acceptance of knowledge is a facile solution but may result in an obstruction to
making further progress. We often realize that we have understood nothing. Starting
from this, we work hard to try to understand something better, and eventually we
may reach the painful realization that there are even more things that we do not un-
derstand! I think this is tough but important training for thinking with concentration.
I would like to cultivate students by giving them the opportunities in my classes to
deeply contemplate rather than giving up quickly or just memorizing. In fact, this is
my primary motivation for giving undergraduate courses in mathematics.
C: I see your point. When studying math, we have to think, not memorize.
K: Exactly. Further, also in economics, I would be afraid an attitude to memorize
“laws” and “formulas” or even the “notions” would end up misusing them in the
real world. This would be dangerous. Instead there is another discipline not to
memorize but to analyze what is crucial by thinking and thinking about what is more
fundamental. It takes an enormous amount of time. In my opinion, deep thinking for
the universal truth is the essence of mathematics.
C: As a high school student, I was trying to solve trigonometry problems. They
were very difficult ones, and I did not even have an idea where to start. I had been
thinking about one of these problems for a whole day and finally found the answer.
After solving the first problem, however, I could solve the remaining five to seven
problems almost instantly. It seemed something inside me had changed after taking a
whole day to solve the first problem. It was unforgettably impressive to me. I learned
from this experience the importance of thinking things through.
K: The whole day that you spent by thinking with “I-don’t-sees” must be a really
special time to bring you the breakthrough afterwards. As a mathematician, I always
Toyota Chairman vs. a Mathematician 5
have many I-don’t-sees while attacking difficult problems that nobody in the world
knows how to solve, or trying to develop a new mathematical theory without existing
methods as if I were in complete darkness. No matter how uncomfortable I feel the
I-don’t-sees, I am still optimistic and want to grapple with them face to face,
C: It seems that mathematical research embraces quite many “whys”.
K: That’s the point! By that very reason, your 5-Whys motto is impressive to me.
It is a clear, nice and encouraging message that will be retained in the memory.
Fujio Cho was born in 1937. Following graduation from the University of Tokyo
with a Bachelor’s Degree in Law, he joined the Toyota Motor Corporation in 1960.
In 1988, he became a company director and was appointed CEO of Toyota Motor
Manufacturing, USA. After holding posts as executive director, senior executive
director and executive vice-president, he became the company CEO in 1999. He
was appointed as the company chairman in 2006 and also served as director of Sony
Corporation and Central Japan Railway Company. He is also the chairman of the
Japan Sports Association.
His awards include the Grand Cordon of the Order of the Rising Sun and the
Medal of Honor with Blue Ribbon.
The University of Tokyo was founded in 1877 by succeeding older schools from the
Edo government, applying the education systems and curriculums introduced from
the West after the restoration of Imperial power in 1868.
Among prominent people graduated from Japan’s oldest university, there are
seven Nobel laureates (Yasunari Kawabata, Reona Ezaki, Eisaku Sato, Kenzaburo
Oe, Masatoshi Koshiba, Yoichiro Nambu and Ei-ichi Negishi) and a Fields Medalist
(Kunihiko Kodaira).
6 F. Cho and T. Kobayashi
Yusuke Yasuda
Former Chief Country Representative
BNP Paribas/Tokyo
Definitions
If I invest X(0) amount of money at time 0, and collect it at time τ , I will receive
X(τ ) determined by the following equation:
where rτ is the spot interest rate for the period [0, τ ]. In addition, X(0) is the
principal which was initially invested, and ΔX ≡ X(τ ) − X(0) is the interest. In
actual business, τ can be a period of 1 day, 1 week, 1 month, 2 months, or even up
to 12 months. The spot interest rate used for the period of 1 day is referred to as
the overnight rate; that of the 1 week period is referred to as the 1-week rate; and
Y. Giga and T. Kobayashi (eds.), What Mathematics Can Do for You: Essays and Tips 7
from Japanese Industry Leaders, DOI 10.1007/978-4-431-54346-6 2,
© Springer Japan 2013
8 Y. Yasuda
so on. Also, I can reserve an interest rate, for example, at a future t, in which case
I promise to invest the amount of X(t) at time t and in return for this investment
at time of t + τ I will receive the amount of money determined by the following
equation:
X(t + τ ) = (1 + Rt→t+τ · τ )X(t). (2)
The interest rate Rt→t+τ for the future time period [t,t + τ ] can be called a “reserved
interest rate”.1 This reserved interest rate at t = 0 is the interest rate for the period
[0, τ ], therefore should equal to rτ , that is:
R0→τ = rτ . (3)
If I decide to invest the principal X(0) for the period [0, τ ] at the spot interest rate
R0→τ (= rτ ), and furthermore decide to reinvest the principal and interest at time τ
for the next period [τ , 2τ ] at the reserved interest rate of Rτ →2τ etc., I will receive
the aggregated interest and principal at time nτ (n is an integer greater than one) in
accordance with the following equation:
Of course, τ can be e.g., 0.5 years, and therefore the amount after 0.5 years, 1 year,
and 1.5 years will be:
Usually, the interest rate will be a positive number greater than zero, and therefore
X(0) < X(0.5y) < X(1y) < X(1.5y) < · · · < X(nτ ) < · · · . (5)
Thus the amount of money will continue to increase as the original principal and
interest will be reinvested over time. In other words, the value of money will
continue to increase with time. The Eq. (4) gives other important information,
namely, that X(0) and X(nτ ) are in proportion:
1 Inmany finance-related books, this is also referred to as forward rate or interest rate futures, but
for the sake of emphasizing the interest rate which is reserved now, I would like to refer to it as
“reserved interest rate”.
BNP Paribas/Tokyo 9
DFnτ is referred to as the discount factor at time nτ . This equation means that the
money at time nτ should be discounted by DFnτ to evaluate its present value. By (4)
and (6), DFnτ can be expressed as follows:
1
DFnτ ≡ . (7)
(1 + R0→τ · τ )(1 + Rτ →2τ · τ ) · · · (1 + R(n−1)τ →nτ · τ )
More practically, the discount factor at each time 0.5, 1, 1.5 years etc. can be
expressed as follows:
1
DF0.5y ≡ , (7)
(1 + R0→0.5y · 0.5y)
1
DF1y ≡ , (7)
(1 + R0→0.5y · 0.5y)(1 + R0.5y→1y · 0.5y)
1
DF1.5y ≡ , · · · . (7)
(1 + R0→0.5y · 0.5y)(1 + R0.5y→1y · 0.5y)(1 + R1y→1.5y · 0.5y)
This concept of the discount factor is very important in finance. For example, when
comparing two different investment schemes, in order to decide which is more
profitable, one cannot decide this by merely comparing the money at the end, but
one needs to match the time of investment. Therefore, we can use the discount factor
and calculate the present value of each investment. By (5) and (6), we can see:
1 > DF0.5y > DF1y > DF1.5y > · · · > DFny > · · · > 0. (8)
This is a universal rule in the world of finance. Please take note of this, as I will
explain in the latter part of the chapter.
Bootstrapping Method
In the case of investing longer than 1 year, it is common to receive interest in regular
intervals. For investments longer than 1 year, it is common to receive interest every
6 months using the following equation:
c1
X(0). (9)
2
The reason c1 is divided by 2 is because the amount of interest for 1 year is c1 X(0),
with half of this amount being paid in 6 months, and the remaining half paid in 1
year. The yield (in this case c1 ) is called annual yield. The amount of money being
received during the investment period is:
10 Y. Yasuda
c1
the interest in 6 months: X(0) and
2
c1
the principal and the interest in 1 year: X(0) + X(0).
2
Using the discount factor of DF0.5y and DF1y , we have
c1 c1
X(0)DF0.5y + X(0) + X(0) DF1y . (10)
2 2
This is the present value of the amount of money which will be received in the
future. In the financial world, this is regarded as the original investment. If c1 is too
small, the investment is not too attractive and no one will invest in it, and if it is
too large, everyone will rush to invest. Therefore there always exists an equilibrium
point, and it is fair to say that the equilibrium point is determined where the present
value of the amount received in the future is equivalent to the initial investment.
This assumption gives us the following equation:
c1 c1
X(0) = X(0)DF0.5y + X(0) + X(0) DF1y .
2 2
c1 c1
1= DF0.5y + 1 + DF1y . (11)
2 2
Referring back to the Eqs. (7) and (7) , we see that DF0.5y can be derived using
the spot interest rate R0→0.5y (= r0.5y ), and if the reserved interest rate R0.5y→1y is
determined, DF1y will be fixed. Using this in (11), we can calculate c1 . However, if
the value of c1 is fixed, we can substitute c1 and DF0.5y into (11) and calculate DF1y ,
and therefore the reserved interest rate R0.5y→1y is determined. In this way, we can
see that c1 and R0.5y→1y determine each other uniquely: if either of them is fixed, we
can compute the other.
In the next example, we will consider an investment of 1.5 years. If we assume
that the annual yield is c1.5 , in which case we receive interest every 6 months for the
period of 1.5 years, the amount of money received in this period will be:
c1.5
the interest in 6 months: X(0),
2
c1.5
the interest in 1 year: X(0), and
2
c1.5
the principal and the interest in 1.5 years: X(0) + X(0).
2
By the same method as before, it can be determined that
c1.5 c1.5 c1.5
1= DF0.5y + DF1y + 1 + DF1.5y . (12)
2 2 2
From this equation, we can see that c1.5 and R1y→1.5y also uniquely determine each
other. Similarly, we can determine the relationship between c2 and R1.5y→2y , and
between c2.5 and R2y→2.5y . This method is called the “bootstrapping method” since
BNP Paribas/Tokyo 11
we tie the relationship between the pairs of variables from the bottom to the top of
the boot.
In the financial industry, the equation for calculating investments longer than 1 year
comes from market convention. Let n be the number of years (n is an integer greater
than two), and let cn be the annual yield, with the assumption that you receive
interest every 6 months for the period of n years. In this case, if pn is the annual yield
if you receive interest every year for the same period, the general market convention
for calculating pn is given by:
cn 2
1 + pn = 1 + . (13)
2
This equation has been used for many years and has become the norm. People not
familiar with finance may think that interest received every year is simply double the
interest received every 6 months (i.e. 2× c2n ), but in the financial world, if the amount
of money received is the same, receiving it earlier is always beneficial. Therefore
the amount of 2 × c2n should be smaller than that of pn , in other words, cn < pn . If we
take (13) and solve for the right-hand side, we can easily confirm this. This equation
is an oversimplified version and therefore problematic. To confirm this, we should
try to solve for cn and pn using the bootstrapping method.
First, the annual yield cn in which you receive the interest every 6 months gives
the following equation:
cn cn cn cn cn
1= DF0.5y + DF1y + DF1.5y + · · · + DF(n−0.5)y + 1 + DFny . (14)
2 2 2 2 2
Solving for cn gives:
2(1 − DFny)
cn = . (15)
DF0.5y + DF1y + DF1.5y + · · · + DF(n−0.5)y + DFny
1 − DFny
pn = . (17)
DF1y + DF2y + · · · + DF(n−1)y + DFny
2
1+DF1y +· · ·+DF(n−2)y +DF(n−1)y 1+DF0.5y +· · ·+DF(n−1)y +DF(n−0.5)y
= .
DF1y +DF2y +· · ·+DF(n−1)y +DFny DF0.5y +DF1y +· · ·+DF(n−0.5)y +DFny
(18)
After a straightforward calculation, we find that (18) can be true only in the
following case:
1 DF0.5y DF(n−0.5)y
= = ··· = . (19)
DF0.5y DF1y DFny
And by substituting the equations of discount factors (7) , (7) , (7), . . . , we find
that
R0→0.5y = R0.5y→1y = · · · = R(n−0.5)y→ny . (20)
This means that the reserved interest rates will all be the same. However, this
contradicts what happens in the real world. Therefore we conclude that Eq. (13) is
not true. Despite all this, we continue to use this equation because it is both simple
and it is very close to being correct. Does this approximation come up with a close
answer every time we use it? I have had this question in mind for a long time, but in
all the books I have read on this topic, I have never found a satisfying answer. But
I finally realized that there is a universal inequality in the financial world, based on
(8), which I will discuss in the next section.
1 > DF0.5y > DF1y > DF1.5y > · · · > DFny > · · · > 0.
If you take this sequence and add every second term starting from DF0.5y and
compare it to the result of doing the same process starting from DF1y , you get
By adding the left-hand side of (21) to both sides of (21), you get
In the real world, cn is small enough to ensure that 1 − c2n > 0 is true, and hence we
can conclude
cn
pn < . (24)
cn
1−
2
Therefore, by combining (22) and (24), we get
cn
cn < p n < . (25)
cn
1−
2
If we take the true value of pn and compare it to the market convention in (13), the
difference of the two can be denoted as εn and satisfies:
c2n
p n = cn + + εn . (26)
4
Substituting this expression in (25), we get
⎞ ⎛
cn
⎜ +
2⎟
c2n 1 c2n
− < εn < ⎜⎝
⎟. (27)
4 4 cn ⎠
1−
2
The right-hand side of this equation is
14 Y. Yasuda
⎛ ⎞
cn
c2n ⎜ 1 + ⎟ c2 2
⎜ 2 ⎟ = n 1 + c + cn + · · · . (28)
4 ⎝ cn ⎠
n
4 2
1−
2
As |cn | 1, cn to the power of 3 or even higher is much smaller than c2n , we see
c2 c2
that (28) is almost equal to 4n . Therefore, we see that |εn |is of an order at most 4n .
By this calculation, I was able to understand that the market convention (13) is
very accurate since the inequality of (25) holds. However, this equation has no
relevance in a financial transaction, so it is difficult to discover this fact in normal
daily life. But it is still very important to know that (13) guarantees a certain degree
of accuracy. Therefore mathematics is a tool of thought essential to doing business
in finance.
Last Words
Equation (13) has been around since before computers were widely used in the
financial industry, and it is still being used today. Nowadays, when more complex
equations are used in the industry, the fact that this concept has survived is simply
remarkable. Behind all this, there exists this universal inequality (8),
1 > DF0.5y > DF1y > DF1.5y > · · · > DFny > · · · > 0
cn
and as a result, this inequality (25) cn < pn < will always be true. Without
cn
1−
2
mathematics, I would not have been able to understand all of this.
Acknowledgements The author would like to express special thanks to Mr. Hiraku Kusaka, who
is a former colleague at BNP Paribas and is currently working for Nomura Securities Co. Ltd.,
for his full support in preparation of the original manuscript in Japanese and its translation into
English.
range of products and services, mainly in the fields of Corporate and Investment
Banking.
Yusuke Yasuda, BNP Paribas’ former Chief Country Representative for Japan, was
born in Tokyo in 1947. He graduated from the University of Tokyo with B.D.
in Economics in 1970 and earned MBA from the Harvard Business School in 1978.
After working for Japan Airlines, he became Senior Representative of Compagnie
Financière de Paribas’ Japanese office in 1985. In 1989, he was assigned as Chief
Country Representative for Japan and General Manager of BNP Paribas Securities’
Tokyo branch.
In 2005, Yasuda was awarded the Chevalier de la Légion d’Honneur by the
French government. He retired from the company in January 2010.
Takeda Pharmaceutical Co., Ltd.
Yasuchika Hasegawa
President and CEO
Takeda Pharmaceutical Company
All companies face a wide spectrum of risks and uncertainties that may impact their
ability to achieve strategic and operational goals. Each and every day, executive
managers confront a myriad of issues that must be analyzed and resolved. Such
analysis is critical in making informed decisions about the future. No matter how
talented or experienced, no company president should make such decisions in
isolation. The organization of a business thus allows a company to allocate roles and
responsibilities and to delegate authority. Through this system and the integration of
specialist knowledge and skills, a company is best able to handle the issues it faces.
Experience is another key factor that can provide valuable insights on future
decisions. Such experience allows businesses to classify problems by type in support
of decision making and to document problems for convenient reference should they
re-emerge. Assessing the policies and outcomes of both successful and unsuccessful
decisions assists future decision making. We at Takeda practice such a policy as part
of our ongoing pursuit of operational excellence. Even so, corporate managers are
called on to deal with many new challenges on a routine basis.
While the above examines one aspect of corporate management, it would be
wrong to suggest that executives and managers simply embrace the qualitative intu-
ition derived from experience in reaching key decisions. It is here that the enormous
Y. Giga and T. Kobayashi (eds.), What Mathematics Can Do for You: Essays and Tips 17
from Japanese Industry Leaders, DOI 10.1007/978-4-431-54346-6 3,
© Springer Japan 2013
18 Y. Hasegawa
A key department in the management of our operations, the Corporate Strategy and
Planning Department uses sophisticated quantitative modeling. Long-range decision
making is essential to the survival and growth of every corporation. These decisions
must be based on both qualitative and quantitative analysis. In isolation, qualitative
analysis is short on clarity and specifics and has a tendency to be based on hunches
and intuition. It is here that numerical data come into play, primarily in the form of
a technique known as discounted cash flow (DCF) analysis.
DCF analysis is a valuation method that generates future free cash flow pro-
jections, which are then discounted at a fixed rate to arrive at a present value. This
technique is used by companies in all industries, and can be elaborated by deducting
initial investment costs and multiplying the probability of success rate at each stage
of the business. The DCF technique makes it possible to ascertain and compare
Takeda Pharmaceutical Co., Ltd. 19
the value of respective businesses with very different revenue and cost profiles over
time. Added to which, we have recently begun making use of so-called Monte Carlo
simulations to estimate the probabilities involved in bringing a successful new drug
to market.
As might be expected, our Sales and Marketing Department handles many fig-
ures, including sales targets, and these data are generated through mathematical
modeling. In drafting sales and marketing plans, binomial distribution and other
mathematical techniques are used to project sales turnover and market size. Math-
ematical methods, including statistical and stochastic techniques, are particularly
valuable when there is a need to perform a quantitative analysis of market needs.
