Вы находитесь на странице: 1из 40

Euler Format theorem: Assume (a,m)=1. Then we have aΦ (m) ≡ 1(modm).

The Exponent of a number modm

PRIMITIVE ROOTS:

If a and m are the integers with m≥1 and (a,m)=1 then by Euler-Fermat theorem, aΦ (m) Ξ
1(modm). Whenever (a,m)=1 ∃ an integer f≥ 1 ∋, af ≡ a(modm). The least positive ‘f’ ∋ af ≡
a(modm) is called the exponent of a modulo m.

Definition:

1. Suppose (a,m)=1, the smallest positive integer f ∋, af ≡ 1(modm) is called the


exponent of a modulo m, and is denoted by writing f = expm(a).
2. Id a and m are relatively prime integers ∋ , exp m(a) =Φ(m), thyen a is called as
primitive root mod m.

RELATIVELY PRIME: if the greatest common divisor of a and b denoted by (a,b)=1 then a
and b are said to be relatively prime.

Definition: Given integers a,b,m with m>0, we say that a is congruent to b modulo m, we
write a≡b(modm), if m divides the difference a-b. The number m is called the modulus of the
congruence.

In other words, congruence a≡b(mod m) is equivalent to the divisibility relation i.e 𝑚⁄𝑎 − 𝑏
.

Divisibility: we say d divides n and we wirte 𝑑⁄𝑛 whenever n=cd For some c. we also say
that n is a multiple of d, that d is a divisor of n or the d is a factor of n. if d does not divide n
we write d҂n.Type equation here.

Euler Totient Property: Φ(Pα)=Pα-Pα-1 for prime P and α ≥ 1.

Euler totient function Φ(n): if n≥1 the euler totient Φ(n) is defined to be the number of

positive integers not exceeding m which are relatively prime to n, thus Φ(n)=∑𝑛𝑘=1 1 where
the ‘ indicates the sum is extended over those k relatively prime to n.

Examples for exponent of a modulo m:

1) A=5, m=8
Then a2≡ 1 mod8
Since 52=25≡1 mod 8
Where 2 is least +ve integer
=> Exp8(5)=2
2) a=7, m=13
Then ∃ least +ve integer 12 ∋ 712≡1(mod13)
=> Exp13(7)=12

Examples for primitive root modm:

1) a=7, m=13
(7,13)=1 => 𝑟𝑒𝑙𝑎𝑡𝑖𝑣𝑒𝑙𝑦 𝑝𝑟𝑖𝑚𝑒
Expm(a)= Φ(m)
Also Exp13(7)=12= Φ(13) => 7 is primitive root mod13
Note: Clearly we have Expm(a)≤ Φ(m) for any relatively prime integers a and m.

2) a=3, m=8
We have 32≡1(mod 8)
∴ exp8(3) =2 <Φ(8) =4
Here 3 is not a primitive root.

Division algorithm:

Given integers a and b with b>0, there exists a unique pair of integers q and r ∋, a=bq+r, with
0≤ r< b, moreover, r=0 if and only if b⁄a

Theorem(1):

Given m≥1(i.e m>0), (a,m)=1, let f= Expm(a) then we have

(a) ak≡ah(modm) <=> k≡h(mod f)


(b) ak≡1(modm) <=> k≡0(mod f). if particular f⁄Φ(m)
(c) the numbers 1,a,a2,---,af-1 are incongruent modm.

Given m≥1, (a,m)=1 that f is least +_ve integer ∋, af≡1(modm)

Suppose ak≡ah(mod m)

ak-1≡1(modm)

let qand r be ∋, k-h=qf+r with 0≤r<f

∴ 1≡ak-h=aqf+r=(af)q.ar=ar.1q(mod m)

= > 1≡ar(mod m) 0≤r<f

If r≠0 this contradicts the minimality of f [ ∵ r < f, ar≡1(mod m) but f I least +ve integer ∋
af≡1(mod m)]
Here r=0 giving k-h=qf, q∈ Z

= > k≡ h(mod f)

Conversely, if k≡h(mod f) then

k-h=qf, q∈∼Z

now, ak-h≡aqf= (af)q≡1q(mod m)= 1(mod m)

= > ak-h≡ 1(mod m)

= > ak≡ ah(mod m)

(ii) ak≡ 1(mod m)  ak≡a0(mod m)

 k≡0(mod f) …………..by (i)

Also since by Euler format theorem,

We know that aΦ(m)≡1(mod m)

But af≡ 1(mod m) where f is least =ve integer.

