Вы находитесь на странице: 1из 24

Finite Differences

P. Sam Johnson

January 9, 2015

P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

1 / 24

Overview
Assume that we have a table of values
(xi , yi ),

i = 0, 1, 2, . . . , n

of any function y = f (x), the values of x being equally spaced.


That is, xi = x0 + ih, i = 0, 1, 2, . . . , n. The step length h is called the
interval of differencing.
Suppose that we are required to recover the values of f (x) for some
intermediate values of x, or to obtain the derivative of f (x) for some x in
the range x0 x xn .
The methods for the solution to these problems are based on the
concept of the differences of a function which we define and disucss
in the lecture.
P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

2 / 24

We are given a table of values


(xi , yi ),

i = 0, 1, 2, . . . , n

of any function (usually unknown) y = f (x).


The values of x being equally spaced. That is,
xi = x0 + ih,

i = 0, 1, 2, . . . , n.

If y0 , y1 , y2 , . . . , yn denote a set of values of y , then


y1 y0
y2 y1
..
.
yn yn1
are called the differences of y .
P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

3 / 24

Forward Difference Operator,


Denoting the differences, (y1 y0 ), (y2 y1 ), . . . , (yn yn1 ) by
y0 , y1 , . . . , yn1
respectively, we have
y0 = y1 y0 , y1 = y2 y1 , . . . , yn1 = yn yn1
where is called forward difference operator and y0 , y1 . . ., are
called first forward differences.
The differences of the first forward differences are called second forward
differences and are denoted by 2 y0 , 2 y1 . . .. Similarly, one can define
third forward differences, fourth forward differences, etc.
The forward difference operator can be operated on function-values,
generally denoted as
f (x) = f (x + h) f (x),
P. Sam Johnson (NITK)

x = x0 , x1 , . . . , xn1 .

Finite Differences

January 9, 2015

4 / 24

Exercises
1 Find the successive differences of f (x) = ab c x and sum the first n
differences.
2 Find the function (with suitable h) whose first difference is
ax + b
sin x
ex
e a+bx

P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

5 / 24

The following result can be proved by the method of induction and it helps
in expressing, any powers of operated on y (x), in terms of yi s.
Theorem
Suppose that we have a table of values (xi , yi ), i = 0, 1, 2, . . . , n of any
function y = f (x), the values of x being equally spaced with h as the
interval of differencing. Then for any non-negative integer k,
 
k
X
ki k
yr =
(1)
yr +i .
i
k

i=0

The theorem refers to the following facts:


1

each higher difference can be expressed in terms of the preceding


differences

it is possible to obtain higher differences in terms of the


function-values yr by continuous substitution.
P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

6 / 24

Shift Operator, E
An operator E defined as
Ef (x) = f (x + h)
is called a shift operator which results in increasing the argument by the
interval of differencing.
E n stands for the operation E being carried n times. Then
E n f (x) = f (x + nh).
The inverse operator E 1 is defined as
E 1 f (x) = f (x h).

P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

7 / 24

Relation between E and

f (x) = f (x + h) f (x) = Ef (x) f (x).


Hence
Ef (x) = f (x) + f (x) = (1 + )f (x).
Here 1 (unity) is an operator just as E and and leaves the function
unaltered when it operates on that function. That is, 1f (x) = f (x).
Since f (x) is arbitrary, we get
E 1 + .

P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

8 / 24

With the help of the operators E and , we can prove the above result
which is restated below.
For any non-negative integer k,
 
k
X
ki k
yr =
(1)
yr +i .
i
k

i=0

Since E 1, by binomial expansion, the above relation can be shown.


Exercises
1

Evaluate E 0 f (x), (c) and E (c) where c is a constant.

True or false: If f (x) = 0, then either 0 or f (x) = 0.

True or false: If E 2 f (x) and E 2 [f (x)]2 are identical.

P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

9 / 24

(Forward) Difference Table

x
x0

y = f (x)
y0

x0 + h

y1

f (x)

2 f (x)

3 f (x)

4 f (x)

y0
2 y0
3 y0

y1
x1 + 2h

2 y1

y2
y2

x1 + 3h

y3

x1 + 4h

y4

4 y0
3 y1

2 y2
y3

P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

10 / 24

Exercise
An operator T is said to be linear if
T [af (x) + bg (x)] = aT [f (x)] + bT [g (x)],
where a, b are constants. Prove that the operators E and are linear.
The following theorem says that any value of the function f (x) can be
expressed in terms of leading term and the leading differences of an
ordinary difference table. It can be proved by the method of induction.
Theorem
For all integral values of n, f (a + nh) =

Pn

n
r =0 r

r f (a).

Exercise
Prove the above result with the help of the operators E and .

P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

11 / 24

Let f (x) be a polynomial of degree n in x. If h is the interval of the


argument, then f (x) is a polynomial of degree (n 1). Thus it is to be
noted that the result of differencing a polynomial once, reduces its degree
by one.
Theorem
Let f (x) be a polynomial of degree n in x. Then the nth difference of this
polynomial is constant. Also the (n + 1)th and higher differences are zero.
Exercise
If f (x) = a0 x n + a1 x n1 + + an1 x + an a0 6= 0, then find n f (x).
Obtain
25 {(x a)(x b) (x z)}
where the operand has only 25 factors and there is no factor of the type
(x x).

P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

12 / 24

Exercises
1 Prove that

e =

2
E

ex .

Ee x
,
2 e x

the interval of differencing being unity.