The previous section examined the business applications for a number of mathemat-
ical techniques. My goal in this section is to address mathematics and its role within
the company from my perspective as president of Takeda. This can be expressed by
the following equation:
aX 3 + bY 2 + cZ = 0
Reference was made to this topic at the beginning of this article, but a company
president is required to deal with numerous issues spanning a diverse spectrum of
operations in which he or she may have little or no experience. Further, the leader
must be able to reach good, timely business decisions that affect the interest of both
employees and stakeholders and the general public. There is no one “solution” to
the above equation. The right “solution” is wholly dependent on the specifics of any
given set of circumstances.
In managing day-to-day operations, there are often times when I get the feeling
that I am subconsciously engaged in solving a system of equations. These can be
multivariate equations that have a significantly higher number of variable parameters
than those given in the above equation, equations with many thousands of constants
and variables. There are times when it is literally impossible to pin all the variables
down.
Often I have attempted to use my judgment to convert a number of the many
variables into named constants. It is a process that involves bringing the various
variables within a defined area in a bid to stabilize the variable parameters and is
based on suggestions from the numerous individuals who offer their experience. It is
also a process of logical thinking that is essential to reaching quick, timely decisions.
In this sense, the mathematics performed by a company president involves taking
certain variables and having the readiness and resolve needed to transform them
into constants.
While my contact with mathematics may pale in comparison with that of those
who devote their professional careers to crunching numbers, the fact that I am
dealing with forms of mathematics might be interesting to some. Already, significant
progress has been made with Enterprise Resource Planning (ERP) and other systems
that support corporate management, and I am sure I am not alone in anticipating
further support for business from the field of mathematics.
Takeda Pharmaceutical Co., Ltd. 21
In closing, I would like to touch upon the subject of the liberal arts. The importance
of a liberal arts/humanities education to corporate leaders and managers is well
established. This is grounded in the fact that in classical antiquity the phrase “liberal
arts” originally denoted those subjects of study that were “intended to liberate the
soul.” Liberal arts refers to the Trivium (the three core subjects taught first) of
grammar, rhetoric, and logic (dialectics) and the Quadrivium (the four subjects
taught following the preparatory work of the Trivium) of arithmetic, geometry,
astronomy, and music, with the three subjects of logic, arithmetic, and geometry
comprising the all-important mathematical elements of the liberal arts.
In reviewing the topic of mathematics in business, I have reaffirmed the
importance to business leaders, who are required to make and be accountable for
sound, independent decisions, of developing reasoning capabilities in mathematical
domains. Going forward as well, I look forward to continuing to forge links with
those who make mathematics their profession.
Born in 1946 and a native of Yamaguchi Prefecture, Mr. Hasegawa joined Takeda
Pharmaceutical Company Limited in 1970 having graduated from the Waseda
University School of Political Science and Economics. He has been a Member of the
Board of Directors at Takeda since 1999 following more than a decade of executive
positions at overseas subsidiaries.
Mr. Hasegawa was appointed President and CEO in 2003 having served as
Managing Director of the Corporate Planning Department as well as the Corporate
22 Y. Hasegawa
Norio Wada
Director and Chairman (2007–2012)
NTT Group
Introduction
It was not until after I began working that I came to understand the significance of
learning mathematics—a subject in which I was out of my depth and on which I was
destined to spend a huge amount of time. A handful of people will delve into this
subject as mathematicians, but what does mathematics mean to the average person?
Think of the following question as an example. There are one red crayon, one
red pencil, and one blue felt-tipped pen in front of you, and you are asked, “How
many?” How do you answer? It is said that, when small children are asked this
question, many are stuck for an answer. However, if they are somewhat older, it
is said that they can answer “Three” based on the reasoning that all the objects
are writing materials, or they focus on color and reply “two red ones and one blue
one.” Answering this question involves two important thinking processes. The first
is to analyze the objects placed in front of you from a variety of perspectives,
such as shape, color, and purpose. The second is to discover what these objects
have in common. By learning mathematics using numbers and graphics, we have
acquired these thinking processes, i.e., the ability to look at objects from a variety
of perspectives, and ultimately discover a truth that suits the intended purpose and
express it in a simple way.
Once we graduate from school, we are quite often confronted with problems
that we have not studied in school, and become highly upset due to the fact that
we don’t have a clue as to what to do. My experiences show that the mathematical
Y. Giga and T. Kobayashi (eds.), What Mathematics Can Do for You: Essays and Tips 23
from Japanese Industry Leaders, DOI 10.1007/978-4-431-54346-6 4,
© Springer Japan 2013
24 N. Wada
It is not an exaggeration to say that it is mathematics that ensures that the telecom-
munications provided by NTT are safe and secure. “Safety” in telecommunications
means that you know for sure that you can be connected to the person you want
to contact at any time you want. To make this possible, it is necessary to forecast
the volume of communications traffic, to invest in the required communications
facilities, and to provide the ability to set up alternative routes automatically when
some facilities fail. These activities are supported by a fundamental research area
called traffic theory. Let me introduce a formula that plays a central role in this
theory.
The probability at which the originator of a call is not connected to the called
person is referred to as the call loss rate. NTT has aimed to keep the call loss rate
below 10% in building its telephone network. One mathematical formula important
in the planning and deployment of telephone network facilities is called Erlang B.
It enables us to calculate how many circuits are needed to keep the call loss rate
below a certain level when serving a specified number of users.
aS
B= S!
S
an
1+ ∑
n=1 n!
B: Blocking Probability
a: Traffic (in erlang)
S: Number of lines
NTT 25
This formula was proposed as far back as 100 years ago by Agner Krarup Erlang,
a Danish mathematician. Even today in the 21st century, the Erlang B formula
remains highly relevant and is playing an important role in telecommunications.
This is due to the fact that, simple as it is, this formula very precisely models the
loss rate of telephone calls.
The Erlang B formula has been supporting NTT’s mission of providing network
services as a social infrastructure, and, moreover, NTT has researchers with suffi-
cient mathematical capability to use the Erlang B formula skillfully and are highly
motivated to apply it to the company’s huge telephone network. The “collaboration”
between mathematics and skilled researchers is the foundation on which NTT stands
today. This is also a prime example of how mathematics has been turned into a
practical tool.
Today, the conventional telephone is well past its heyday, and the Internet has
taken over. However, today’s Internet is mostly best-effort-based, and so it is taken
for granted that communication becomes slow or even unavailable when there is
traffic congestion. What is desired from now on is the construction of the NGN
(Next Generation Network), which adds high reliability and a quality of service
(QoS) guarantee to IP communication used in the Internet. In this new environment,
there is a renewed recognition of the importance of the Erlang B formula. Today,
a small number of users who are highly knowledgeable in ICT (information and
communications technology) can use most of the available network bandwidth.
If we can create an environment that retains the beneficial features of the current
Internet, namely, imposing few constraints and allowing use for any purpose, but
can also control traffic in such a way that anyone, irrespective of his or her level
of ICT literacy, can impartially receive network services whenever he or she wants,
then we can greatly contribute to building a better society.
Mathematical thinking plays an important role in creating services that open new
horizons. It is hoped that ICT will be harnessed to create new innovative services
and build new social infrastructures, with a view to resolving the mountain of
social issues that confront Japan and the world. In this section, let us consider
the importance of mathematical thinking in launching a new business. Starting
a new business requires a variety of capabilities, such as the ability to sniff out
business seeds, willingness to meet the challenge of exploring untrodden paths
as a trailblazer, and the ability to discover some rules in the business you are
exploring (even though you may be groping for them in a kind of fog) and to
develop best practices from them. As I wondered how and what kind of people
acquire these capabilities, I came to realize that there is a clue in geometry. From
two-dimensional figures that you draw on paper with a pair of compasses and rulers,
three-dimensional figures, such as a sphere, to complex multi-dimensional figures
that you can draw only in the imagination, geometry has expressed its intrinsic rigor,
26 N. Wada
unpredictability, and beauty. I believe that the ability to create businesses arises from
geometric thinking. When you look closely at the individual tasks you handle every
day, you’ll find that they exist more or less independently of each other, and it is
often assumed that the ability to judge and solve problems on the spot is important.
This is particularly true at the initial stage of launching a business. However,
this approach is not conducive to creating services or businesses characterized by
consistently high quality and that make a meaningful social contribution. Rather,
it is advisable to take a renewed look at the flows and relationships among people
and data, and create a geometric figure from them. This will enable you to see what
has previously been hidden from you, and see an efficient and reasonable workflow
emerge as a beautiful figure. In other words, I believe that, rather than considering
tasks as a series of steps, imagining a geometric figure that represents tasks as parts
of an ecosystem is essential to solving various issues that arise in our globalized
complex society.
As a specific example, let me introduce a current research project.
One of the core technologies that support service creation is video coding.
MPEG2, which is now widely used in digital television and DVD, was standardized
by ISO/IEC in 1994. The activities of Dr. Hiroshi Yasuda (then an NTT employee),
who worked from a global perspective, played a large role in this standardization.
Today, nearly 20 years on, as data traffic increases day by day, research on
video coding with a higher level of compression is being undertaken globally, with
NTT playing a significant part in it. The current video coding is based on two
technologies. Simply put, one technology reduces the amount of data by regarding
changes in shading within each video frame as waveforms, and by expressing the
frame as a set of superimposed simple waveforms using discrete cosine transform,
and the other technology decreases the volume of data by extracting only changes
from the previous frame in encoding a frame. Conventional video compression
technology has been developed through trial and error by many pioneers in the
field who relied on physical phenomena that people can recognize, such as color
and motion. In contrast, NTT is now doing research on the ultimate compression
technology. This is not an extension of any conventional method but is entirely
different. It uses what is called genetic programming, which can, through iteration of
computation, evolve a video coding function into one providing higher compression.
In other words, unlike conventional technology that provides average compression
performance for any type of video, a technology that can be compared to ready-
made clothes, the new technology aims to provide ideal compression performance
for each type of video, which can be likened to tailor-made clothes.
When seen from a different angle, this research is expected to bring benefits
beyond just high compression of video. Since the coding function derived by
this technology is a mathematical formula that represents a recognizable object,
it has the potential to shed light on the mechanisms by which humans recognize
color, shape, or even an object. A study that was begun to remove reliance on
human perceptual recognition may turn out to help illuminate the mechanisms of
recognition in humans. I have a strong sense of how mysterious this ecosystem-like
NTT 27
or geometric approach is. It is intriguing to wonder what this research, albeit still
in its infancy, which was initiated to compress images or video, will bring us in the
future.
In 2004 when I was its president and CEO, NTT announced its medium-term
management strategy. The main thrusts of the strategy were the promotion of
optical access, construction of the NGN, and the provision of new services that
exploit these. At that time, despite its eventual success, aggressive investment in the
construction of optical access networks was derided as crazy by our competitors and
investors around the world. However, I had a firm belief that progress in broadband
communication made possible by the penetration of optical access would usher in a
paradigm shift, the impact of which would be as profound as that of the industrial
revolution. However, when I needed to make a critical decision as the leader of a
huge enterprise, sometimes my determination almost crumbled under the pressure.
Yoshida Shoin led a private school, Shoka Sonjuku, which exerted a great
influence on the Meiji Restoration. Two days before his execution, he began to write
Ryukonroku, which is considered to be his last will and testament addressed to his
students. In this essay he said, “Form a resolution and make it the foundation for
everything.” I assume he meant that everything hangs on how you ask yourself what
you should do and how you form a resolution based on that. This is the most difficult
thing to do today when things change dramatically in a discontinuous manner, but as
such, I believe, it is all the more important for leaders to show this kind of resolve.
Time and again I was encouraged by Shoin’s words, which were backed by
his firm conviction. Today, 7 years after I made that decision, seeing broadband
access widespread and the Internet flourishing, I feel relieved to know that my
determination and actions in those days were not misdirected.
The first part of this chapter described how mathematics and mathematical
thinking play an important role in various activities of NTT and are critical to
ensuring the sustained development of our business. In the concluding part, I
explained how it is necessary in corporate management to forge a future vision
and have the strong resolve to pursue it. The message I wish to convey here is that,
regardless of whether your business is big or small or which industry you are in, it is
only when the above-mentioned two elements—mathematical thinking, and vision
linked to determination—work together like two wheels of a cart that they provide
a great driving power to open new horizons.
NTT had been publically managed as Nippon Telegraph and Telephone Public
Corporation until it was privatized in 1985 under the liberalization policy for
telecommunications. In 1999, it created a holding company and shifted to group
management. The group has a significant presence in various fields of business, in-
cluding Regional Communication, Long-distance & International Communication,
Mobile Communication and System Integration.
In the fourth quarter of FY2009, NTT made consolidated net sales of 10.1 trillion
yen and a group operating profit of 1.1 trillion yen.
Norio Wada was born in 1940. Following graduation from Kyoto University with a
B.A. in Economics, he joined NTT in 1964. After he became a Senior Vice President
in 1992, he held various important posts, including the manager of the Tohoku
Branch and a group Executive Vice President. He became a Senior Executive Vice
President in 1999 and President and CEO in 2002.
Under Wada’s leadership, the NTT Group announced in 2004 the Medium-Term
Management Strategy, which set a company goal to establish the Next-Generation
Network (NGN) based on optical access networks and full-IP telephony. It con-
tributed to the spread of broadband services in Japan.
He was the NTT Group’s Director and Chairman from 2007 to 2012. He had also
experienced being a Vice Chairman of the Japan Business Federation (Keidanren)
for 4 years beginning in 2003. In that position he was in charge of policies for
housing problems and the aging society. Because he has held various important
posts in employers’ associations, he is celebrated in Japan’s business communities.
Nomura Holdings, Inc.
Kenichi Watanabe
Group CEO (2008–2012)
Nomura Holdings
As far back as the Edo period (1603–1868), rice futures were already being traded at
the Osaka Dojima Rice Exchange. Japan in those days was clearly highly innovative,
given that it was more than 100 years before Europe began conducting similar
financial transactions at the end of the 19th century. However, it was North America
and Europe that led the way in derivatives, starting in the 1970s. Through the
remarkable developments in the field of financial engineering, complex derivatives
and other financial instruments have only become widely used in the last 20 or 30
years. Today, with its extensive use of mathematics, the discipline is probably better
described as “mathematical finance”. Indeed, without mathematical application
finance would not be as sophisticated or globalized as it is now (although the
development of rules, legislations and other infrastructure has also been vital).
Mathematics is essential for analysing financial instruments related to deriva-
tives, a prime example being the application of probability theory. “Ito calculus,” the
stochastic integral and stochastic differential developed by Dr. Kiyoshi Ito, stands
up alongside Newtonian calculus in terms of its importance to probability theory.
“Ito’s lemma” (Ito’s formula) and the Ito representation theorem in particular are
extremely useful in the area of finance, and it would be no exaggeration to say that
researchers and professionals worldwide use these formulae on a daily basis.1
1 The Nomura Group has endowed chairs and research centers at universities in Japan and abroad,
contributing to society. In 2007, to commemorate the 2006 Carl Friedrich Gauss Prize awarded to
Y. Giga and T. Kobayashi (eds.), What Mathematics Can Do for You: Essays and Tips 29
from Japanese Industry Leaders, DOI 10.1007/978-4-431-54346-6 5,
© Springer Japan 2013
30 K. Watanabe
Derivatives function as an effective risk transfer tool because they bring about
the composition and decomposition of future cash flows linked to a reference index
with a specific risk. Derivative transactions between parties wanting to hedge risks
such as equity, interest rate, and foreign exchange rate and parties wanting to take
on that exposure make it possible to effectively transfer the risks. Derivatives can
also be used to vary leverage,2 enabling a financial institution to take on exposure
many times greater than the actual amount of equity capital.
However, as seen with the collapse of Long-Term Capital Management back
in 1998 and more recently at major investment banks where leverage exceeded
30 times, a financial crisis induced a credit crunch, leaving these institutions with
insufficient capital buffers. This highlights that a key management issue for financial
institutions is to ensure the efficient allocation of risk capital. This requires precisely
calculating the potential profit, loss, and degree of risk associated with liquidity
procurement, asset valuations (mark to market), as well as the necessary amount of
equity capital, conducting a factor analysis, and calculating the investment efficiency
(expected excess return per unit of risk). In recent years, the need to upgrade risk
management has become increasingly pressing, but this is not just a simple matter
of calculating the value at risk (VaR),3 the Greeks,4 or credit exposure.5
Returning to derivatives, various types of exotic derivatives (derivatives with
complex path dependencies) have been developed since the end of the last century,
where the application of mathematics has enabled numeric calculations for pricing.
Even for standard (plain vanilla) products like European options, the market has
transformed into one, whereby different strike prices and maturities change the
implied volatility (a phenomenon known as “volatility skew” or “volatility smile”),
spurring competition to develop pricing models with a greater level of sophistication
than the Black–Scholes model. New credit derivatives, such as credit default swaps
(CDSs) and collateralized debt obligations (CDOs), that use credit as the underlying
asset (reference index) have emerged, causing a sudden burgeoning in market
size, but the subprime mortgage meltdown has also made apparent a number of
Dr. Kiyoshi Ito, Nomura Holdings endowed a research division known by the acronym KIMF at
Kyoto University. The division, which operated from 2007 to 2010, comprised two subdivisions
belonging to the university’s Research Institute for Mathematical Sciences and Institute of
Economic Research, respectively. Offshore, the company has established the Nomura Centre for
Mathematical Finance within the Mathematical Institute at the University of Oxford.
2 Leverage is the reciprocal of the equity capital ratio, using the lever principle to increase
investment efficiency. Debt can be used to procure funds to create asset holdings many times greater
than equity capital, boosting the expected return on equity (if the credit risk is excluded).
3 Value at Risk is the maximum expected loss over a certain period of time within a certain
confidence interval. VaR is used in practice to determine standards for the necessary amount of
equity capital.