= > f⁄Φ(m)

(iii) since { 0,1,2,……., f-1} is a set of incongruent integers mod f it follows from (i) that
a0=1, a,a2,…….,af-1 are incongruent mod’m’ .

Incongruent mod m:

a≡ 0(mod m) if and only if m⁄a . Hence a≡b(mod m)  a-b≡0(mod m). If m⁄(a − b) we


write a≢b(mod m) say that a and b are incongruent mod m.
PRIMITIVE ROOTS AND REDUCED RESIDUE SYSTEMS:

Reduced Residue Systems: By a reduced residue system modulo m we mean any set of Φ(m)
integer, incongruent modulo m, each of which is relatively prime to m.

Theorem (2): let (a,m)=1 then ‘a’ is a primitive root mod’m’, ,=. The numbers a, a2,….aΦ(m)
form a reduce residue syste mod m.

Let ‘a’ be a positive root mod ‘m’

 expm(a)=Φ(m)

from theorem (1)(iii), a, a2,….aΦ(m) are incongruent mod m which are Φ(m) in numbers.

They form a reduced residue system mod m.

Conversely, let a, a2,….aΦ(m) is a reduced residue system => aΦ(m)≡1(mod m) no smaller


power of ‘a’ is congruent to 1 => expm(a)=Φ(m)

∴ ‘a’ is a positive root

Note:

1) we know that the reduced residue classes mod m form a group. If m has primitive root a,
then by above them this group is a cyclic group generated by the residue class â.
2) If m has a primitive root then by above theorem, each reduced residue system mod m can
be expressed as a geometric progression. This gives a powerful tool that can be used in
problems involving reduced residue systems.
3) We now show that primitive roots exist only if m=1,2,3,4,pα or 2pα, where p is an od
prime and α≥1.
i) For m=1 is trivial

For m=2 the number 1 is a primitive root.

For m=2 we have Φ(4)=2 and 32 ≡ (mod 4), so 3 is a primitive root.

ii) We now show that if m=2α with α≥2 then no primitive root mod m exist.

The non-existence of primitive roots mod 2α for α≥3

Theorem (3):

Φ(2ᶛ)⁄ α
Let x be an odd integer. If α≥3 we have x 2 ≡ 1(mod 2 ) so there are no primitive roots
mod 2α.

Φ(2ᶛ)⁄ α
x 2 ≡ 1(mod 2 )………………………………….(1)
Φ(2³)⁄
If α=3 congruence (1) => x 2 ≡ 1(mod 2 )
3
8 − (4)⁄
 x 2 ≡ 1(mod 2 )
3

 x2 ≡ 1(mod 8) for x odd

i.e we show for x=1,3,5,7

(2k+1)2=4K2+4k+1 =4k(k+1)+ 1 and k(k+1) is even.

We now prove the theorem by induction on α

We assume (1) holds for α and prove for equation (1) holds for α+1. By induction hypothesis
Φ(2ᶛ)⁄ α
x 2 =1+2 t where t is an integer.
squaring on both sides

Φ(2ᶛ)⁄ α+1 2α 2
x 2 = 1+2 t +2 t
Φ(2ᶛ)⁄ α+1 α-1 2
x 2 = 1+2 (t+2 t )
Φ(2ᶛ)⁄ α+1
x 2 = 1(mod 2 )
but,

Φ(2α+1) = 2 Φ(22)

= > Φ(2α) = 2 Φ(2α+1) /2

Φ(2ᶛ⁺¹)⁄ α+1
=>x 2 ≡ 1(mod 2 )
Equation (1) holds for α+1

Hence by induction theorem is proved.

The existence of primitive roots mod p for odd primes p.

Lemma (1): Given (a,m)=1, let f=expm(a)

i.e af≡1(mod m) where f is a least positive integer.

If expm(ak)=e then ‘e’ is the least positive integer.