2

Prove that

n1
X

yx = yn y0 .

x=0
3

Prove that y4 = y3 + y2 + 2 y1 + 3 y1 .

Show that
y4 = y0 + 4y0 + 62 y1 + 103 y1
if fourth and higher differences are zero.

P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

13 / 24

Exercises
1 Prove that






n+1
n+1
n+1
y0 + y1 + + yn =
y0 +
y0 + +
n y0 .
1
2
n+1
2

Prove the following identity

X
x=0

y2x

1X
1
=
yx +
2
4
x=0

4
1 +

2
4


y0 .

Find the sum of the series,


1.2 x n 2.3 2 x n + 3.4 3 x n 4.5 4 x n +
the interval of differencing being unity.

P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

14 / 24

Backward Difference Operator, (called nabla)


Central Difference Operator,
The differences
y1 y0 , y2 y1 , . . . , yn yn1
are called first backward differences and they are denoted by
y1 , y2 , , yn
respectively, so that y1 = y1 y0 , y2 = y2 y1 , . . . yn = yn yn1 ,
where is called the backward difference operator.
The central difference operator is defined by the relations
y1 y0 = y1/2 , y2 y1 = y3/2 , . . . , yn yn1 = yn1/2 .
Similarly, higher-order central difference can be defined.
P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

15 / 24

Averaging Operator,
We can observe the following
y0 = y1 = y1/2 ,

3 y2 = 3 y5 = 3 y7/2 .

The averaging (mean) operator is defined by the equation:



1
yr =
yr +1/2 + yr 1/2 .
2
Exercise
Deduce the following:
1 E 1
E 1/2 E 1/2
(1/2)(E 1/2 + E 1/2 )
2 1 + (1/4) 2
E E
P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

16 / 24

Differential Operator, D

The differential operator D is defined by


Df (x) =

d
(f (x)).
dx

Exercise
Deduce the following:
1

E e hD

hD

2
2

P. Sam Johnson (NITK)

3
3

Finite Differences

January 9, 2015

17 / 24

Factorial Notation

The continued product of factors whose first factor is x and successive


factors decrease by a constant difference h, is called a factorial function
That is denoted by x [n] where n is a positive integer, which determines the
number of factors in x [n] .
Thus
x [n] = x(x h)(x 2h) (x [n 1]h).
When h = 1,
x

P. Sam Johnson (NITK)

[n]

 
x!
x
=
= n!
(x > n).
(x n)!
n

Finite Differences

January 9, 2015

18 / 24

In fact f (x) = x [n] where n is a positive integer, is called the factorial


polynomial of degree n and the various differences of a factorial
polynomial are again factorial polynomials.
x [n] = nhx [n1] is a polynomial of degree (n 1) in factorial notation.
In general, n x [n] = n!hn , a constant term and k x [n] = 0 for k > n.
When h = 1,
x [n] = nx [n1]
2 x [n] = n(n 1)x [n2]
..
..
.
.
n x [n] = n!.
Hence differencing of functions in factorial notations resembles the
differentiating of functions when h = 1.
P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

19 / 24

We now discuss the factorial function x [n] when n is any negative integer.
A simple calculation shows that for any positive integer n
x [n] = (x [n 1]h)x [n1] .

(1)

Putting n = 1 in the equation (1) we get x [0] = (x + h)x [1] .


For convention, we take x [0] = 1. Hence x [1] =

1
x+h .

In general,
x [n] =

1
(x + nh)[n]

x [n] = nhx [(n+1)]


2 x [n] = ()2 n(n + 1)h2 x [(n+2)]
and so on.
P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

20 / 24

Exercises
1 Express the function
x 4 12x 3 + 24x 2 30x + 9
in factorial notation, the interval of differencing being unity.
2

Using factorial notation, obtain the function whose first difference is


x 3 + 4x 2 + 9x + 12.

Express 2x 3 3x 2 + 3x 10 and its successive difference in factorial


notation.
A second degree polynomial passes through (0, 1), (1, 3), (2, 7) and
(3, 13). Find the polynomial.

P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

21 / 24

Exercises
1 Find the relation between , and in order that
+ x + x 2
may be expressible in one term in the factorial notation.

P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

22 / 24

Exercises
1 Using the method of separation of symbols, prove the following
identities:
1
2
3

yx n yxn = yx1 + yx2 + 2 yx3 + + n1 yxn .


x2
x3
x
2
y1 + (1x)
y1 x + y2 x 2 + y2 x 3 + = x1
2 y1 + (1x)3 y1 + .
Taking fifth order differences of yx to be constant and given
y0 , y1 , y2 , y3 , y4 , y5 prove that
y2 12 =

1
25(c b) + 3(a c)
c+
2
256

where a = y0 + y5 , b = y1 + y4 , c = y2 + y5 .
2

For any positive integer n, prove the following:


1
2
3



x
r nn = nr
, r <n

n xn = 1
x [r ] (x rh)[n] = x [r +n] .

P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

23 / 24

References

Richard L. Burden and J. Douglas Faires, Numerical Analysis Theory and Applications, Cengage Learning, Singapore.

Kendall E. Atkinson, An Introduction to Numerical Analysis, Wiley


India.

David Kincaid and Ward Cheney, Numerical Analysis Mathematics of Scientific Computing, American Mathematical
Society, Providence, Rhode Island.

S.S. Sastry, Introductory Methods of Numerical Analysis, Fourth


Edition, Prentice-Hall, India.

P. Sam Johnson (NITK)

Finite Differences

January 9, 2015

24 / 24

Вам также может понравиться