4 “The Greeks” are the quantities representing the sensitivities (partial differential) of the price of
financial instruments to a change in the underlying asset prices or parameters on which the value
of an instrument is dependent.
5 The amount of exposure to the credit risk of borrowers, counterparties, etc.
Nomura Holdings, Inc. 31
embedded problems with securitized products like CDOs. One example is the
normal Copula model widely used as a de facto standard for risk assessment
(although the assumptions are slightly unrealistic).
The recent financial crisis, spurred by a credit crunch, has served as a reminder
of something that has in fact long been known, namely, that risk diversification
based on the “Law of large numbers” does not always work. Obviously, the crisis
was less about the pros and cons of Copula models, and more to do with the
ways in which securitized products were structured and distributed, spawning
conflicts of interest and fraud-like moral hazards. Market participants as a whole
(as well as policymakers, regulatory authorities, rating agencies, and home loan
borrowers) were also implicated for professing themselves to be diversifying risk,
while ignoring the possibility of credit risk arising, based on the assumption that
real estate prices would continue to increase.
Mathematics leads to simple and beautiful conclusions by ignoring and idealizing
various factors in order to abstract. However, reality is both more complex and
discrete, and in many cases mathematical conclusions are only approximately valid.
There is also the risk that the assumptions used for valuation models differ from the
market (model risk), as well as issues with the reliability of the parameters used in
the model (parameter risk).
Naturally, mathematics is only one of many tools used in finance. Simply
applying mathematics without analyzing economic fundamentals can be not only
futile, but even dangerous. For example, quantitative analysts, or “quants”, tend
to concentrate on formulaic mathematical analyses, putting less importance on the
underlying assets, financial markets, or the real economy. Investors only look at
actual probability (physical measure), but derivative traders and quants analysts
only look at the risk-neutral probability (risk-neutral measure). It could be confused
that in principle mark to market should be done under the risk neutral measure Q,
and risk sensitivities should be measured under the physical measure P. (Even this
dualistic approach is of course overly simplistic.)
Neoclassical economists believe that as the formation of future rational expec-
tations determines the market price at the current point in time, all information is
reflected in the market price. Financial mathematics in relation to derivatives has a
similar conception, where pricing is done under a valuation measure based on no
arbitrage opportunity in the market (risk neutral measure),6 but that is not sufficient
when risk enters the picture. In fact, underlying asset price movements can only
be observed retroactively under physical measure P. In measuring risk, distribution
under physical measure becomes critical.
When it comes down to it, the definition of risk is not obvious, and how to
estimate risk is actually a major challenge. To simulate possible future scenarios
requires many sample paths, but the actual price path is no more than the realization
of one of the envisaged sample paths. Even if past investment management
performance has been strong, it is difficult to determine whether that was through
6 “No arbitrage” means that reliable positive excess returns cannot be obtained without risk.
32 K. Watanabe
skill or luck. This would imply that understanding risk is far from straightforward.
As noted in Nassim Nicholas Taleb’s “The Black Swan: The Impact of the Highly
Improbable”, it is difficult to envisage risks that have never been seen before, but
which could potentially occur.
The recent financial crisis has highlighted exposure towards the risk of liquidity
drying up, whereby a decent asset ceases to sell at an acceptable price on the market,
and the correlation risk, which tends to grow contagiously. Risk factors abstracted
in normal valuation models emerged as much greater than estimated, requiring a
rethink of risk management. It can be said that over the last 10 years, as financial
markets worldwide have all been heavily biased towards derivatives, this has meant,
failure towards observing proper risk management.
In recent years, neoclassical economics has fallen slightly out of favor, shoul-
dered aside by new fields such as “Econophysics” and “Behavioral Finance”.
“Econophysics” studies statistical estimation using actuarial science, extreme value
theory or statistical physics, and consequently does not rely on an idealized
stochastic process of the underlying asset. Experience bears out that the return (the
rate of return) on many financial assets does not follow a normal distribution but
rather a fat-tailed one,7 which exhibits extremely higher values at its tail than the
normal distribution does. Not only the diffusion process seen in Brownian motion,
but also, in the short term, the risk of asset price jumps (defaults, booms and
busts) must be taken into consideration. In the case of non-tradable assets, such
as real physical investments, catastrophe bonds, insurance, or illiquid assets, risk
management needs to be even more conservative.
7 However, this fact was known before econophysics, and the hypothesis that stable distribution and
some law have a dominant effect, with some parameter such as a physical constant existing, seems
a little thinly backed.
Nomura Holdings, Inc. 33
In 1925, the securities arm of the Osaka Nomura Bank spun off to establish
Nomura Securities. It reorganized into a holding company and was listed on the
New York Stock Exchange in 2001, while Nomura Securities remained the group’s
core company. Since the acquisition of Lehman Brothers’ Asia-Pacific franchise and
certain operations in Europe and the Middle East, the Nomura group has accelerated
its overseas operations, providing integrated financial services in more than 30
countries or regions. The group not only has achieved a dominant share in Japan
but has solidified its position as a top-ranking investment bank in the world.
Mr. Watanabe started his career at Nomura in 1975 in corporate finance and has
held a number of key positions in finance as well as international planning.
He was promoted to the Board of Nomura Securities in June 1998 and has been
responsible for a wide range of roles including finance (CFO), credit evaluation
(CRO), and administrative affairs (CAO), which included general services, compli-
ance, and investor relations.
He is also a member of the Board of Trustees of the Nomura Foundation, which
contributes to advancing the social sciences and the world economy, cultivating the
arts and culture, fostering international talent, encouraging international understand-
ing, and supporting the progress of human society.
Outside the firm, he is an Executive Member of the Policy Board of the
Japan Business Federation (Keidanren), contributing to the growth of the Japanese
business community.
Mr. Watanabe graduated from Kobe University with a B.A. in economics. Away
from the office, he is an avid reader and particularly enjoys Chinese classics.
HORIBA, Ltd.
Atsushi Horiba
Chairman, President and CEO
HORIBA
Y. Giga and T. Kobayashi (eds.), What Mathematics Can Do for You: Essays and Tips 35
from Japanese Industry Leaders, DOI 10.1007/978-4-431-54346-6 6,
© Springer Japan 2013
36 A. Horiba
HORIBA Group products use chemical and physical reactions to measure the
parameters of substances in various forms including atmospheric air, the water in
our environment, and solid state matters. Therefore, many personnel ranging from
ordinary employees to executives have scientific backgrounds, which means the
education they received in high school and university included mathematics, with
no exception. The development and design of measuring instruments inevitably
involves a large amount of mathematical content. Some of these instruments work
by using mathematics itself.
The pH meter, our company’s first product, is based on electrochemistry. This
product measures hydrogen potential. An analysis device that is one of our key
products measures gases based on the absorption of infrared light, which is a part
of the electromagnetic spectrum. In recent years, efforts have been made to develop
laser-based gas-measurement technology as well. A laser beam whose wavelength is
adjusted to a target absorption peak is radiated on the basis of quantum mechanics,
and gas concentration is calculated from the extinction intensity of the light beam.
As another example, we have produced spectrometers utilizing Fourier transform
technology, which is based on general mathematical principles. This method, which
is called “Fourier transform spectroscopy,” can measure absorption spectra of
substances with high sensitivity within a short time because it uses light intensity at
high efficiency. Here, owing to the wave characteristics of light, beams interfere
with each other. As shown in Fig. 1, by splitting a light beam from a light
source into two sub-beams and subsequently combining them back into a single
beam with a beam splitter, interfered light can be generated and detected. Such
a device is called a Michelson interferometer. If the mirror on one of the split
paths moves back and forth continuously, the optical path difference between the
two sub-beams changes according to the mirror position. In the case of single-
wavelength light, the intensity profile of the combined beam at a detector is
represented by a cosine wave because of interference. In the case of wide-band light,
which consists of continuously different wavelengths, the light intensity becomes
strongest when the optical path difference is zero and weakens with the path
difference. Figure 2 shows the light-intensity profile caused by interference, referred
HORIBA, Ltd. 37
Fig. 1 Schematic of a M1
Michelson interferometer
S
BS
L1 M2
L2
D
S: Light source, BS: Beam splitter, L1, L2: Lenses,
M1: Fixed mirror, M2: Moving mirror, D: Detector
When describing natural phenomena, we human beings have been using mathemat-
ical expressions to simplify the principle, not only in physics and chemistry but also
in biology, thereby expressing phenomena intuitively and in an objective fashion.
Here, “to ascertain the essence of things” and “to identify the true nature of things”
which underlie Kyoto culture seem to coincide amazingly well with methods of
mathematical expression.
38 A. Horiba
Intensity
Wavenumber (cm−1)
Communications
Among a total of nearly 5,000 employees within the HORIBA Group, about
3,000 speak languages other than Japanese. The number of French employees is
particularly large at 1,000. Ranking next are 700 Americans and 700 Germans,
followed by 200 in China and 150 in the United Kingdom. All the people working at
HORIBA, Ltd. 39
in Engineering and has held a V.P. position with a top semiconductor and related
device manufacturer in the United States, which adds a global perspective in our
management team.
The HORIBA Group has become a large corporate entity. But we aim to remain a
“great medium-sized enterprise” so that “big company disease” will not develop.
Our company is strictly a vessel for human wisdom, and we place a high value
on this fact. In the HORIBA Group, we call human resources “human assets.”
Engineers tend to portray an inorganic image as their work is based on mathematical
principles. However, when combined with the human nature of being able to
concentrate and focus on one thing, they can work enthusiastically. The skill to
understand and manage the engineers’ minds is a part of the know-how in our
company’s management style.
After studying physics at a Japanese university, I enrolled as a junior in
the Department of Electrical Engineering in the College of Engineering at the
University of California Irvine (UCI). At the time, I had a very hard time coping with
the English language. However, the mathematics I studied at the university in Japan
was helpful. There are no language barriers in mathematics. After graduating from
UCI, I worked as the first HORIBA service engineer stationed in a foreign country
and dealt with a considerable amount of our products that experienced problems.
When I made reports to the head office in Kyoto, the R&D team’s response was,
“We have never had such problems in Japan.” The pain I experienced at that time
could be likened to what must have been experienced by Japanese border soldiers
who were stationed far away from home more than 10,000 years ago. I felt caught
in a crisis, thinking, “Operations that give no thought to the front-line troops will
stagnate.” As a result of this experience, I dared to become “anti-headquarters.” One
result of this has been our spirit of placing importance on the front line that supports
this global company.
In large enterprises, job assignments tend to be divided up according to em-
ployees’ specialties. For example, electricians are assigned to electrical tasks and
physicists are assigned to physics tasks. However, in the HORIBA Group, our
engineers are trained so that every engineer can understand electricity, physics,
chemistry, and software. At present, the number of our product models in headquar-
ters exceeds 1,000. The number increases considerably when the group companies’
products are combined. If 5,000 employees are to handle operations ranging from
development to after-sales service for all our product models, every employee needs
to be multi-functional and become a multi-skilled worker so that they can deal
with any situation. Here also, a mathematical way of thinking seems to be of help.
We have invested significant time and effort in in-house education.
HORIBA, Ltd. 41
Training and education are provided at a variety of levels and fields to develop
human assets that support “technical meshes = chains.” A training program called
“HORIBA College” was also established, consisting of 260 classes spreading across
all departments, strengthening the power of our human assets, and leading to
improved product power and organizational power. Each year 10 or more employees
from group companies in Japan are sent to our overseas group companies for a
1-year internship to develop a global point of view. These employees are selected
not on the basis of linguistic proficiency but on the basis of what they want to do. So
far, this system has given 10% of our employees experience working with different
cultures and ways of thinking in foreign countries. And now, in any department in
our company there is at least one employee who can communicate in English with
experience from living in a foreign country.
In any company, things will not go well with the effort of only a single talented
person. Even if not every employee is a superman, we need to work together to
ensure that all our employees in combination will be like panes of stained glass and
radiate beauty when illuminated by light, making the HORIBA Group as a whole a
super dream team.
Sixty-seven years have passed since the HORIBA Radio Laboratory came into being
in 1945. Sales have risen to over 100 billion yen per year, and 60% of the employees
in the HORIBA Group are non-Japanese. There are R&D bases in foreign countries
as well, where managers of local nationalities are in charge of the business. In spite
of the expansion, I am happy to say that the venture spirit cherished since the early
days of our founding has been passed on. This fact is expressed symbolically in
our company motto: “Joy and Fun.” This motto contains the hope that the work
on which we spend the most active period of our lives should be addressed with a
spirit of pride and challenge. In an R&D-oriented enterprise like ours, added value
is created only when employees find their work worthwhile.
The spirit of “Joy and Fun” has penetrated our group companies abroad as well.
They differ not only in ethnic groups and cultures, but also in business fields.
A center-of-gravity force is required to unify these differences and diversities into
a single brand. Unification leads to corporate values and produces a corporate
culture. The products of our company are based on science including mathematical
principles. Our executives and our managers as well as many employees have
scientific backgrounds. Furthermore, about 50 of our employees in Japan and about
100 of our overseas employees have PhDs. The HORIBA Group is basically run on
a global basis by around 10 top executives, including some non-Japanese. Most of
them have engineering experience.
Executives in the HORIBA Group learn their business management strategy by
trial and error. Since they have scientific backgrounds, they can naturally conduct
experiments and go on to implement the outcome on the basis of the resulting
42 A. Horiba
data. The process of explaining the natural or social sciences using a mathematical
calculation is called simulation. When an actual simulation is performed, the result
often turns out to be different from reality. Lessons are learned from these failures
and improvements are made. As a result, the simulation approaches reality more and
more closely.
Following the spirit of “Joy and Fun,” Japanese and French R&D personnel
or American and German R&D personnel cooperate internationally to raise the
competitiveness of products bearing the HORIBA brand. Many of the products of
the HORIBA Group hold the largest market shares in the world. These products
cover various fields, including the automotive industry, the medical field, the field
of analytical devices used for environmental measurement and advanced research,
and the semiconductor manufacturing field. The absence of these products would
disrupt automotive development, and the manufacture of semiconductors would be
discontinued all over the world. This reflects our social responsibility and our pride.
HORIBA engineers across the world support our products.
Saying that “the HORIBA Group is one company” means that business is
conducted on the basis of a single set of values and a single corporate cul-
ture. However, the localities of each different region are also respected. Taking
environmental measurements as an example, there are some differences in the
regulations between Japan and Europe. However, engineers and top management
overcome these differences between regions while being connected on the basis of
mathematical principles and one corporate culture. Another important point is that
all entities take pride in their own cultures and learn from one another’s experiences,
forming a strong team. In that sense, it is fortunate that the head office is located in
Kyoto, which has a long history of cultural diversity.
HORIBA, Ltd. 43
Atsushi Horiba was born in Kyoto in 1948. He graduated from the College of
Engineering, University of California, in 1977. He joined HORIBA in 1972 and
worked as Overseas Executive General Manager, Senior Sales Manager, and Senior
Executive Director before becoming President & CEO in January 1992. Since June
2005, he has been Chairman, President & CEO. He has also held various important
posts outside the company, including President of the Japan Analytical Instruments
Manufacturers’ Association (JAIMA) and a member of the Board of Advisors at
UC Irvine. He received the Officier de l’ordre national du Mérite in 1998 and the
Chevalier de la Légion d’Honneur in 2010 from the French government.
Railway Technical Research Institute (RTRI)
Eisuke Masada
President
Railway Technical Research Institute
Y. Giga and T. Kobayashi (eds.), What Mathematics Can Do for You: Essays and Tips 45
from Japanese Industry Leaders, DOI 10.1007/978-4-431-54346-6 7,
© Springer Japan 2013
46 E. Masada
Objectives such as increasing the speed of trains, better riding comfort for pas-
sengers, and improvement of maintainability of rails are pursued in railway
technologies at all times. The explication of causes of abnormal phenomena such as
derailment is also important. The mathematical analysis of train motion on the track
is the basis for these studies. The train consists of a car body, bogies, wheels, traction
motors, and other parts. It runs following the track profile, and its components are
connected to each other with suspension systems such as springs and dampers to
suppress vibration caused by running. Physically, the running state of a train can
be expressed in analytical dynamics as a multi-body motion, where it is restrained
at the contact point between wheels and rails, and its components are connected to
each other with suspension systems. The nonlinear elements as the friction force
working at the contact point of the wheel and the elastic motion of the car body
itself also need to be considered. Then the motion of the vehicle is given by a group
of high-order nonlinear differential equations. It can be solved only by numerical
analysis. The multi-body dynamics program is usually applied for this purpose.
Final verification is carried out by operational tests utilizing a real train, seeing
whether the analytical results obtained with such a numerical method show practical
phenomena correctly. Because it is difficult to ascertain the mechanical construction
and to adjust characteristics of components in actual operation, the test equipment
called the rolling stock test stand, as shown in Fig. 1, is utilized. Operational tests
are carried out on rolling stock, in which longitudinal motion is restricted and
which operates with wheels driven by rotating discs simulating rails. Their results
are compared with analytical ones obtained with a real-time simulation program,
which runs parallel to and coincident with the operation on the stand. It is possible
to prove the design of rolling stock through such an arrangement before actually
manufacturing and testing it.
In the case of a magnetically levitated transportation system, which is called the
“linear motor car” in Japan and is shown in Fig. 2, the thrust, lift, and guide force
working on the train are determined with the electromagnetic forces generated by
the current flowing in coils equipped on the guide way. Therefore its characteristics
cannot be fully dealt with unless both the dynamics equations mentioned above and
electromagnetic equations are solved simultaneously. It requires a more sophisti-
cated scheme for numerical analysis.
Fig. 3 Aseismatic Evaluation and Measures for Existing Railway Facilities considering their
Interaction
swelling river behaves around a bridge pier, and so on. With recent unstable weather
conditions, new tasks are created such as evaluating the influence of strong winds
against a moving train.