(ak)e≡1(mod m)

∴ e>0 is the smallest positive integer ∋ ake≡1(mod m)

Then by part (ii) of theorem (1), e is the least positive integer with ke≡0(mod f) and this
convergence has solution

e≡0(mod f/d)…………….(1)
where d=(k,f)

∵ e>0, the least integer satisfying (1) is f/d

Hence e=f/d

= > expm(ak) = f/ (k,f)

Theorem (4):

Let p be any odd prime and let d be any positive divisor od p-1. Then

(i) In every reduced residue system mod p there are exactly Φ(d) numbers ‘a’ ∋,
expp(a)=d. in particular, when d=Φ(p)=P-1
(ii) There are exactly Φ(p-1) primitive roots mod p.

Let p be an odd prime

Consider the set S given by S={1,2,3,……P-1}

For any divisor of P-1, let Ad={x∈S: expp(x)=d} and nd be the number of elements in Ad.

Then nd≥0 for each nd

Also, Ad ⋂ Aδ = Φ for d≠δ where d and δ are divisors of P-1 and ⋃ 𝑑 𝐴d =S and ∑ 𝑑 𝑛d=P-
𝑝−1 𝑝−1

1……(1)

We know that ∑ 𝑑 𝛷(𝑑)=P-1…………(2)


𝑝−1

Where Φ(n) is the Eulers function

∴∑ 𝑑{𝛷(𝑑) − 𝑛𝑑}=0……………………..(3)
𝑝−1

If we are able to s.t nd≤Φ(d)…………….(4)

It follows that each term on left hand side of (3) is non negative.

∴ Φ(d)-nd =0 i.e, Φ(d) =nd for each divisor d of P-1 which is required in part (i) of the
theorem.

Hence it remains to prove only (4)

To do this we s.t either nd=0 or else nd=Φ(d)

If nd≠0 then Ad≠Φ

Choose a ∈ Ad then since expp(a) =d.


We get ad≡ 1(mod P) every power of ‘a’ also satisfies the same congruence.

∴ each of the ‘d’ numbers a,a2,…..ad……..(5) is a solution of the poly congruence

Xd-1≡0(mod P) [expp(x)=d, xd≡ 1(mod P) => xd-1≡0(mod P)]

∵ the numbers in the list (5) are in congruent mod P because expp(a)=d

Since the congruence (6) can have atmost ‘d’ solutions by lagrange theorem. It follows that
numbers in the list (5) are all the solution of (6) . Hence member of Ad is the form
ak(k=1,2,…….d).

Now by above lemma, expp(ak)=d < => (k,d)=1. In other words among the ‘d’ numbers in (5)
there are Φ(d) which have exponent or modulo P.

Thus nd=Φ(d) if nd≠0.Thus theorem is complete.

Note: Suppose P is an odd prime and n≢0(mod P). If x2≡n(mod P) has a solution, we say that
n is a quadratic residue modulo P.

Primitive roots and quadratic residues

Theorem (5): let g be a primitive root mod P, where P is an odd prime. Then the even powers
g2, g4,…gP-1 are the quadratic residues mod P and the odd powers g, g3,…gp-2 are the
quadratic non residues mod P.

Suppose n is even say n=2m. then gn = g2m

 gn = x2 (mod P) where x= gm
 gn is a quadratic residues mod P.
( g is primitive root mod P = > g Φ(P) ≡ 1 (mod P) where P is odd prime)
𝑃−1
But there are distinct even powers g2, g4,…gP-1 mod P and the same number of quadratic
2
residues mod P.

∴The even powers are the quadratic residues & odd powers are the non-residues.

The existence of primitive roots mod Pα.

We shall prone that primitive roots modulo Pα exists, whenever P is an odd prime and α ≥1 is
an integer. Also we establish that primitive roots mod m where m is of the form m=2. P α in
which P is an odd prime and α is an integer ≥1 exist.

Theorem (6): suppose P is an odd prime.Then

(i) A Primitive root g mod P is a primitive root mod Pα for all, α ≥1 if and only if gP-1
≢1(mod P2)………(1)
(ii) There is atleast one primitive root g mod P satisfying the condition (1)

(iii) There is atleast one primitive root g mod Pα whenever α ≥2

Given that P is an odd prime and that g is a primitive root mod P (existence by theorem 4)

(i) if g is a primitive root mod . Pα for α ≥1 then, in particular g is a primitive root mod P2
also.

If g satisfies (i) the condition is necessary.

If g doesnot satisfy (i) then we can show that g1 = g+p is a primitive root mod P and that is
satisfies the condition that g1P-1≢1(mod P2)………..(2)

In fact we have,
𝑃−1
g1P-1 = (g+P)P-1 = gP-1+(P-1)gP-2.P+ 2
gP-3P2+…..+PP-1

= gP-1+(P-1)gP-2.P+ P2.t
𝑃−1 𝑃−1
Where t= gP-3 + gP-4.P+….+PP-3 is an integer;
2 3

So that g1P-1 = gP-1+P2(gP-2+t)-PgP-2

≡ gP-1-PgP-2(mod P2) ≡ 1-PgP-2(mod P2)

Which gives that

g1P-1≡1(mod P2) < => PgP-2≡ 0(mod P2)

< => gP-2≡ 0(mod P) and this contradicts the fact that g is a primitive root mod P.

∴ g is a primitive root mod P satisfying (2)

Thus if ‘g’ is a primitive root mod Pα for each α ≥1, then there is a primitive root mod P
which satisfies (1) conversely, suppose that there is a primitive root g mod P, satisfying (1).

Then we have to show that g is also a primitive root mod Pα for each α ≥2………..(3)

Let t= expα(g).

Then to prove (3), we have to show that

t=Φ(Pα) = Pα-1(P-1) ………..(4)

since gt≡1(mod Pα), we have gt≡1(mod P) and since g is a primitive root mod P, it follows
that Φ(P)/t

so that t= q.Φ(P) for some integer q


Now t/Φ(Pα) = > qΦ(P)/ Φ(Pα)

= > q(P-1) / Pα-1(P-1)


= > q/ Pα-1
= > q= Pβ for some β≤α-1………..(5)

Hence t= Pβ(P-1) so that (3) follows if we show β=α-1

If possible β≠α-1 so that β<α-1( in view of (5)) and hence β ≤α-2.

Then t= Pβ(P-1) divides Pα-2(P-1) =Φ(P)α-1.

Now since Φ(P)α-1 is a multiple of t, we get gΦ(Pα-1)≡1(mod Pα)…..(6)

But, now we how that, if g is a primitive root mod P ∋, gP-1≢1(mod P2) then.

gΦ(Pα-1)≢1(mod Pα) for each α≥2 ……(7)

clearly (7) holds for α=2.

Assume (7) holds for some α

Then because gΦ(Pα-1)≡1(mod Pα-1)

By Euler-Fermet theorem, we can write

gΦ(Pα-1)≡1+kPα-1 for some integer k, with P+k . ….(8)

raising both sided of (8) to the Pth power we get

gΦ(Pα-1)≡(1+kPα-1)P
𝑃(𝑃−1)
= 1+kPα+k2 P2(α-1) +2.P3(α-1)
2

Where r is some integer.

Now 2α-1 ≥ α+1 and 3α-1≥ α+1, since α≥2, the equation (9) gives the congruence.

gΦ(P)≡1+kPα( mod Pα+1), where P/k and this gives

gΦ(Pα)≢ 1 (mod Pα+1), proving that (7) holds for α+1 and hence for all α.

But if β≠α-1 then (6) and (7) contradict each other

Hence β=α-1 and this proves (4).

Thus g is a primitive root mod Pα also.

(ii) Suppose g is a primitive root mod P.

If g satisfies (1) there is nothing to prove otherwise take g1=g+p then, we have proved in the
first part that g1P-1≢ 1(mod P2) and that g1 is a primitive root mod P.
(iii) By theorem (4) , for any odd prime P there is a primitive root mod P.
By part (ii) of this theorem there is a primitive root mod P satisfying (1).
Then by part (i) of this theorem ‘g’ is a primitive root mod Pα for each α≥1

Hence theorem is proved.

Theorem (7):

(i) If P is an odd prime and α ≥1 then there is an odd primitive root g mod Pα.
(ii) Each g of part (i) is a primitive root mod 2 Pα .

(i) suppose ‘p’ is an odd prime and α ≥1.

Then by theorem (6) , there is a primitive root g mod Pα .

Then g1=g+pα is also a primitive root mod Pα

Now on of g and g1 is clearly odd, proving part (i) of the theorem.

(ii) Suppose g is an odd primitive root mod Pα and suppose f= exp2Pα(g).


we wish to prove f=Φ(2Pα)…..(1) from which part (ii) follows.