Figure 3 shows a model of a situation, where a seismic wave is generated with a
dislocation, propagates in a stratum with a complicated structure, vibrates structures,
and exerts an influence on a running train. If such a phenomenon can be analyzed
numerically with the finite element method, and the state of vibration as well as
its frequency characteristics are clarified, we can consider the proper measures for
secure structures and safe operations. It makes it possible to evaluate measures
against derailment for a train operating relatively near a seismic center.
An environmental problem caused by the railway system is the audible noise gener-
ated by train operations in areas neighboring the railway line. The aerodynamic
noise from a moving train increases with its velocity. Therefore, it is important
to introduce measures to suppress aerodynamic noise for further increasing the
operational velocity of the high-speed system. The aerodynamic noise is caused
by exfoliating eddies generated at discontinuities in the shape and projections of the
car body as well as by the oscillation of the turbulent boundary layer, which exists
along the whole surface of a train. The sound source can be identified and evaluated
clearly with the results of measurements in a wind tunnel as shown on the left-hand
Railway Technical Research Institute (RTRI) 49
−104 104
side of Fig. 4, or with a running test of a train with a microphone array, in which a
large number of microphones are arranged along the cross section of the car. On this
basis the measures are studied in the design of the shape of a train. Because the level
of audible noise is important in the frequency area, mathematical methods become
important for revising differences in observation windows in the measurement of
each microphone caused by its position and the class of sound source.
Although it is a rather localized matter, the audible noise from the pantograph
(current collector) is remarkable in high-speed operation. Applying porous material
to the surface of the pantograph is being considered in the study of measures to
suppress noise. The right-hand side of Fig. 4 shows the field of air flow obtained in
experiments and simulations in a case where this material is affixed to the surface
of a cylindrical component. It has been proven by solving the field of air flow with
numerical simulation, that only a limited number of stable eddies are generated in
the case where the porous material is affixed. It confirms the improvement of the
performance of the air flow and makes it possible to select the appropriate porosity
of material and its thickness.
Eisuke Masada completed his doctoral work in the Graduate School of Mathematics
and Physics, The University of Tokyo, in 1965. As soon as he received his PhD in
Engineering, he started teaching in the university’s Department of Electrical Engi-
neering. He became a professor in the Department of Engineering in 1983. After
his retirement in 1998, he became a professor in the Department of Engineering,
Tokyo University of Science. He has been President of RTRI since 2007 and has
held various posts, such as President of the Institute of Electrical Engineers of
Japan (IEEJ), President of the Electric Technology Research Association (ETRA),
a director of Fuji Electric Holdings and Vice-President of the Japanese Industrial
Standards Committee (JISC). His awards include the IEEJ Meritorious Contribution
Award, the Transport Culture Award (from the Ministry of Transport), and the
EPE-PEMC Council Award.
Japan Oil, Gas and Metals National Corporation
(JOGMEC)
Hirobumi Kawano
President
JOGMEC
Introduction
In this article, how tasks are theorized, quantized, or transacted in various fields and
areas of petroleum and natural gas exploitation is described. The requirements for
future mathematical development are also explained. Firstly, a brief description of
the features of petroleum and natural gas exploitation is given.
A schematic view showing oil and natural gas fields is given in Fig. 1. The green-
colored area is often described as an oil reservoir, and the figure may be interpreted
as an underground cave filled with crude oil. The reality, however, is that the green-
colored area is a rock layer as seen in Fig. 2. In many cases, the rock that collects
oil or gas is sandstone or limestone (see Fig. 3). There are very small continuous
cavities inside them, and collectively they constitute 20–30% of the volume of all
rocks. Inside the cavity, there may be oil, gas, or water, and together they form an
oil field or a natural gas field.
Y. Giga and T. Kobayashi (eds.), What Mathematics Can Do for You: Essays and Tips 53
from Japanese Industry Leaders, DOI 10.1007/978-4-431-54346-6 8,
© Springer Japan 2013
54 H. Kawano
Fig. 1 Oil–gas field a few thousand meters underground [ref. “Petroleum Development Technol-
ogy Guide”, Japan Petroleum Development Association (in Japanese)]
Fig. 2 Oil–gas reservoir is a rock layer [ref. “Petroleum Development Technology Guide”, Japan
Petroleum Development Association (in Japanese)]
Plankton that accumulates at the bottom of seas and lakes is sedimented deep into
the ground through siltation. During this process, the increasing ground temperature
and pressure with depth causes some portion to change into sources of petroleum
and natural gas (i.e. hydrocarbon compounds). Rock generated by sedimentation
of plankton with soil, typically mudstone, is called source rock. Because of the
Japan Oil, Gas and Metals National Corporation (JOGMEC) 55
limestone
Fig. 3 Types of rock constituting an oil–gas reservoir and its cavity [ref. “Petroleum Development
Technology Guide”, Japan Petroleum Development Association (in Japanese)]
difference in specific gravity from the surrounding water, the source of petroleum
and natural gas thus generated drifts in the rock layer for millions of years until it
reaches its present position, forming an oil or natural gas field.
As the specific gravity of petroleum and natural gas is generally lower than that of
the surrounding water, the probability of their sedimentation is higher in convex-
shaped, low-permeability, dense rock with an upper rock layer which acts as a
cap (the dark-colored seal layer in Fig. 4), a rock layer caused by dislocation or
by protrusion of rock salt. Thus, petroleum exploitation begins with a search for
the aforementioned rocks and rock layers deep in the earth. Exploration for rocks
and rock layers capable of petroleum sedimentation at depths of a few thousand
meters underground and of several hundred meters underwater is primarily done
using seismic waves (see Fig. 5). Vibrations generated by dynamite or an air gun
propagate underground and reflect at the interface of rock layers. The underground
structure is then analyzed using data, such as elapsed time, phase, and amplitude,
from the received reflected wave. The principle resembles that of ultrasonography
used in medical practice (see Fig. 6). There are great differences, however, between
searching through subcutaneous fat in a human body, typically on the order of few
centimeters thick under the skin, and through rock(s), which may contain salt water
or oil, typically on the order of few thousand meters under the surface of the earth.
There are many cases in which a slight error may cause considerable changes in the
interpreted profile of the underground structure.
56 H. Kawano
seal
reservoir
rock
Oil field
4 3
2
source rock
1
basement rock
Fig. 4 Drift of petroleum gas inside rock layer [ref. “Petroleum Development Technology Guide”,
Japan Petroleum Development Association (in Japanese)]
Even when a rock layer that may be an oil–gas field is found, there is no current
technology to determine whether the rock layer a few thousand meters underground
contains crude oil, natural gas, or just water. The presence of crude oil or natural gas
can be only found by drilling a well into the rock layer. Drilling an exploratory well
costs 100,000,000 to 1,000,000,000+ yen, and may take 3–6 months or more.
Once crude oil or natural gas has been found by drilling some wells, an estimation
of the size, extent, and quality of the reservoir, calculation of reserves, and
development expenditure is made using the limited amount of information. In oil
exploitation, a huge investment is made initially, and return on the investment
is sought during the production period, which may last for several decades. The
rough shape and extent of the oil reservoir rock is obtained from the seismic
exploration. Porosity and the oil/water ratio is estimated from the electrical and
acoustic properties of rock samples taken from the wells and from the approximately
1 m well-hole made by dropping a logging tool into the well. The average fluidity
of the oil inside the reservoir is estimated from test production and pressure data.
Putting the information together, an oil reservoir model is constructed. The future
Japan Oil, Gas and Metals National Corporation (JOGMEC) 57
satellite
survey
vessel
seabed
sound wave
reflects at the
interface of
rock layers
crude oil and natural gas production for various exploitation conditions is estimated
by numerical simulation, and basic data for calculating the economic viability
of oil field development is determined (see Fig. 7). The technology and know-
how manifests itself in performing production estimation with high accuracy from
limited amounts of information.
Numerical simulations are carried out for various problems, ranging from exploita-
tion to development and production of petroleum and natural gas resources. In
general, handling strata a few thousand meters underground and the fluids contained
within them comes with a high level of uncertainty and differences in scale,
resolution and amounts of data. For example, though seismic exploration data may
capture the vertical and horizontal extent of reservoirs, it has a rough resolution
on the order of dozens of meters, and is called soft data. Data from wells, on the
other hand, is obtained from rock samples from the oil–gas reservoir, and is on
the order of a few centimeters. In the case of a well into which a logging tool is
dropped, electrical, acoustic and density data around the well wall is obtained every
30 cm vertically. The interval of wells, however, is on the order of a few kilometers.
Consequently, the so-called hard data is very discrete. A geological strata model
is constructed by using these data appropriately and applying statistical methods.
Japan Oil, Gas and Metals National Corporation (JOGMEC) 59
– Development/production stage:
Multiphase flow analysis using distinct element method; oil–gas–water manage-
ment process design; unsteady multiphase flow analysis in pipeline; offshore
production system mooring/riser tensioner behavior analysis; dynamic behavior
analysis of floating production system with dynamic positioning system on
board; etc.
Specific Cases
One of the major achievements in petroleum and natural gas exploitation in recent
years is three-dimensional seismic exploration. In conventional two-dimensional
exploration, with data on a lattice grid at intervals of 500 m to a few kilometers,
data inside the gridlines is estimated using various interpolation methods, such as
spline interpolation or kriging. However, there is a high degree of uncertainty in
underground structure, which makes accurate evaluation difficult. Moreover, the
uncertainty is further affected by the surrounding underground structure, leading
to false images. Three-dimensional seismic exploration is implemented by placing
seismometers on a finer lattice with intervals of 12.5–15 m and recording the
reflected wave from every direction across the plane. This method can eliminate the
uncertainty accompanying interpolation, and increases the accuracy of underground
structure imaging by placing false images in the right position. However, capturing
the reflected wave on a plane increases the amount of seismic wave data recorded
considerably. A single survey conducted offshore from Abukuma over an area
of 70 km × 30 km by the three-dimensional geophysical survey vessel “Shigen”,
introduced by the Ministry of Economy, Trade and Industry (METI) in 2007 and
operated by JOGMEC, produced 12 terabytes of data. Optimization of parallel
computation and the algorithms used became essential to many processes, such as
noise reduction in the data, deconvolution to eliminate the effect of the hypocenter
(dynamite or air gun), and migration, which is a method for describing underground
structure based on seismic wave data (Fig. 9).
a finite number of cells, and the behavior of each portion of the reservoir is
estimated by solving the mass transport equations between cells for gas, oil, and
water phases using Darcy’s law. Multicomponent simulation is used for the more
complex interactions of carbon dioxide flooding and hydrocarbon miscible flooding,
treating the fluid as a mixture of several components and permitting component
62 H. Kawano
transfer between gas and liquid. In addition to calculating oil and gas phase flow
between cells, phase equilibrium calculations are performed. Variants of this type
of simulation include the thermal flooding model, which requires energy balance
equations, and the chemical flooding model, which expresses various chemical
phenomena. When it comes to multicomponent calculations with the order of a
few million cells requiring fast calculation and large memory, the calculation time
is large even with parallel computers. There are still problems with determining the
proper upscheduling, comparisons with actual production history, and automatic and
high-accuracy parameter calibration. A recent example of successful development
of a numerical simulator is a Japanese model that performed excellently in an
international competition for simulating the production of an unconventional natural
gas resource (methane hydrate). In this model, new routines were incorporated to
reproduce specific phenomena that had not been treated, such as melting gasification
of solid hydrate, rehydration, and consolidation of the unconsolidated sand layer and
changes in permeability.
In 2004, the Japan National Oil Corporation (JNOC) and the Metal Mining Agency
of Japan (MMAJ) were integrated to establish Japan Oil, Gas and Metals National
Corporation (JOGMEC) and ensure a stable supply of oil and natural gas, as well
Japan Oil, Gas and Metals National Corporation (JOGMEC) 63
as metal and mineral resources. Its primary roles are to implement exploitation
projects and technology development programs for promoting the exploration and
production of these resources, to provide necessary funding for exploration, and to
secure a stable supply of natural resources and energy through storage of petroleum,
natural gas and rare metals. JOGMEC also deals with protection of public health and
conservation of the living environment through mine pollution control measures,
and contributes to the sound development of metal mining.
Hirobumi Kawano was born in Tokyo in 1946. Following graduation from The
University of Tokyo with a B.D. in Economics, he joined the Ministry of Interna-
tional Trade and Industry, which was the predecessor of the Ministry of Economy,
Trade and Industry (METI). After holding director’s posts in the Petroleum Division
and the Basic Industries Bureau in the Agency for Natural Resources and Energy
(ANRE), he became Director-General of ANRE in 1999 until his retirement from
METI in 2002.
Kawano became an external director of Sony Corporation in 2003 and a senior
corporate officer of JFE Steel Corporation in 2004. He has been President of
JOGMEC since 2008.
His hobby is yacht racing. He is Vice President of the Japan Sailing Federation.
Iwane Laboratories, Ltd.
Waro Iwane
President and CEO
Iwane Laboratories
A Camera Vector (CV) value has six degrees of freedom and is derived from the
three-dimensional (3D) position T and 3D posture R of an object. The important
thing is that a CV value can be obtained by statistically analyzing a two-dimensional
(2D) image that is taken by a camera. This process is called CV computation.
All 2D images that have CV values (which are called CV images) make it
possible for us to conduct 3D space analyses on every object appearing in the
Y. Giga and T. Kobayashi (eds.), What Mathematics Can Do for You: Essays and Tips 65
from Japanese Industry Leaders, DOI 10.1007/978-4-431-54346-6 9,
© Springer Japan 2013
66 W. Iwane
images. Considering such images as perfect dioramas of the actual world, we are
allowed to identify 3D position or measure any size and distance between the
features.
When CV computation and CV images are applied to some technology for 3D
space analysis, we call such technology CV technology. CV technology is the
fundamental tool used in any solutions that our company provides.
Although CV computation can be performed with images taken by a normal
camera, we usually use an omni-directional camera to make the processing easier.
The picture taken by an omni-directional camera is of the 360◦ surrounding the
camera’s focal point (Fig. 1, left). We can then obtain a planar picture such as Fig. 1
(right) by applying an equirectangular projection to the spherical image.
Formula 4 holds at the feature points, of which there are usually hundreds.
Although the formula is very simple, thanks to the power of the notation it
becomes a huge system of equations when treated computationally. Moreover,
strictly speaking, there is no solution to the equations because there are more
equations than the unknowns and each of them has slight error. Instead, we process
the errors statistically and find a solution P, Tn , Rn , for which the error is minimized.
T R (P − T )
n n
∼
= Pn ···
4
The goal of our company is to achieve artificial intelligence (A.I.), and the
development of A.I. technology requires researching the function of the human brain
while describing the findings in mathematical language.
The fundamental ability required for A.I. is recognition ability. To recognize the
3D space around us, we must begin by describing this space mathematically.
Furthermore, to develop an A.I. recognition system, we must study recognition
in humans and describe it mathematically so that we can derive “solutions” from the
mechanisms.
Although we are only standing on the threshold of A.I. research, our next step is
to directly describe “intelligence” itself.
An important point is that the actual “intelligence” does not use mathematics,
although A.I. can be realized only with the aid of mathematics.
This reveals the uniqueness of mathematics, as well as its limitations.
So far, the “solution” has been obtained mathematically with the aid of a
computer. As a result, the 2D image can be analyzed as a 3D diorama, so that we
can understand the 3D space.
Aided by these methods, consequent images are analyzed automatically. Solving
the equations allows us access to 3D information extracted from the original 2D
images. We have developed the following products out of this research.
The Virtual Geographic Information System (VGIS) links together the CV images
and the Geographic Information System (GIS) digital vector maps. It is used for
improving road safety and in the operation and maintenance of roads. Since the
system allows 2D images to be used as 3D data, we can conduct 3D measurements
on the images, stabilize them, and add CG images to movies.
Iwane Laboratories, Ltd. 69
Fig. 3 3D-VGIS
WEBGIS (Fig. 4)
RCG (Fig. 5)
Fig. 4 WEBGIS
Fig. 5 RCG
Iwane Laboratories, Ltd. 71
Conclusion
In both our personal and professional lives, we cannot avoid scribbling letters and
writing sentences if we wish to express meaning. We can also use numbers, symbols,
formulas and equations to express universal truths, such as human thoughts and
natural phenomena, in an abstract manner.
Certainly, such thoughts and phenomena cannot be expressed without using the
power of mathematics. Since mathematical language is just as familiar as natural
language to us, however, we do not believe it has a special power. Rather, we tend to
see it as a trivial tool and sometimes do not even notice how vitally important it is.
However, we find it surprising that thoughts and phenomena themselves are made
from neither mathematics nor natural language.
Although thoughts and phenomena are not mathematics, they may be represented
by mathematics. It is important to remember that, for example, nature decides the
movement of an object without solving differential equations.
Our civilization could not have emerged without the invention of written
language, and we can say the same thing about mathematics.
It would be incorrect to think of mathematics as being used exclusively in
company activities. Rather, it is like air and water: both are indispensable for our
lives, but we rarely think about them.
When we encounter a difficult problem in human affairs, we often try to solve
it by asking questions in natural language, so that we are better able to understand
the essence of the problem. Similarly, when we wish to understand unknown natural
phenomena, we do the same thing with mathematical language.
When we take this approach, we use an everyday expression and say that we are
“solving equations.” However, although it may sound trivial, “solving equations” is
not a matter of indifference, but actually shows the unparalleled logical power of
mathematics.
Mathematics does not merely describe the problem in an abstract way, it allows
us to find a previously unknown “solution” from the abstract description. It is
surprising that the unknown can be transformed into the well known when we
succeed in describing the problem mathematically. This is a huge accomplishment
of mathematics that awakens our emotion.
As we have seen, running a mathematics-driven company is full of exciting
experiences. I am grateful for our human knowledge as well as our ancestors who
have developed mathematics to today’s level.
Acknowledgements I would like to thank Mr. Akifumi Arayashiki and Mr. Tomoyuki Wakakuwa
of our company, who provided valuable assistance while I was writing this chapter.
Waro Iwane was born in Sendai in 1943. After graduating from Yamagata University
with a B.D. in Physics in 1966, he started working as an assistant researcher at
the Research Institute of Applied Electricity (RIAE), Hokkaido University. His
achievements with RIAE include “Study on Information Processing Mechanisms
of the Human Visual System” and “Study on Band Compression Techniques in TV
Signal Transmission.”
He resigned from the institute in 1978 to start his own business. In the following
year, he established Iwane Laboratories and became the company’s President
and CEO.
Mathematics and I
Kaoru Yosano
Minister of Education (1994–1995)
Minister of International Trade and Industry (1998–1999)
Minister of Finance (2009)
After graduating from Azabu Elementary School, I went to Azabu Junior High
School. Today it is one of the most prestigious schools in Japan, but this was not
yet the case in my childhood. When I was in the seventh and eighth grades, I was a
poor student. One day my teacher called me out and said, “How long do you study
at home?” I said, “I work hard at school, so I don’t study at home.” At last, he said,
“You should study 15 minutes, at least.”
When I was an eighth grader, my father, who was a diplomat, was dispatched
to Egypt, and I was put into a British boarding school in suburban Cairo. From the
ninth grade, I started taking classes in English. Since the school rules were strict and
I could not understand English well, I had a hard time of it.
In those days, however, I made the wonderful discovery that math was easy when
taught in English. To solve a word problem in math, for example, you have to
ask yourself questions such as “what are the given conditions?” and “what is the
unknown value?” I could understand these processes more clearly than when I was
studying math in Japanese, because both math and English were languages that
emphasized the importance of clarity. After a short time, I became the best math
student in my grade.
I was not satisfied with only the official curriculum, so I sometimes bought math
textbooks myself and read them privately. It was a surprisingly big change for a
student like me, who had not studied at home in my home country. Among the
books I read in those days, for example, Do It Yourself Calculus gave me a good
understanding of calculus. I also became interested in physics and read George
Y. Giga and T. Kobayashi (eds.), What Mathematics Can Do for You: Essays and Tips 73
from Japanese Industry Leaders, DOI 10.1007/978-4-431-54346-6 10,
© Springer Japan 2013
74 K. Yosano
Gamow’s One Two Three . . . Infinity, in the eleventh or twelfth grade. While reading
these books, I experienced a lot of fun, as well as a vague sense of understanding of
the basics of math and science.
Hooked by Science
After spending 4 years in Egypt, I returned to Azabu High School as a twelfth grader.
I graduated from the Department of Law in The University of Tokyo, and then
started working for the Japan Atomic Power Company. Contrary to my expectation,
I was assigned to a research division in the engineering department. My supervisor
Ryukichi Imai was a graduate of the Department of Mathematics in The University
of Tokyo. He was the most intelligent man I had ever met. One day I complained,
“I do not understand what engineers are talking about,” to which he replied, “It is
nothing much to worry about. Anything will be clear if you study a little of it.” This
sparked my motivation, and I started learning nuclear power engineering.
I studied topics such as atomic fusion, fuel and, generally what was happening
inside the nuclear reactor. Some topics were beyond my understanding, as I lacked
the requisite mathematical background. For example, I did not understand the
diffusion equation used to describe the process of neutron diffusion. Despite these
difficulties, however, my interest in mathematics did not die. I studied hard on
nuclear power, and when I was tired of it, I enjoyed reading books about the history
of mathematics.
After I became a member of congress, I was interested in theories of elementary
particles and held onto the dream that I would settle down to study this field. Then,
shortly before my appointment as the Minister of Education, I started learning about
it from a book titled The Adventure of Quantum Mechanics (1991, Hippo Family
Club). This book contains many cartoons to help the reader’s understanding, and
was really a good book to start with, even though it did not describe every detail. So
I read other textbooks too, including Shin’ichiro Tomonaga’s Quantum Mechanics.
I was also interested in outer space and read works by Katsuhiko Sato, who was
known for studies of cosmic inflation. Although my understanding of quantum
mechanics and cosmology was very limited, I had a hunger for knowledge in these
fields.
There are two types of problem in the world. Problems of the first type can be solved
if they are attacked by a million people. The second type of problem can only be
solved by excellent people, regardless of how many there are. The proverb “Two
heads are better than one” is useless when it comes to the most difficult problems
such as proving Fermat’s Last Theorem. One million people of average intelligence,
Mathematics and I 75
working together, cannot reach the insight of one genius like Andrew Wiles. If we
want to apply the proverb, we have to alter it to “Two brilliant heads are better than
one equally brilliant one,” because brilliant discoveries only come from brilliant
brains.
I am not claiming, of course, that a few brilliant people can do anything. It is
true that a genius called Johannes Kepler discovered laws of planetary motion in the
17th century, but this discovery depended on the many anonymous people who kept
track of planetary movements each night and added to the huge amounts of data.
The best example is Tycho Brahe, a scrupulous scientist whose data led to Nicolaus
Copernicus’ heliocentric ideas, as well as to Kepler’s Law. Without such people, we
would have had neither Kepler nor Wiles.
To put it another way, there need to be a certain number of exceptionally brilliant
people for the efforts of ordinary people not to be wasted. If there was a revival of
so-called “elite education,” this would not be for the “elite” people themselves but
for society in general. In other words, we should raise elites and make them serve
the ordinary people.
I strongly agree that education should be given to all people equally. However, it
is also true that every child has both strengths and weaknesses. Some can run fast
while others can study well. We have to develop the abilities of talented children for
Japan’s science and economy to progress further.
Keep Learning
One of the reasons why I study math and physics is that I want to keep my brain
active as I get older. Mostly though it is because I have a hunger to understand how
the world is constructed. I am learning out of curiosity.
Math is quite fun when told as a story. For example, one day the following
question popped into my mind: “How did humans discover the general solution of
quadratic equations?” After reading a history book, I learned that it was discovered
in Arabia. The book also explained that the general solutions of cubic and quartic
equations were found in Italy, but that quintic equations were proved not to be
solvable by radicals. It is purely for fun to know such stories, so while some people
say that math is boring and has nothing to do with our amusement, I do not agree.
Kaoru Yosano was born in Tokyo in 1938. Following graduation from The Univer-
sity of Tokyo with a B.D. in Law, he joined the Japan Atomic Power Company. After
working as a secretary to Yasuhiro Nakasone, who would later become the Prime
Minister, he was in 1976 elected for the first time to the House of Representatives.
Known as one of the most distinguished policy experts, he held prominent
positions such as the Minister of Education in the Murayama cabinet, the Minister
of International Trade and Industry in the Obuchi cabinet and the chairman of the
LDP’s Policy Research Council in the Koizumi cabinet.
He was the Chief Cabinet Secretary in the Abe cabinet, and in the Aso cabinet
tripled as the Minister of Economic and Fiscal Policy, the Minister of Finance and
the Minister of Financial Services.
He is a grandchild of the noted poets Tekkan and Akiko Yosano.
Mathematics for Industry: Principle, Reality
and Practice, from the Point of View of a
Mathematician
Masahiro Yamamoto
Professor,
Graduate School of Mathematical Sciences,
The University of Tokyo
We describe the principles of mathematics for industry and discuss schemes for
promoting activities. Then we present two cases of the practice of mathematics for
industry and argue on a bilateral collaboration style for effective joint projects by
mathematicians and industry.
Introduction
Originally mathematics was not only aimed at abstraction but also was applied to
solving real problems, and it has been developed in both respects. We can understand
that if we recall several prominent mathematicians such as Archimedes (287–217
BC) and J.C.F. Gauss (1777–1855). They worked for both abstract mathematics
and real applications of mathematics. In particular, Gauss invented the least-squares
method, which is a major numerical method nowadays for various problems.
The least-squares method is the basis of the Tikhonov regularization, which is a
fundamental method for inverse problems as explained in section Case Study I: Risk
Management and Innovation for Operation of a Blast Furnace.
Reviewing mathematical history, we can recognize that great knowledge of
mathematics, which was originally exploited only for itself and not necessarily
aimed at applications, has eventually played essential roles in other disciplines, as
in quantum physics and the theory of relativity. Thus it is natural to expect that
mathematics can work for industry. This is “mathematics for industry”.
Y. Giga and T. Kobayashi (eds.), What Mathematics Can Do for You: Essays and Tips 77
from Japanese Industry Leaders, DOI 10.1007/978-4-431-54346-6 11,
© Springer Japan 2013
78 M. Yamamoto
For solving problems in the real world, not only a knowledge of mathematics is
useful, but also the mathematician is meaningful: the mathematician uses specific
ways of thinking such as abstraction, generalization, and extraction of truth from
various realities to crystallize to a simple statement as a theorem. Moreover, in
various applications, mathematicians do not merely apply knowledge in an existing
specific area of mathematics, but can build up a new theoretical system of math-
ematics for solving concrete problems. Therefore, these intrinsic characteristics of
mathematicians may frequently yield breakthroughs for interdisciplinary projects.
In particular, we can expect innovation in industry and more economic profits by
mathematics after proposals of problems from industries, because:
• The mathematical approach may be different from conventional cooperative
research, and we can find more fundamental solutions by mathematics which are
based on principles and theorems. The ways of thinking of mathematicians are
less dependent on existing hardware and methodologies, compared with those
of engineers, and can be expected to propose initiating ideas for flexible, non-
conventional, and innovative methods.
• With the help of the universality of mathematics, we can find general solutions
in various issues and themes which emerge across various industrial aspects.
Therefore we can solve several industrial problems simultaneously. Such help
can be offered only by mathematics.
• The activity range of mathematicians is totally international. With such inter-
national networks of researchers, if an industrial problem is suitably given to
mathematicians, then the mathematicians can invite members from their human
resources to place the right persons in the right places and to activate their
knowledge and accomplish the mission.
• Recently we can collect more data and we have to treat big data which increase
day by day. Among the collected data, noisy data or meaningless data may be
mixed in, and we should often select more meaningful data among the total data
set. For this, we can expect new mathematical models on the basis of geometry,
analysis, etc., which can distinguish relevant data sets from meaningless data and
make more effective analyses and designs. In particular, by more collaboration
Mathematics for Industry: Principle and Practice 79
Next we can analyze possible reasons for the current status in Japan as follows.
1. There is a deficiency of know-how on industry–university cooperation: a diffi-
culty in building constant credibility between industry and universities.
2. There is a small number of mathematicians as coordinators. Here, by a
“coordinator” we mean a person who can understand not only the requirements
about time-frame and desired outputs from industry, but who also can exercise
strong leadership within research groups and networks, which can build bridges
between industries and mathematicians. The same is happening on the industrial
side. As a result, research projects tend to be carried out with a unilateral
direction and do not last for a long period of time.
3. The timescale mindset on the university side is much slower than on the industry
side after proposals of certain missions from industry. The university side is too
much inclined to regard a topic put forth by the industry side as a part of the
educational and recruitment activities in graduate schools. Mutual understanding
about timescales and purposes by both sides is missing.
4. There is no flexible system on the university side which enables mathematicians
to solve missions presented by the industry side.
80 M. Yamamoto
5. In Japan, there are not many post-doctoral mathematicians who are competent
in theory, programming, and computation. That is, in all three, a great number
of students at departments of mathematics are not well motivated to apply
mathematics to industrial missions, although their mathematical abilities are very
high.
6. The actual results delivered through industry–university cooperation have not yet
been sufficiently evaluated on the university side. That is, there is a gap between
evaluation standards in the university and those in industry: the standard of
evaluation within the university is based on publications in international journals,
while the standard within industry is usually by patents or real use. Moreover, it
may not be possible to publish some achievements in view of confidentiality
requested by the industrial side, so that the achievements by mathematics may be
invisible.
These difficulties may exist everywhere in industrial countries, but in the
countries where these are better solved, activities by mathematics for industry are
conducted more systematically to yield constant outputs. One definite solution is
organization of institutes of mathematics for industry.
The author proposes short-term policies for overcoming these difficulties from
a general viewpoint, which will be conducted appropriately by the presence of
institute of mathematics for industry:
• Assigning mathematician–coordinators in the above sense.
• International–interdisciplinary networks conducted by the coordinators in order
to solve missions within a requested time limitation, by inviting appropriate re-
searchers on appropriate spots. This network is needed for overcoming deadends
or bottlenecks which may appear during limited-duration research.
• Pool of young talent: We should always provide suitable opportunities for
graduate students in order for them to understand problems in the real world,
as in industry.
• Understanding of time frames and scheduling by both the university side and the
industry side.
Example: It takes a huge amount of time to translate real-world problems into
mathematical language. Even for bright mathematicians, it may take a substantial
amount of time for solving this issue.
• Understanding by the industrial side of the value of fundamental research in
various mathematics, because we cannot predict which mathematics will be
useful.
In the previous section, we briefly stated policies for overcoming the difficulties.
Needless to say, for their realization, we need more concrete efforts. As one
example, we mention current efforts at Graduate School of Mathematical Sciences
Mathematics for Industry: Principle and Practice 81
of The University of Tokyo by the Global COE Program ”The Research and
Training Center for New Development in Mathematics,” which started in 2008. The
GCOE Program contains the following activities resulting from the financial support
and human resources of the program.
Since 2009, we have offered a course of mathematics in industry every year for
graduate students, and a student can gain credit units from attending. The course
is shared with about seven speakers from IT, insurance, finance, and manufacturing
companies. The purpose is for graduate students to see how mathematics is applied
in various actual worlds of industry and to make better plans for future career paths.
Since 2011, we have organized guided tours to plants and institutes of Kao
Corporation and Nippon Steel Corporation. The purpose is not only to see the plants,
but also have the students present overviews of their own research to members of
the visited companies, which may be useful in considering future possible industrial
collaboration.
4. Research Internship
Since 2011, two graduate students have stayed 2 or 4 weeks at Advanced Tech-
nology Research Laboratories of Nippon Steel Corporation (at that time) and
worked for mathematical solutions and formulation of real problems posed by the
Mathematics for Industry: Principle and Practice 83
company. The internship for recruitment is common but this internship is aimed at
mathematical activities by graduate students in company. We expect development
of this program in the next year.
As Professor Masato Wakayama writes in this volume, there are comprehensive
activities at Kyushu University organized by Institute of Mathematics for Industry
and the GCOE Program ”Education and Research Hub for Mathematics-for-
Industry.” Also at Graduate School of Mathematical Sciences of The University of
Tokyo, we are substantially expanding related activities. The combination of these
multiple activities should be useful for promoting mathematics for industry involv-
ing graduate students and conducting an open platform for industrial collaboration
with mathematicians.
The author has been working on inverse problems. By inverse problem, we mean
a problem of determining a cause or detecting invisible objects by available data.
For example, medical diagnosis and non-destructive testing are inverse problems.
Also one can schematically understand it in Fig. 1. We can interpret “input” as
a cause or inaccessible object, and “output” as results or measurement data. The
“model” describes the law which governs the phenomenon and connects input
to output, and is often given by a differential equation. The inverse problem is
the determination of an input and/or a system by an output, while the forward
problem is the determination of output for a given system and input, which is
typically a prediction of the future state of the system. Moreover, a problem of
designing the input or system in order to realize the output, is usually called a
control problem. The inverse problem and the control problem are based on the
same concept. A tremendous number of problems in industry can be considered as
inverse problems. Since the inverse problem is a determination problem of some
states by incomplete data, the inverse problem commonly has intrinsic instability.
That is, even if noises in measurement data are small, huge deviations may occur in
resulting solutions reconstructed by noisy data.
Already we have considered the general advantages of mathematics for in-
dustry and here we add other advantages of mathematics of inverse problems
for industry.
• By industrial inverse problems, one can expect nice cost performance by the
participation of mathematicians.
In the inverse problem discussed in section Case Study I: Risk Management
and Innovation for Operation of a Blast Furnace, the demanded mission is to
A blast furnace controls a basic process for the iron-making process, which produces
molten iron from sintered ore and coke through the deoxidizing process. The
inside temperature distribution is highly nonhomogeneous and is around 1,500 ◦ C.
Mathematics for Industry: Principle and Practice 85
Fig. 2 Cross section of blast furnace: raw data by thermocouples, reconstructed heat-flux [2]
A large-size blast furnace is about 100 m high in total and produces approximately
12,000 tons per day and, moreover, should work without any break down for more
than 15 years.
We are concerned with the basement of the furnace, which is covered by special
bricks, and the brick directly touches the molten iron (see Fig. 2a from Nakagawa
and Yamamoto [2]). The brickworks are expected to withstand the high temperature
and are designed to protect the plant and avoid damage to the furnace. It is extremely
necessary for us to estimate the heat-flux on the inner surface of the brickworks
because the interior heat-flux can be a good index for a secure controlled process.
If the heat-flux behaves irregularly, then one has to reduce the activity level of the
furnace. By considering the daily production output (i.e., about 12,000 tons per day),
the reduction of the activity level implies significant economic loss, and should be
avoided as long as safety is assured.
However, direct observation of the interior heat-flux is impossible owing to the
size and structure of the furnace and the high temperatures, so that it is possible
only to estimate them by temperature data near the bottom of the furnace which are
observed by thermocouples (Fig. 2a).