Now f/ Φ(2Pα) = Φ(2). Φ(Pα)= Φ(Pα) and since

gf≡1(mod 2.pα) we have gf≡1(mod pα) so that Φ(Pα)/f ; Proving f=Φ(Pα), which is (1).

thus g is primitive root of mod 2P α.

The non existence of primitive roost mod m when m≠2P α

We prove that if m is not an element of the set { x: x=1,2,4, Pα or 2. Pα where P is an


odd prime } then primitive roots mod m do not exist.

Theorem (8): Suppose m is a natural number other than 1,2,4, Pα or 2 Pα where P is an


odd prime. Then for any integer a with greatest common divisor (a,m) =1 we have
aΦ(m)/2≡1(mod m) ( which shows that there are no primitive roots mod m)

Since m≥1, it has canonical representation m=qα.p1α1.p2α2…..prαr where α≥0, pi are odd
primes and αi≥1 for i=1,2,3…..r

In case r=0(i.e, m is of the form m=2α); we have α≥3(∵ m≠1,2,4) and in this case
primitive roots mod m do not exist (by theorem (3)).
Also m is not of the form 1,2,4,pα or 2pα it follows that α≥2 if S=1 and r≥2 if α=0 or1.

To prove (1), suppose a is an integer which is relatively prime to m. choose a primitive


root mod p1α (the existence which is generated by theorem (6)) and then choose k∋,
a≡gk(mod p1α1). This is possible by theorem (5).

Then we have aΦ(m)/2≡ gΦ(m)/2(mod p1α1)


Φ(2α).Φ(pα1)….Φ(prαr)
≡ gk (mod p1α1)
2

≡gtΦ(P1α1)(mod P1α1)

Where t= kΦ(2α)Φ(p2α2)…….Φ(prαr) /2
Φ(2α)
Then we observe that t is an integer. In fact, if α≥ 2 then Φ(2α) is even so that is
2
an integer proving t is also an integer.

Also , if α=0 or 1 then r≥2 and the factor Φ(p2α2) is even so that and hence t is an integer.

Then aΦ(m)/2≡ [gΦ(p1α1)]t(mod p1α1) ≡1(mod p1α1)

Hence aΦ(m)/2≡1(mod p1α1)

Similarly we can prove aΦ(m)/2≡1(mod piαi)

For i=1,2,…..r …………..(2)

Now we show that aΦ(m)/2≡1(mod 2α) …………….(3)

In fact, if α≥3 , then (a,m)=1 requires a to be odd and therefore , by (2) in the proof of
theorem (3) we get aΦ(2α)/2≡1(mod 2α)……………..(4)

Now since Φ(2α)/ Φ(m) , we get (3) if α≥ 3, in view of (4)

If α=0 or 1 we have aΦ(2α)≡1(mod 2α) ……(5)

Also in this case r≥1 so that

Where S is an integer.
Φ(m)
Hence Φ(2α)/ and therefore(5) implies (4). Thus (3) holds for all α.
2

Multiplying the congruence in (2) and (3) we obtain aΦ(m)/2≡1(mod m)

And this shows that a cannot be a primitive root mod m.


Φ(m)
(iii) if m is not of the form 1,2,4,pα or 2pα where p is an odd prime, then < Φ(m)
2

Hence expm(a) <Φ(m) for any ‘a’ with greatest common divisor (a,m)=1 . This is no integer
‘a’ with (a,m)=1 is a primitive root mod m, proving (ii) of the theorem.
The number of primitive roots mod m:

Theorem 9: if m is∋, a primitive root g mod m exists then there are exactly Φ(Φ(m))
incongruent primitive roots. The set S of these primitive roots is given by S={gn: 1≤n≤Φ(m);
(n,Φ(m) )=1}

Since g is a primitive root mod m, we have expm(g) = Φ(m)

Therefore by lemma (1) expm(gn) = expm(g)

< = > (n,Φ(m)) =1

∴ each element of S is a primitive root mod m conversely, if ‘a’ is a primitive root mod m
then a≡ gk(mod m) for some k∈( 1,2,…., Φ(m)).

Hence expm(gk) = expm(g) = Φ(m) and lemma (1) implies (k1Φ(m)).

Thus every primitive root is a member of S. Hence S is the set of all primitive roots mod m.