When irregular behavior of such temperature data is observed (e.g., Fig. 2b,
between 600 and 2,000 h) and, considering such an abnormal state as an omen
of worse irregularity, the furnace engineers conventionally carried out shutdown
operations for reducing the furnace activity (Fig. 2b from [2]: especially during the
86 M. Yamamoto
Inside the blast furnace, complex processes go with multiphases involving solid
and liquid phases under high temperature, and the modeling itself is a very difficult
problem. However, our main purpose is to find an index for a secure operation of the
furnace, not to know the inside process itself. Thus our strategy for establishing a
model equation is the choice of a minimal model explaining raw data of observations
by thermocouple. Indeed we simplify the shape of the concerned part of the blast
furnace (Fig. 2a) and choose a surprisingly simple one-dimensional heat equation:
where the constant α > 0 is the thermal diffusion coefficient, 0 < x < is the depth
variable of the brick, and x = and x = 0 correspond respectively to the bottom
end of the brick and the interface with the molten iron. We take finite differences
of measured data by two thermocouples located in the depth direction and we can
approximate the derivative ∂∂ ux (,t), 0 < t < T . Therefore our available measurement
data are
Mathematics for Industry: Principle and Practice 87
The inverse heat conduction problem belongs to a category of inverse problems for
partial differential equations.
The first mathematical subject is uniqueness. Uniqueness means: h(t) := u(,t)
= 0 and g(t) := ux (,t) = 0, 0 < t < T imply f (t) := −ux (0,t) = 0, 0 < t < T for
u satisfying (1). Uniqueness guarantees that there exists at most one heat-flux f at
x = 0 realizing given data g, h. We note that there may not exist such f . The proof of
88 M. Yamamoto
√ −2μn T
while fn = ε n n + 12 π 1−e √
2 μn
−→ ∞ as n → ∞. By choosing ε > 0 suffi-
ciently small, the solutions fn become very large as n → ∞ even if the data gn , hn are
small for all n, so that the inverse heat conduction problem is instable.
Instability can be understood intuitively by the fact that the heat conduction
process has a strong averaging effect of the temperature distribution: even if
the variations in f (t), 0 < t < T , are very large, the specific profile of f (t) is
greatly smoothened during conducting from x = 0 to the measurement point x = .
Therefore we have to reconstruct the original f (t), 0 < t < T by such smoothened
filtered data g(t) and h(t), and any large deviation in f (t) must be detected from
very small changes in observation data of g, h.
Remark 2 (The Restoration of Stability). It is known that if the uniqueness holds
for an inverse problem and the map from inputs to outputs is continuous, then we
can restore stability provided that solutions are restricted to a suitable bounded set.
In the above case with the L2 -norm, we can more precisely state it as follows. We
define a space L2 (0, T ) by
Mathematics for Industry: Principle and Practice 89
T
L (0, T ) := g;
2
g(t) dt < ∞
2
0
1
T 2
with the L2 -norm g := g(t)2 dt . Let M ⊂ L2 (0, T ) be a compact subset.
0
For example, as M we can choose g; g + dg dt ≤ C with an arbitrarily fixed
constant C > 0. Let un = un (x,t), n = 1, 2, 3, . . ., satisfy (1)–(3) with gn and hn . Then
we can conclude that
∂ un
(0, ·)
lim
n→∞ ∂ x =0
if limn→∞ ( hn + gn ) = 0 and
∂ un
(0, ·) ∈ M . (4)
∂x
This is a consequence of a general theorem in the topology which is also called the
Tikhonov theorem. This result implies that we can restore stability under an a priori
condition (4), and is called the conditional stability. For many inverse problems, we
can prove the conditional stability, and it is an important and attractive mathematical
subject to prove it with a concrete stability rate. We do not state the details.
We will give a sketch of our numerical method, assuming for simplicity that
g(t) ≡ 0. For the details, see Wang, Cheng, Nakagawa and Yamamoto [4]. The keys
are
• The Fourier method
• Discretization
• The Tikhonov regularization
For the initial/boundary value problem (1), (3) with given f (t) := −ux (0,t) and
initial value u(x, 0) = a(x), 0 < x < , we can use the Fourier method. By the Fourier
method, one can find the solution to the initial/boundary problem as superposition
of the trigonometric functions in x, which was established in 1822 by J.B.J. Fourier
(1768–1830) and is widely applicable to linear partial differential equations. The
Fourier method is classical and every undergraduate student must know it, and
thanks to our choice of a minimal model, we can apply it to write the solution as
follows. We set
n2 π 2
λn = α , n≥0
2
90 M. Yamamoto
and
2 ∞ −λn t nπ x nπ y 1
G(t, x, y) = ∑ e cos cos + .
n=1
Then we have
∞ t
u(x,t) = ∑ An (x)e−λnt − 0
G(t − s, x, 0) f (s)ds (5)
n=0
Now let us fix discrete times: 0 < t1 < · · · < tM = T . Our numerical task is to
construct approximate solutions of f (t) and a(x) := u(x, 0) by data g(t j ), h(t j ), j =
1, 2, . . . , M. We discretize the inverse problem. That is, first we truncate the series in
(5) to obtain
N tj
u(,t j ) ∼ ∑ An ()e−λnt j − 0
G(t j − s, , 0) f (s)ds
n=0
N j ti
= ∑ An()e −λn t j
− ∑ f (ti )
ti−1
G(t j − s, , 0)ds
n=0 i=1
ti−1
N j ti
= ∑ An()e −λn t j
−∑
0
G(t j − s, , 0)ds −
0
G(t j − s, , 0)ds f (ti ).
n=0 i=1
We set
i (x,t) = − 0t G(t − s, x, 0)ds, 0 < t ≤ ti ,
G
− 0ti G(t − s, x, 0)ds, t > ti
and
G i (,t j ), j ≥ i,
j,i := G j,i = 0, j < i.
G
Then we can rewrite the above approximate equation with respect to u(,t j ) to
obtain a system of linear equations:
PX = W,
Mathematics for Industry: Principle and Practice 91
where P = {P1|P2 },
⎛ ⎞
e−λ0t1 e−λ1t1 ··· e−λN t1
⎜ e−λ0t2 e−λ1t2 ··· e−λN t2 ⎟
⎜ ⎟
⎜ ⎟
P1 = ⎜ e−λ0t3 e−λ1t3 ··· e−λN t3 ⎟
⎜ ⎟
⎝ ··· ··· ··· ··· ⎠
e−λ0tM e−λ1tM ··· e−λN tM
⎛ ⎞
1,1
G 0 0 ··· 0
⎜G ⎟
⎜ 2,1 G2,2 − G2,1 0 ··· 0 ⎟
⎜ 3,2 − G3,1 G3,3 − G3,2 ⎟
P2 = ⎜ G G ··· 0 ⎟
⎜ 3,1 ⎟
⎝ ··· ··· ··· ··· ··· ⎠
M,1
G M,2 − G
G M,3 − G
M,1 G M,2 M,M − G
··· G M,M−1
and
⎛ ⎞
A0 ()
⎛ ⎞ ⎜ A1 () ⎟
u(,t1 ) ⎜ ⎟
⎜ ··· ⎟
⎜ u(,t ) ⎟ ⎜ ⎟
⎜ 2 ⎟ ⎜ ⎟
⎜ ⎟ ⎜ AN () ⎟
W = ⎜ u(,t3 ) ⎟, X =⎜ ⎟.
⎜ ⎟ ⎜ f (t1 ) ⎟
⎝ ··· ⎠ ⎜ ⎟
⎜ f (t2 ) ⎟
u(,tM ) ⎜ ⎟
⎝ ··· ⎠
f (tM )
We have tested our numerical method for a miniature model built at a laboratory
of Nippon Steel Corporation and we have confirmed that our numerical method can
give better numerical results than other existing methods. Thus we can conclude that
our method is reliable.
Mathematics for Industry: Principle and Practice 93
We note that one can never know the interior status, but the reconstructed interior
heat-flux can be used as an index for detecting indications of potential dangers and
that the reconstruction is by an inverse problem method.
1. Our problem is an inverse problem and we cannot expect stability in the usual
sense. Therefore we have to apply a specific method called the Tikhonov
regularization.
2. For a realistic numerical method, we need other careful consideration for the
choice of discretization. In view of the intrinsic instability and accuracy of
available real data, we should not exploit any numerical methods which yield
highly accurate numerical solutions if data can be very accurate. In general, a
numerical method of high precision may well reflect the original instability and
is too sensitive to data, and the numerical results may not be reliable.
3. In many applications, available data are not of high precision. In the case of
the furnace inverse problem, the available data are not highly accurate because
the number of thermocouples is small compared with the size of the furnace.
Thus by such realistic situations, our numerical method should produce results
of reasonable precision even from medium-accurate data.
4. The Tikhonov regularization is a relevant method to produce reasonable nu-
merical results from not very accurate data. For it, the clever choice of the
regularization parameter γ is essential in view of the conditional stability, original
noise levels of data, and the accuracy of discretization. The key to the optimal
choice is the “golden mean,” that is, “Too much is as bad as too little.”
5. Joint research of the inverse heat conduction problem in industry is composed of
the three different steps:
(a) The mathematical part about the uniqueness and the conditional stability.
(b) The development of relevant numerical methods.
(c) The adjustment of the numerical method to real use.
Both mathematicians and industry are often inexperienced with part (c) (see
Fig. 3).
6. For the inverse problem, the most important mathematical subject is unique-
ness and then conditional stability. We note that usually the existence of the
input, realizing given output, is not necessary to be considered in the inverse
problem, because we are discussing a phenomenon which is clear and present.
Reconstruction by the numerical method on the assumption of its existence,
is much more important than the existence issue. For the well-posed forward
problem such as an initial/boundary value problem for the heat equation (1),
we note that uniqueness, existence, and stability are fundamental mathematical
Mathematics for Industry: Principle and Practice 95
h(t) is constant, which is below the melting temperature of the workpiece material
and is independent of t. However, when the thickness of the workpieces is not
uniform, a constant target temperature h(t) does not yield the uniform penetration
depth of the laser beam along curves under consideration. In other words, in a place
with little thickness, the constant power is too strong and the penetration is too
much, and one cannot have a product with uniform hardening. Therefore we need
to determine the laser power p(t) as a function in t in order to realize the set-up
temperature h(t) over some time interval 0 < t < T .
We formulate this problem mathematically, omitting the details. For a workpiece
with a flat upper face, by Ω we denote the flat face where the laser beam is radiated.
Thus we assume that Ω is a domain on the plane. In Fig. 4, the upper side of the
workpiece is the cross section of Ω in the depth direction and the laser is radiated
along a curve on the planar domain Ω . By x = (x1 , x2 ) we denote a point on the
plane with suitable orthogonal coordinates. The temperature evolution u(x,t), x ∈ Ω ,
0 < t < T is described by the following heat equation:
2
∂ 2u
ut (x,t) = ∑ ∂ x2 (x,t) + F(u(x,t)) + p(t)G(x,t), x ∈ Ω , 0 < t < T, (7)
k=1 k
Here (ν1 (x), ν2 (x)) is the unit outward normal vector at x to ∂ Ω and (9) describes
no heat-flux through the boundary ∂ Ω . In (7)–(9), all physical constants are
normalized. We assume that the absorption of laser energy per unit area is modeled
by p(t)G(x,t) where G(x,t) is a normalized intensity profile and we suppose that
we know G(x,t) physically. In the case of a gas laser, the function G is of a
bell shape like a Gauss function. Moreover, high heating induces further effects
inside the workpiece such as solid - solid phase transitions and for the modeling,
we introduce a nonlinear function F in (7) satisfying suitable conditions. Now our
problem of achieving the uniform hardening is stated as follows:
Let γ (t), 0 < t < T , be a curve in Ω where the laser beam is radiated and let
h(t), 0 < t < T be a given target temperature. Then determine the laser power p(t),
0 < t < T , such that u(γ (t),t) = h(t), 0 < t < T .
We consider p(t) and u(γ (t),t) respectively, as input and output, as explained in
Fig. 1. This is a kind of inverse problem: we do not know p(t) in the heat source
term in (7) and we are requested to determine p(t) by means of desired output h(t),
0 < t < T.
We can solve this inverse problem by a modern theory of partial differential
equations called the semigroup theory, but we omit the mathematical details.
The semigroup theory has been developed since the 1940s mainly by Japanese
Mathematics for Industry: Principle and Practice 97
mathematicians, one pioneer of whom was Professor Kôsaku Yosida, and its
systematic applications to inverse problems were done mainly in the Soviet Union.
We show numerical examples for the surface hardening of a workpiece with a
rectangular cavity (the lower part of Fig. 5 from [1] indicates a workpiece with
an originally created cavity shown in white). Figure 5 is a result of a numerical
simulation with constant laser power p(t) = p0 and constant velocity, that is, the
power is not chosen by an inverse problem method.
The top part of the figure indicates the temperature uhs of the hot spot on the
surface, while in the lower part the black shows a hardened surface layer (more
exactly speaking, this is the resulting fraction of austenite). This case is not good
because (a) as the graph of uhs (t) shows, the high temperatures above the cavity and
at the end of the workpiece lead to exceeding surface melting, and (b) the too-thick
hardening above the cavity may foster fatigue effects which may eventually lead to
destruction.
Next we show a hardening process with the inverse problem. In Fig. 6 [1], the
profile uhs is the temperature of the hot spot on the surface, and our control by
solving the inverse problem makes uhs lower above the cavity in order to avoid
exceeding surface melting. On the other hand, by usub we denote the temperature
profile close to the lower end of the hardening depth. Thanks to the inverse problem,
usub is uniform, which guarantees highly uniform hardening. Figure 7 [1] is a
photograph of a real product where the white layer near the upper surface is the
unform surface hardening part and the inside black part is of course the cavity. The
depth of the white layer does not depend on the locations, whether above the cavity
or not.
Am I permitted to write that Fig. 7 is evidence of one triumph of mathematics
for industry as well as the achievement described in section Case Study I: Risk
Management and Innovation for Operation of a Blast Furnace?
98 M. Yamamoto
Conclusions
be used. Mathematicians will be useful for industry provided that they maintain
a high level of mathematical knowledge and experience.
• Collaboration with industry frequently inspires mathematical research and even
creates new mathematical subjects.
• Bilateral collaboration between mathematicians and industry is highly desirable
and can produce a win–win relation. One scheme of such bilateral collaboration
is indicated in Fig. 3.
• Serious activities recently started in Japan for promoting mathematics for
industry. The above two greatly successful cases were based on more individual
grounds before those activities were begun, and so we can greatly expect
much more output from our current systematic efforts as described in the
previous section. On this occasion the author asks industry to remain interested
in mathematics and support the above-mentioned activities, and ask graduate
students to join the activities with a positive attitude to gain a wider vision.
References
Masato Wakayama
Professor,
Director of the Institute of Mathematics for Industry
Kyushu University
Y. Giga and T. Kobayashi (eds.), What Mathematics Can Do for You: Essays and Tips 101
from Japanese Industry Leaders, DOI 10.1007/978-4-431-54346-6 12,
© Springer Japan 2013
102 M. Wakayama
A few decades ago, before computing capabilities became so great, there were
many Japanese engineers and professors, in particular, in such fields as mechan-
ical engineering, aeronautical engineering, precision instruments engineering and
electrical engineering and, of course, physicists who could function as applied
mathematicians when visiting Western countries. In addition, in the past, the relation
between mathematics and engineering/technology in some sense holds via scientific
fields, in particular, physics. Today, however, with the development of high-speed
computing, the direct connection of mathematics with all fields of science and
engineering has increased strikingly, and mathematics has come to play the role
of both the language and the guiding framework for research in an unprecedentedly
broad range of scientific, engineering and technology fields. However, despite this
trend, it has been found that with the vast and rapid development of computer
software (mainly in foreign countries), the number of researchers in scientific
fields with a strong interest in mathematics has decreased greatly. Also, while this
trend is seen even in industrialized Western countries, it is more striking in Japan.
Given this situation, the training of young researchers with mathematical skills
necessary to solve real-world mathematical problems should be given a priority,
both domestically and internationally [1, 3].
In Japan, in the past, the education of graduate students in mathematics and
mathematical sciences took place primarily within the mathematics departments in
Faculties of Science. Also, before the shift in Japan, which was initiated by the
Government, to an increased role of graduate school education, the major focus of
such mathematics programs was the training of college instructors and professors
in mathematics. In those days, the number of students who entered Ph.D. programs
was extremely small, and even the number of people entering master’s programs was
very small, in comparison to other scientific fields. This trend continued even after
the shift to a greater emphasis on graduate school education. However, now, even
in mathematics and the mathematical sciences, following the government policy to
increase the number of people with Ph.D. degrees, there has been a large increase
in the number of students entering graduate school, including in Ph.D. programs.
This would seem to suggest that a balance has been attained between the potential
demand for mathematicians with industry, scientific or technological experience
and the number of graduates coming from Ph.D. programs in mathematics and
mathematical sciences, as specified by government policy. In relation to this, it
should be noted that, according to an article [4] in SIAM News [SIAM: the
world’s largest industrial and applied mathematics academic society, the Society
for Industrial and Applied Mathematics (USA)], in Brazil, which is experiencing
remarkable economic development, there are presently about 150 graduates from
Ph.D. programs in mathematics and applied mathematics per year, but this number
is insufficient to support the industrial development of that country [4]. For that
reason, plans are now being developed to increase the number of Ph.D. graduates
while maintaining the quality of their training. (For additional historical comments
about Japanese mathematics, research and education, see also [5].)
Importance and Unpredictable Effectiveness of Mathematics 105
One of the intelligent ideas for promoting scientific research and technology
development was proposed by Hendrik Bruget Casimir (1910–2000). Casimir was
a Dutch physicist, a student of Paul Ehrenfest at Leiden, and an excellent leader of
industrial research at Royal Philips Electronics. He made fundamental contributions
to the foundations of quantum mechanics and solid state physics, and is also quite
well known in mathematics communities for the Casimir operators. He is actually
well known for predicting a true effect of quantum mechanics; i.e. even in the
seemingly emptiness of the vacuum between two metal plates an electromagnetic
force is generated that pulls the two plates toward each other. After the prediction,
now referred to as the Casimir force, it was necessary to wait for some 60 years
before the experimental verification was achieved by Lamoreaux [6] in 1996
with 1% accuracy. It is worth noting that, in order to make the prediction he
computed some divergence series, which is equivalent to computing the special
value of the Riemann zeta function ζ (s) at s = −3 (Leonhard Euler computed this
divergent series around 1749 before the explicit formulation of the notion of analytic
continuation) (Fig. 1):
106 M. Wakayama
Fig. 1 Euler’s moon and sun; functional equation of the zeta function
1
ζ (−3) = “1 + 8 + 27 + 64 + 125 + . . .” = .
120
From the fall of 2009, DARPA (the U.S. Defense Department’s Defense
Advanced Research Projects Agency) has launched a project encouraging scientists
to work on ways to manipulate this quirk of quantum electrodynamics and recent
research has already found that the manipulation of the Casimir force is a major
challenge.
It was already realized by the middle of the 1980s that innovation is not linear and
certainly not driven only by experimentation and theoretical research in science but
is also strongly driven through the exploitation of technology. This observation had
already been made by Casimir in the seventies at the time when the linear model was
still considered largely valid. Technology uses science with a time delay of about 10
years, he argued. In turn, science is driven by new development in technology. And,
both progress together. This is the so-called Casimir spiral and is a key component
of innovation today (Figs. 2 and 3). Of course, the spiral pitch of the scientific and
technological Casimir spiral differs considerably depending on the field (see [7]).
As described above, with the remarkable increase in the capabilities of computers
that has occurred in the last two decades, the era in which the contribution of
mathematics to technology is communicated only through physics and related fields
Importance and Unpredictable Effectiveness of Mathematics 107
Inspired by the Casimir effect (which is also called the vacuum force), we composed
the following waka ( : a 5-line 31-syllable Japanese poem) by the standard
technique called the “adaptation of a famous poem .” It exploits the fact
−s −1
that ζ (s) is given by the infinite product ∏ p (1 − p ) over all prime numbers p.
108 M. Wakayama
One may notice immediately that this is actually an adaptation of the following
famous poem composed by Fujiwara no Toshiyuki Ason :
Fujiwara no Toshiyuki Ason (birth date unknown −901 or 907) was a middle
Heian waka poet and nobleman. This poem is the 169th of the 1111 poems in the
Kokin Waka-shu , which is an early Heian waka Imperial anthology,
envisioned by Emperor Uda (r. 887–897) and ordered by his son Emperor Daigo
(r. 897–930) in approximately 905. The number 169 = 132 is reminiscent of a prime
factor of 1859, the year of the Riemann Hypothesis’ birth (Fig. 4).
The meaning of the title “Mathematics for Industry (MfI)” is different from that
for “Mathematics in Industry”. That for the latter phrase is restrictive—it tends
to be identified with the actual mathematics that specifically arises in the daily
management and operation of manufacturing. The former, however, denotes a new
Importance and Unpredictable Effectiveness of Mathematics 109
research field in mathematics that may serve as a foundation for creating future
technologies. This title/notion was born from the integration and reorganization of
pure and applied mathematics into a fluidic and versatile form capable of stimulating
awareness in industry of the importance of mathematics as well as responding to the
needs of industrial technologies. The history of this integration and reorganization
has led us to the belief that this basic idea will someday find increasing utility. (See
[9] for interesting and important observations about the application of mathematics
in industry.)
The Institute of Mathematics for Industry (IMI) was established in April 2011
as Kyushu University’s fifth research institute. IMI was specifically planned to
be an offshoot of the Global COE Program “Education and Research Hub for
Mathematics-for-Industry” (FY2008–2012). [MEXT promotes the Global COE
(Centers of Excellence) program which provides funding support for establishing
education and research centers that perform at the apex of global excellence to
elevate the international competitiveness of the Japanese universities in various
fields of study.]
The idea behind establishing IMI can be traced to the 21st Century COE Program
“Development of Dynamic Mathematics with High Functionality” (FY2003–2007).
The aim of this program—a precursor to our Global COE Program MfI—was to
provide a broad view about mathematics to doctoral students and to open new
career paths for them. This is being achieved through, for example, an introduction
into the program of long-term internships in industry and the establishment of
the Mathematics Research Center for Industrial Technology (MRIT, April 2007–
March 2011), which is one of the common-use education/research centers at Kyushu
University. Since its birth in 1939, the Department of Mathematics in the Faculty of
Science at Kyushu University has enjoyed a long tradition of respect for striking a
harmonious balance between pure and applied mathematics. This balance serves as
the framework for the organization and management of the IMI.
Kyushu University has instituted a Graduate School/Graduate Faculty System.
According to its rules, University (faculty) members, in principle, belong to a
Graduate School or a Research Institute. A Graduate School is an educational
organization (education body) within the different graduate school of studies. IMI
and the Faculty of Mathematics are the partner bodies responsible for the Graduate
School of Mathematics.
Beside the standard educational activities at the graduate school etc, the ac-
tivities of IMI include (a) promoting collaboration with industry, (b) organizing
international conference, workshops and study groups, (c) holding seminars for
industry-academia partnerships and mathematical tutorials, and (d) fostering under-
graduate, graduate and Ph.D. programs which deliver students with a comprehensive
mathematical understanding of the two-way linking between mathematics and
110 M. Wakayama
Fig. 5 Development of
mathematics
Real-World applications (the Casimir-spiral for mathematics (Fig. 5)). The insti-
tute has 27 regular faculty members and consists of three main divisions plus
one laboratory: Advanced Mathematics Technology; Applied Mathematics and
Fundamental Mathematics; Laboratory of Advanced Software in Mathematics. In
addition, in order to efficiently promote the above projects/activities, a Visiting
Scholars Division and a Partnership Promotion and Technical Consultation Room
have been established.
As the reader can understand from the organizational structure depicted above,
IMI is staffed by both applied mathematicians (specializing in such traditional
fields as numerical analysis, partial differential equations, integrable systems and
probability theory, as well as more modern fields, such as data encryption, nu-
merical simulations, statistics and optimization) and mathematicians specializing in
traditional pure mathematics fields (topology, differential geometry, function theory,
representation theory and algebraic analysis). Most of these members, including
those with backgrounds in pure mathematics, conduct collaboration in joint research
projects which involve researchers in industry. It is particularly noteworthy that
the pursuit of this type of research activity is in perfect alignment with the fact
Importance and Unpredictable Effectiveness of Mathematics 111
fostered the development of the new theoretical field of information criteria. This
in turn has led to great advances in the field of statistics and its applications.
3. It was thought for several hundred years that the field of number theory had
no applications. However, this is no longer the case. Number theory is of
vital importance in modern information security technology, which is based on
cryptographic theory. The genesis of cryptographic theory lay in the observation
of the computational hardness of the factorization of integers into prime numbers.
The theoretical need provided by such information technologies including data
compression and code theory, has stimulated progress in number theory and
algebraic geometry, which serves as the foundation of the latter-mentioned
research fields. To keep pace with advances in these developments, both number
theory and algebraic geometry have indeed achieved a highly advanced level of
abstraction that certainly far exceeds the dreams of Gauss (widely considered
the greatest mathematician of all time). Thus, in a manner that no one could have
predicted, pure mathematics has realized a great application that took 2,500 years
for the making, and through feedback from this application, pure mathematics
has itself advanced. This is an archetypical example of the cooperative advance
of mathematics and technology.
4. The concept of a statistical (or an information) manifold was introduced for
the purpose of facilitating the insightful treatment of probability distributions
Importance and Unpredictable Effectiveness of Mathematics 113
The following table lists the joint research projects, coming from various industries,
which have been conducted at IMI (Fig. 9).
Importance and Unpredictable Effectiveness of Mathematics 117
We launched an electronic journal (but have published a hard copy once a year), the
Journal of Math-for-Industry in April 2009, which is published at the institutional
repositories. This journal JMI is dedicated to the broadening of the horizons of MfI,
and publicizes swiftly and internationally achievements in education and research
on MfI. JMI presents original research papers and survey papers with original
viewpoints in all scientific disciplines where mathematics and applications jointly
play a basic or seminal role. Articles by scientists in a variety of interdisciplinary
areas are published. Research areas include significant applications of mathematics
to, in and for industry, including feedback from industry to mathematics, which is a
new initiative for MfI (Fig. 10).
In Japan, it has been only about 10 years since the time that joint research
between academia and industry began on a significant scale (to this time, mainly in
engineering, agricultural and chemical fields). However, as seen in the three graphs
below, it appears that the number of such joint projects, the number of filings for
and acquisitions of patents, and other related measures of such joint research have
leveled off in recent years. According to the analysis carried out by the Ministry
of Economy, Trade and Industry, in the past, joint research between academia and
industry was not common in Japan, compared with Europe and the US. However,
because Japan had strong foundations for high-level research in both academia and
118 M. Wakayama
industry, it was not difficult to transfer the knowledge that already existed in the
academic world to industry and thereby achieve rapid technological development.
This situation has now existed for 10 years. However, today, we are grappling with
more fundamental problems. We have reached the stage where problems concerning
the real foundations of research must be investigated.
It is often pointed out, even outside of Japan, that after the Meiji Restoration,
and again after World War II, Japan made great efforts to catch up with the
world’s advanced countries, in particular with the US. Consequently, nowadays,
it is highly regarded as a country at the cutting edge of technology. Around the
1980s, Japan had gotten the unenviable reputation (as “Free ride on basic research”)
of embracing new proven national and international technology with resounding
success and turning it into an innovation with a Japanese emphasis. This could
not be, however, seen as so obvious. Actually, for instance, beginning around
the mid-1980s, Japanese companies did invest a lot in fundamental research and,
of course, basic scientific research was always greatly appreciated especially in
academia. Indeed, its economic success throughout history is directly related to
this inherent Japanese originality and creativity to be innovative. That success was
achieved through technological innovation. However, the research and development
style that has been used until this time (particularly, after mid-1990) is no longer
Importance and Unpredictable Effectiveness of Mathematics 119
sufficient. Indeed, for instance, before Japan became one of the world’s strong
leaders in technology, research and development in industry that had engineering
as its foundation, in particular manufacturing (monozukuri), was often sufficient,
but this is no longer the situation (Fig. 11).
As is happening in developed industrial countries around the world now, in
response to the demands and needs for the new technologies such as information
security, life science, etc., and even manufacturing, the new challenge for Japanese
industry is to embrace properly the linking of mathematics with industrial discovery
and innovation. Given this situation, we must also reconsider the importance of
mathematics education in Japan, while educating scientists and technologists with
strong backgrounds in mathematics and promoting a broad range of research by
these researchers. In this way, top-level mathematicians and mathematical scientists
as well as the fostering such researchers will become a vital part of the infrastructure
of Japanese society.
Acknowledgements The author thanks Bob Anderssen in CSIRO, Australia, for stimulating
discussion about mathematics in- and for industry and for his helpful comments on the article.
He also thanks my colleague Yasuhide Fukumoto in IMI for his useful comments on the article.
He finally thanks Seiko Sasaguri for preparations of figures in the text.
References
Masato Wakayama received his Ph.D. at Hiroshima University in 1985. His main
research fields are representation theory and number theory. He is a Professor of
Mathematics, Kyushu University since 1997 and now a Distinguished Professor.
He is the founding Director of the Institute of Mathematics for Industry (IMI),
which is the first research hub to bring together mathematics and industrial
technologies in Asia. In addition to the Advanced Mathematics Technology, Applied
Mathematics and Fundamental Mathematics divisions, the Laboratory of Advanced
Software in Mathematics it includes the Office for Promotion of Collaboration and
Consultation, which promotes smooth interaction and valuable joint research with
the industry.
He currently serves as a Vice-President of Kyushu University, and is also
the Program Leader, Global COE Program “Education and Research Hub for
Mathematics-for-Industry.”
Mathematics for Business and
Business Leaders Based on Mathematical
Intelligence
Hiroshi Fujita
Emeritus Professor,
University of Tokyo
Introduction
Japanese business leaders who are authors of the chapters of this book report the
role and contributions of mathematics for their companies or corporations and
relate their experiences of making serious decisions as business heads with the aid
of mathematical thinking. Their comments encourage us to develop mathematics
for business and urge on our efforts to improve mathematics education to foster
potential business leaders.
The present chapter is intended to analyze and respond to their views and
suggestions from the standpoint of mathematics. Except for a few remarks, no
specific mathematical background in readers is assumed below.
We shall begin with recognition of the basic two ways in which mathematics
contributes to business, namely, one as a mathematical tool and the other in
mathematical thinking. For the latter, which is the main target of this chapter, I
propose renaming it mathematical intelligence with two components: a component
of mathematical literacy and a component of mathematical thinking, which will be
described in the following section, Mathematics for Business and Businesspeople.
The remaining part of this chapter is devoted to relevant topics of a somewhat
specific nature, which include subjects such as business mathematics viewed
from the methodology of applied mathematics (in the Methodology of Applied
Mathematics section) and a historical view of great peaks of mathematics as origins
of useful methods (in A Historical View of Great Peaks of Mathematics). Then,
mathematics education to cultivate mathematical intelligence will be discussed
Y. Giga and T. Kobayashi (eds.), What Mathematics Can Do for You: Essays and Tips 123
from Japanese Industry Leaders, DOI 10.1007/978-4-431-54346-6 13,
© Springer Japan 2013
124 H. Fujita
in regard to high school education and liberal arts education (in Education to
Foster Mathematical Intelligence). The chapter concludes with some comments on
the Japanese cultural background concerning mathematics as compared with the
Western one.
Finally, I am afraid that, because of my specialty and career, the scope and
content of this chapter is rather biased to applied analysis, if it should speak for
mathematics in general.
The following two quotations from chapters in this book are lively examples of
ideal collaboration between mathematical tools and creative minds of leaders, which
is realized in typical venture companies.
When describing natural phenomena, we human beings have been using mathematical
expressions to simplify the principle, not only in physics and chemistry but also in biology,
thereby expressing phenomena intuitively and in an objective fashion. Here, “to ascertain
the essence of things” and “to identify the true nature of things” which underlie Kyoto
culture seem to coincide amazingly well with methods of mathematical expression.
(Horiba)
The goal of our company is to actualize artificial intelligence. Inventing A.I. technology
means to research the function of the brain while describing the finding in mathematical
language. (Iwane)
In the next section I shall discuss the second category, i.e., the way of mathemat-
ical thinking, in more detail.
with numbers, such as bulk weight, noise level, vibration level, etc. By “numbers,” I am
not talking about technical calculations that are conducted in laboratories. I am talking
about simple arithmetic, such as addition, subtraction, multiplication and division. Numbers
and arithmetic are what Toyota’s 300 thousand employees share as the fundamental
communication tool. (Toyota)
While mathematical thinking should play its role in every phase of business and in
every corner of companies, its most serious and challenging stage is cases where
business leaders need to make the best decision in a complicated situation, taking
future risk factors into account and foreseeing the big picture, within a limited time,
as many of the authors of this book who lead their businesses would agree.
Needless to say, aside from some expectation that activities based on mathemat-
ics for business may stimulate the interest of mathematicians as new targets, the goal
of mathematical thinking in business is not aimed at the progress of mathematics
itself. Nonetheless, the ability of mathematical thinking that business leaders are
required to exercise is comparable, in its style and depth, to that of professional
mathematicians in research. In fact, both mathematicians and businesspeople are
dedicated to finding the essence of problems and thinking through them until they
reach a significant solution.
The capacity enabling such efforts is so basic in nature that it can be developed
initially by eager learning of high school mathematics. This point is well stated by
Norio Wada, the chairman of NTT, as below.
By learning school mathematics using numbers and figures, we have acquired these thinking
processes, i.e., the ability to look at objects from a variety of perspectives, and ultimately
discover a truth that suits the intended purpose and express it in a simple way. Once we
Mathematics for Business and Business Leaders Based on Mathematical Intelligence 127
graduate from school, we are quite often confronted with problems that we have not studied
in school, and become highly upset due to the fact that we don’t have a clue as to what to
do. My experiences show that the mathematical thinking mentioned above is very helpful
in solving problems in such situations. In other words, mathematics is a tool that enables
us to broaden our ability to deal with problems that confront us, and makes our conclusions
highly convincing. . . . I’d like to emphasize the importance of mathematics education in
laying a foundation for teaching people to look at and think about things from perspectives
not confined to any existing framework. (NTT)
The business leaders in this book confess that the difficulty of finding a solution
in business under given conditions is reminiscent of “equation solving” that they
experienced in their high schools. Let me quote Yasuchika Hasegawa of Takeda
Pharmaceutical Company as an example.
In managing day-to-day operations, there are often times when I get the feeling that I
am subconsciously engaged in solving a system of equations. These can be multivariate
equations that have a significantly higher number of variable parameters than those given
in the above equation, equations with many thousands of constants and variables. There are
times when it is literally impossible to pin all the variables down. (Takeda)
I will also quote from Koichi Hori’s (Dream Incubator Inc.) essay “Mathematics
Is Indispensable to Business Managers,” which appeared in the Japanese edition of
this book. He makes a proposal on the basis of his experience of having studied at
The University of Tokyo and Harvard Business School and succeeded as a business
consultant.
In our time, when we can so easily benefit from the progress of computer technology as
well as mathematical tools such as accounting and business science, we would give up
any privileges of the era if we steered the business without recognizing the importance of
mathematical thinking. . . . I dare say that universities only have to provide science courses
(in the broadest sense of the term, which includes engineering) to undergraduates. Study
of specialties such as law and economics should start in graduate school (master courses).