Since S has Φ(Φ(m)) elements the theorem is complete.

Legender’s Symbol:

Let p be an odd prime if n≢0(mod p) we define legender’s symbol (n/p) as follows:

+1 if nRp
(n/p) ={
−1 if nȒp

If n≡0(mod p) we define (n/p) =0

Jacobi Symbol:

Suppose p is a positive odd integer with prime factorization p=∏ri=1 piαi and n is any integer.
n
Then the Jacobi symbol (p ) is the legender’s symbol for each i.

n
We define ( 1) =1

n
The Jacobi symbol (p ) is also denoted by (n/p)

NOTE :
n n
(i) The possible values of (p ) are -1,0 or 1, the possible values of (pi ) are so for
each i.
n
(ii) (p )=0 < => (n,p) >1
Suppose (n,p) >1
Then ∃ pk/n , pk/pa for some k with 1≤k≤r.
n n n
= > (pa ) = (∏ri=1;i≠k(pi)αi) (pk)
n
=0 (∵(pk)=0 as pk/n )
n
Conversely , if (p) =0
n
Then (∏ri=1(pi)αi) =0
n
This implies k∋, (pk) = 0 for some k with 1≤k≤r
= > pk/n = > pk/n, pk/p
= > pk/ (n,p) = > (n,p) > 1
n
Thus (p) =0 < = > (n, P) > 1
n n
(iii) If the congruence x2≡n(mod P) has a solution then (pi) =1 for each I and hence (p)
=1
n n n
But if (p) =1 , it is not necessary that (pi) =1, for each I, for (p) can be 1 if an even
n
number of factors -1 appears in (∏ri=1(pi)αi)
(iv) When p is prime , the Jacobi symbol is same as the legender’s symbol.
Hence Jacobi symbol is a generalization of legendr’s symbol.

Properties of Jacobi symbol by using the properties of legendre’s symbol.

Theorem (10): if P and Q are odd positive integer, then


m n mn
(i) ( p )(p) = ( )
p
n n n
(ii) (p)(q) = (pq)
m n
(iii) ( p )=(p), whenever m=n(mod p)
a²n n
(iv) ( p )=(p), whenever (a,p)=1

let P= ∏ri=1 piαi , Q= ∏rj=1 qjβj ; where pi are distinct, qj are distinct odd primes ( while some of
pi any be equal to qj and vice versa)
mn mn αi
(i) By definition ( ) =∏ri=1( )
p pi
mn m
But since the lengendre’s symbol is completely multiplicative, we get ( pi ) =(pi)
n
(pi)
mn mn αi n αi m n m n
∴( ) =∏ri=1( ) (pi) =∏ri=1( pi)αi . ∏ri=1( pi)αi = ( p ) (p)
p pi
n n n
(ii) (pq) = ∏ri=1( pi)αi ∏sj=1( qj)βi , by definition

n n
= (p)(q)

(iii) m≡n(mod p) = > m≡n(mod pi) for each i, 1≤i≤r


m n
= > (pi) = (pi) , i-1,2,….r
m n
= > ∏ri=1( pi)αi = ∏ri=1( pi)αi
m n
= > ( p ) = ( p)
a²n a²n αi
(iv) ( p ) =∏ri=1( ) ; by definition.
pi
a² αi n αi
= ∏ri=1( pi) (pi) ; because the legendre symbol is completely multiplicative
n a² n
= ∏ri=1( pi)αi (∵ (pi) =1) = (p) )

Properties of legendre symbol:

(i) (1/p) = 1 for all odd primes P


(ii) (m2/P) = 1 for all integers on all odd primes P
(iii) (m/P) = (n/P) is a periodic function of n with period P.

NOTE: legendre symbol (n/p) is a completely multiplicative function of n. that is, if P is an


m n mn
odd prime then ( p )(p) = ( p ) for all m and n

Theorem (11): if P is an odd positive integer then


−1
(i) ( p ) = (-1)P-½
2
(ii) (P) = (-1)P²-⅛

Write P=p1p2…….pm where Pi are odd primes not necessarily distinct.

Then P=p1p2…….pm = ∏m m
i=1 Pi = ∏i=1(1 + (Pi − 1))

= 1+∑m
i=1(Pi-1) +∑i≠j(Pi-1)(Pj-1)+……….