Mathematics for Business and Business Leaders Based on Mathematical Intelligence 129
Undergraduate curricula should be organized with their cores in science and with emphasis
on mathematics and logic, and further on learning the scientific methodology of natural
science to set up a hypothesis and test it. (Dream Incubator)
Most of the business leaders who appear in this book spent their high school
and university years in the 1960s and 1970s; it was the era when mathematical
education and the social environment in Japan were in good shape. Students at
that time used to read more books and study harder than today. With remnants of
the pre-World War II educational framework still in evidence in schools, teachers,
and parents, high expectation for learning was shared among students. Besides, the
open atmosphere after the educational reform following the war had made it easier
for excellent students to hold big dreams to realize diverse possibilities. Since the
ability of critical thinking was highly weighted in entrance exams of top universities,
which already were seriously competitive then, high school students were required
to have an academic attitude even in their exam studies. It is unfortunate that the
current condition of Japan’s mathematical education and social environment are
deteriorating noticeably.
Mathematical literacy is, just as the term indicates, the ability to utilize basic
mathematics linguistically. The word is now widely known as one of the survey
items in the Program for International Student Assessment (PISA). I was the first
advocate of mathematical literacy who introduced the word in the mid-1980s [2, 3]
and maintained that it should be a part of the attainment target of secondary school
mathematics.
I have stressed that the principal goal of mathematical education in high schools
is to help students develop their mathematical intelligence, which is, in parallel to
discussions made above (in Mathematical Intelligence Composed of Mathematical
Thinking and Mathematical Literacy), composed of mathematical thinking ability
and mathematical literacy. In this section, I will verify the importance of mathemat-
ical literacy for businesspeople.
As stated earlier, by mathematical literacy for businesspeople, I mean the ability
to leverage mathematical concepts and expressions to nail down a problem in a
quantitative way, to ensure an objective explanation, and to improve the persua-
siveness of business judgments. The merit of mathematical exposition, including
numerical expressions, is certainly its clarity and objectivity. It helps us differ-
entiate true and false unambiguously. Being free from subjectivity, cheating, and
fabrication, mathematically proved statements are strongly persuasive to anyone.
They are not only useful in inter-organizational and interdisciplinary cooperation,
but also provide a service in international communications where people do not
necessarily share the same cultural background and the same language. This is one
130 H. Fujita
of the reasons why the business leaders in this book highly regard mathematical
literacy, as exemplified by the following quotation.
No matter how talented or experienced, no company president should make such decisions
in isolation. The organization of a business thus allows a company to allocate roles and
responsibilities and to delegate authority. Through this system and the integration of
specialist knowledge and skills, a company is best able to handle the issues it faces. . . .
It would be wrong to suggest that executives and managers simply embrace the qualitative
intuition derived from experience in reaching key decisions. It is here that the enormous
contributions are made by quantitative factors. Here, mathematical modeling, which is the
subject of this discussion, comes into play. (Takeda)
I wonder if the reason that Japanese business leaders emphasize the importance
of mathematical logic so strongly might originate from the fact that the Japanese lan-
guage is more implicit and feeling-based than explicit and logical when compared
with European languages. Even a Japanese mathematician can sometimes explain
his research more clearly in English than in Japanese. Kaoru Yosano, the only
politician who contributes to this book, writes about how English and mathematics
can be harmonized so easily.
When I was an eighth grader, my father, who was a diplomat, was dispatched to Egypt, and
I was put into a British boarding school in suburban Cairo. From the ninth grade, I started
taking classes in English. Since the school rules were strict and I could not understand
English well, I had a hard time of it. In those days, however, I made the wonderful discovery
that math was easy when taught in English. To solve a word problem in math, for example,
you have to ask yourself questions such as “what are the given conditions?” and “what
is the unknown value?” I could understand these processes more clearly than when I was
Mathematics for Business and Business Leaders Based on Mathematical Intelligence 131
studying math in Japanese, because both math and English were languages that emphasized
the importance of clarity. After a short time, I became the best math student in my grade.
(Yosano)
The “concept” that enabled such modeling is, on the mathematical side, differential
calculus (invented by Newton himself) and acceleration, etc.; and on the physics
side, particles and universal gravitation (the law discovered by Newton himself).
The “method” used by Newton to solve the problem was differential and integral
calculus by hand.
Next, as an example from advanced technology, I would like to mention the
chapter “Improvement of Railway System Technology” by Mr. Eisuke Masada,
chairman of the Railway Technical Research Institute, in this book. In this case,
“object” is the task to realize a high-performance train including a linear motor
train. To build up mathematical models that help them realize the expected high
performance, they use concepts from physics and engineering. The conditions
posed in order to formulate the task are equations of motion in complex dynamical
systems and partial differential equations such as electromagnetic equations. As for
“method” solving them, they use the theory of partial differential equations and
computer simulations.
Although the situation has been exemplified through the preceding examples, here
I would like to analyze the methodology of mathematical sciences, or modern
applied mathematics from a general point of view [4, 5], and in accordance with
the main subject of this book. In doing so, I hope that the here-stated arguments on
methodology of applied mathematics would be suggestive also in examining the use
of mathematics for business and in confirming the way of mathematical thinking by
business leaders.
Actually, when we encounter phenomena to study in the light of mathematics
or problems to solve with the aid of mathematics concerning certain objects, we
start with mathematical modeling of the latter. Namely, modeling is the first stage
of research (and development).
There, proper concepts must be called for, and we formulate the problem into a
mathematical model. In carrying out this process, it can happen that some concepts
are taken from outside of mathematics and that the model is formed with the aid
of laws or formulas in related fields of science. In these cases, a good model
is an outcome of the researcher’s multidisciplinary scholarship or his/her close
cooperation with fellow experts in relevant fields.
Next to modeling, which formulates the object mathematically, the second stage
comes up. Namely, the mathematical model thus obtained is analyzed or solved
by means of mathematical methods that are chosen by the researcher, often in
consideration of computing cost and time. The necessary mathematical work in
order to calculate or to solve something is carried out, very often in the form of
computer simulation. Thus this stage of mathematical analysis is sometimes called
the simulation stage.
134 H. Fujita
more satisfactory solutions of the problem. Indeed, J.L. Lions, who was a leading
scientist of applied mathematics, referred to the third stage of the above-mentioned
methodology as feedback and control. The positive meaning of feedback seems to
have great significance for problems of time evolution systems, like the prediction
problems about investment and management in business mathematics, where one
needs to calculate the solution of a problem at every instant along the flow of time.
Many of the mathematical concepts, methods, and ways of thinking that support
the application of mathematics are the outcomes of outstanding great works of
mathematics, so to speak, the great peaks in history of mathematics from ancient
138 H. Fujita
times. Reviewing these great peaks would be useful and inspiring for us to found
reliable and progressive mathematics for business.
With an intention to project the result on our efforts to develop business
mathematics in the 21st century, and to encourage mathematics education to
cultivate mathematical intelligence in students who are potential business leaders,
we shall here examine such alpine scenery of great achievements of mathematics
and recall characteristics of each of these peaks briefly [4].
It is well known that ancient mathematicians in Arabic counties, China, India,
and other parts of the world made great contributions. We could, however, count
the following three as giant peaks in the history of mathematics, as far as direct
influences on modern mathematics and its application are concerned.
1st peak: Birth of Euclidean Geometry in about 300 BC.
2nd peak: Discovery and Development of Calculus (Differential and Integral
Calculus) in the 17th and 18th centuries.
3rd peak: Development of Modern Axiomatic Mathematics in the twentieth
century.
Incidentally, recognition of the importance of these three peaks in regard to
mathematical education was first proposed by the late Professor Yukiyoshi Kawada
around 1970, when the wave of the New Math movement reached Japan. The
suggestion of Prof. Kawada was seriously considered in organizing the Japanese
national curriculum of high school mathematics for those periods, when most of the
business leaders who are authors of this book learned mathematics as high school
students.
(b) Calculus
The word “calculus” as used here means infinitesimal calculus and is used to refer
jointly to differential and integral calculus. In addition, mathematical fields built
on differential and integral calculus are called analysis. The theory of differential
equations and the theory of complex functions are primary branches of analysis. We
shall use here the terms “analysis” and “analytic” in this sense.
The objective of analysis is to study change (variation): namely, change of
numerical variables. “Movement of a point,” which Newton studied at the beginning,
is nothing but the change in the position of a point in the passage of time (an
independent variable). As we have learned in high school mathematics, a change
of values of an independent variable gives rise to a corresponding change of the
dependent variable. According to modern terminology, this correspondence between
the two variables is itself called a function. As for the function describing the
movement of a point, coordinates of a point are dependent variables. Analysis is
a field of mathematics about functions.
Changes include local changes and global changes. When the variable is time,
i.e., when the object under consideration is a system evolving in time, what matters
locally is the changes in the vicinity of a given instant. A typical example is the rate
of change in that instant. If the function represents coordinates of a moving point,
140 H. Fujita
its instantaneous rate of change is none other than velocity. In contrast, in global
changes, we ask the amount of variation of a function when variables vary over a
specified finite or infinite region.
In problems encountered in mathematics for business, we ask the variation of
a quantity in time, make predictions about the future, and try to cope with future
risk factors. Such kinds of problems are those of a time evolution system, and
the treatment will be essentially analytic. In the problem of coding of pictures, as
introduced in this book by Mr. Wada, the president of NTT, they handle variations
in the shades of colors of a picture. It is quite natural that they are using analytic
methods to handle such spatially extended variations. One of the interesting methods
that is realized in this imaging technology is to consider a small increment Δ F of a
function over a small duration of time Δt in tracking the change of an image along a
consecutive frame, a method that is reminiscent of the idea of differential calculus.
In addition, as we know from high school mathematics, one of the central
problems around functions is to find the maximum/minimum of a function. The
extension of the latter, namely, the problems of maximization and optimization, are
often encountered in mathematics for business. They are essentially analytic as well.
Coming back to reflections on the historical scale, and as a main point of this section,
we claim that a new peak has shown up recently. It is the computer-driven advance
of mathematical sciences. In its wide range and its potential height, the new peak
Mathematics for Business and Business Leaders Based on Mathematical Intelligence 141
deserves the title of the fourth giant peak in the history of mathematics, although it
might be better to use the term “a great alpine plateau” instead of “a giant peak.”
Thus we regard the following four as the towering high-rises of the mathematics
concerned:
Euclidean Geometry, Calculus, Axiomatic Mathematics, Mathematical Sciences.
In this direction, we believe, mathematics will find its promising way with its
own vivid creativity and with recovered links between mathematics and the real
world. Formation of advanced and trustworthy mathematics for business should be
strived for in this spirit.
At this point, however, we note that as to mathematics, the rise of a big new
peak does not mean that the former peaks are wiped out, which the very history
of mathematics shows. As for mathematics, such a big change takes place, so to
say, as sublation (Aufheben) rather than a drastic paradigm shift observed in some
fields. The heritages of modern mathematics remain important and will be respected
even in the era of mathematical sciences, although stagnation caused by excess
ramification inside and by unconcerned isolation outside must be cured.
In this connection let us quote a passage by Poincaré [9].
If I may be allowed to continue my comparison with the fine arts, the pure mathematician
who should forget the existence of the exterior world would be like a painter who knew
how to harmoniously combine colors and forms, but who lacked models. His creative power
would soon be exhausted.
The second point we want to emphasize here is that this new plateau, the fourth
peak, shares basic features with the second one, the peak of calculus, in the sense
that both are characterized by (1) realistic concepts, (2) powerful methods for rich
and far-reaching applications, and (3) tight linkages with outer fields and objects. In
short, the fourth peak of unified mathematical sciences is outgoing in its character
just as the second peak of calculus was.
There is no doubt that mathematical education in high schools and universities plays
an important role in training students, who would become experts in science or
engineering in industry, in order to enable them to acquire mathematical capability
at the required level. On the other hand, I have already discussed briefly in the
second section above, Mathematics for Business and Businesspeople, the role of
school education to foster mathematical thinking ability and mathematical literacy
of relevance to general business leaders. The present section is a supplement to my
earlier argument there.
142 H. Fujita
Prof. Kodaira also argued as follows. The talent for mathematics, namely, the
talent to think mathematics and understand it, is not quite the same as brightness in
the generic sense, and it varies greatly from one person to another. It is as though
we need to introduce a new kind of human sense to account for such a situation.
We shall call this math-sense. Like the sense of direction, there is no particular
organ of the body that produces math-sense. The inborn level of math-sense varies
considerably and hence not every person is endowed with it or with the same level
of math-sense. This may be because math-sense grew up at quite a later stage of
human evolution. (Prof. Kodaira further told me informally that, in ancient times,
those who were too weak-sighted to distinguish their mother from a wolf must have
died out due to natural selection. On the contrary, those with weak math-sense
were not wiped out because math-sense did not make much difference in ancient
environments.) As the individual math-sense is considerably different among us,
we cannot justify the so-called uniform principle to be applied to mathematics
education.
Moreover, even if one is endowed potentially with good math-sense, it will not
be awakened unless he/she continues proper studying of mathematics earnestly from
childhood. In this connection, we must enrich the language education in elementary
schools as a prerequisite for a good start toward math-sense. This was the claim of
Prof. Kodaira, which I support fully.
I feel that the concept of math-sense is useful as a model that helps us study the
manifestation and development of the ability to think mathematically.
I propose that high school education should aim at awakening the math-sense of
students and letting them acquire the habit of thinking mathematics deeply, without
discriminating between science-oriented students and the others. Moreover, people
in charge of evaluation of students’ scholastic achievements should try to invent an
Mathematics for Business and Business Leaders Based on Mathematical Intelligence 143
appropriate way to appreciate (i.e., diagnose the growth of math-sense of) students
who have done such proper study of mathematics and have reinforced the ability to
think mathematically. This kind of effort would be valuable from the point of view
to train future business leaders endowed with mathematical intelligence.
In Japan before World War II, the entrance into leading national universities
(imperial universities) was practically limited to graduates of high schools under
the prewar educational system. According to this educational system, the total quota
of the imperial universities equaled the total quota of high schools. In other words,
the number of high schools was so limited under the prewar education system. In
those days, liberal arts education used to be given at high schools as one of the two
backbones together with the character building education, and it was the dominant
part of elite education then.
Under the postwar education system in Japan, although liberal arts education is
supposed to be given in the first year of university normally, it currently shows a
tapering off trend due to pressure from hectic professional education. Nevertheless,
as the bulk of professional education is being shifted to graduate schools these
days, some people even go as far as to argue that liberal arts education should be
expanded to cover the entire period of undergraduate studies, while modern liberal
arts education must keep in touch with ideas and methodology at the fronts of
academic fields. (The essay of President Kohichi Hori already cited is one of those
voices.)
In liberal arts education today, mathematics plays an important role. It was
the case in the old days as well; originally, of the seven subjects constituting the
classic liberal arts, more than half of them (arithmetic, geometry, logic, grammar,
rhetoric) were related to mathematics, with mathematical literacy to be considered
as a linguistic ability. The other subjects were music and astronomy, while I like to
consider that the latter can be included in mathematical sciences.
Even putting aside such old history, mathematics is still expected to play a
major role in future liberal arts education. That said, mathematical education
expected today for the university level must not limit itself to a blind transmission
of practical methods and techniques, nor should it be too much concerned with
detailed conventions of professional mathematicians. Rather, it should be aimed
at acquisition of mathematical literacy as stated above in the section Mathematics
for Business and Businesspeople, and should help students understand the method-
ology of mathematical sciences as stated in the section Methodology of Applied
Mathematics, thereby allowing them to attain a unified view on the fundamentals
of mathematics and its applications. This knowledge of mathematics can be said
to be a passport to the frontier of our present civilization. Before World War II,
high school students used to eagerly study Western languages such as English and
German as a “passport” that enabled them to access the frontier of knowledge
144 H. Fujita
(although they were not so concerned with the ability to communicate orally). I think
that, nowadays, acquiring mathematical intelligence plays the role of a “passport”
for all students, whether they major in science or not, that paves the way toward
international leadership at the forefront of various fields including business.
Concluding Remarks
On the other hand, in the modern information society, where human relation-
ships have diversified, it is more necessary than before for leaders of teams and
organizations to be warmhearted with human sensitivity to others. Moreover, the
world of integrated sciences has entered such a phase that “Teamwork can produce
creativity,” as was described by Koichi Tanaka, who received the Nobel Prize in
Chemistry for his research on mass spectroscopy.
I would like to conclude this chapter with the hope that mathematics for
business in Japan will be built and flourish on the arching combination of Western
intelligence and Japanese sentiment, and will contribute much to the world as well
as to our future.
References
1. Fujita H (1966) On the blowing-up of solutions of the Cauchy problem for ut = Δ u + u1+α .
J Fac Sci Univ Tokyo 13:109–124
2. Fujita H (1985) The present state and a proposed reform of mathematical education at senior
secondary level in Japan. J Sci Educ 9(2):39–52
3. Fujita H (1987) The present state and current problems of mathematics education at the senior
secondary level in Japan. In: Development in school mathematics education around the world,
vol. 1. National Council of Teachers of Mathematics and The University of Chicago School
Mathematics Project, Chicago, pp 191–224
4. Fujita H (2004) Goals of mathematical education and methodology of applied mathematics. In:
Proceedings of the 9th international congress on mathematical education, Tokyo/Makuhari, 31
July–6 August 2000. Kluwer Academic, Dordrecht, pp 19–36
5. Fujita H (2007) A suggestion to statistics education on its spirit and approach—from viewpoint
of mathematics education. Nihon Tōkei Gakkai shi 36(2):219–230 (in Japanese)
6. Giga Y, Kohn RV (1985) Asymptotically self-similar blowing-up of semilinear heat equations.
Comm Pure Appl Math 38:297–319
7. Kodaira K (1984) Gensoku wo wasureta syotou/chutou kyouiku—nan no tame, soshite dare no
tameni isogunoka. Iwanami Kagaku 54(1):4–9 (in Japanese)
8. Kagaku to gijyutsu no aida, International Institute for Advanced Studies, Kyoto (2000)
(Koutouken sensyo no.7) (in Japanese)
9. Poincaré H (1905) La Valeur de la Science. Flammarion, Paris [English translation by Halsted
GB (1929) The value of science in The Foundations of Science. Science Press]