Each (Pi-1) being even , each sum after the 1st in the above will be divisible by 4.

P≡= 1+∑m
i=1(Pi-1)(mod 4)

P−1 Pi−1
Or ≡ ∑m
i=1 (mod 2)
2 2
Pi−1 P−1
∴ (-1) ∑m
i=1 = (-1)
2 2

−1 −1 Pi−1
∴ ( p ) = ∏m m
i=1 Pi = ∏i=1(−1) 2

Pi−1 P−1
= (-1) ∑m
i=1 = (-1) . this proves (i)
2 2

Now P2 = ∏m 2 m 2
i=1 Pi = ∏i=1(1+ (Pi -1))

= 1+ ∑m 2 2 2
i=1(Pi -1) + ∑i≠j(Pi -1)(Pj -1)+….

Since each P is odd , we have Pi2-1 ≡0(mod 8)

So that P2-1 ≡ ∑m 2
i=1(Pi -1) (mod 64)

P²−1 Pi²−1
This implies ≡ ∑m
i=1 (mod 8)
8 8

Pi−1 P²−1
∴ (-1) ∑m
i=1 = (-1)
8 8

2 2 Pi²−1 Pi²−1 P²−1


Hence (P) = ∏m m
i=1(Pi) = ∏i=1(−1) = (-1)∑m
i=1 = (-1)
8 8 8

This proves (ii)

Quadratic reciprocating law for Jacobi symbols:


P Q
Theorem (12): If P and Q are positive odd integers with (P,Q)=1, then (Q) ( P ) =
P−1 Q−1
(−1) 2 2

Write P=p1p2…….pm , Q=q1q2...........qn, where Pi, qi are odd primes.


P Q pi qi
Then (Q) (P ) = ∏m n
i=1 ∏j=1( qi)(pi) = (-1) ……………. (say)
r

pi qi
But (qi)(pi) = (-1)½(Pi-1)½(qi-1)

(pi−1)(qi−1)
Hence r= ∑m n
i=1 ∑j=1 4

(pi−1) (qi−1)
= (∑m
i=1 )(∑nj=1 )
2 2

But since pi,qi are odd primes, we have


(pi−1) P−1
(∑m
i=1 )= (mod 2)
2 2

(qi−1) Q−1
And (∑nj=1 )= (mod 2)
2 2

Using these in the above , we get


P−1 Q−1
r= 2 2

P Q P−1 Q−1
Hence (Q) (P ) = (−1) 2 2

Definition: Diophantine equation

A Polynomial equation to be solved in integers is called Diophantine equation against


Diophantus of Alexandria

Example:

y2 = x3+k for a given integer k,

x2+2xy2+3y3=10, x2+y3=100 are some Diophantine equation.

Application of Legendry and Jacobin symbols in solving the Diophantine equation:

Theorem (13): the Diophantine equation y2 = x3+k…….(1) has no solution if k is of the form
k=(4n-1)3 -4m2 ……(2) where m and n are integers ∋, no prime p≡ -1(mod 4) divides m.

Suppose if possible (1) has a solution say x,y i.e., y2= x3+k, where k is given by (2)

Since k≡(4n-1)3-4m2, we get

k≡-1(mod 4) so that

y2 = x3-1(mod 4)……………(3)

For any integer ‘z’ we have always z2 ≡0 or 1(mod 4) in particular y2 ≡0 or 1(mod 4)

If x is even or x≡-1(mod 4) then by (3) we get y2 ≡1(mod 4) or y2≡-2(mod 4)

That is y2≡ 3 or 2(mod 4), which is impossible.

∴ x must be of the form x≡1(mod 4).

Write a = 4n-1, so that K=a3-4 m2

Then , by (1), we get y2=x3+(4n-1)3-4m2

or y2+4m2 =
x3+a3 = (x + a) (x2 - ax+ a2)

Now, since x ≡ 1 (mod 4) & a ≡ -1 (mod 4),

we get x2 - ax+ a2 ≡ 1-a+ a2 ≡ 3 ≡ 1 (mod 4) …….. (4)


Hence x2 - ax+ a2 is odd.

By (4), x2 - ax+ a2 being an odd no. of the form 4s-1,

All its prime factors. Can’t be of the form 4l+1.

∴ For some prime P, of the form P ≡ -1 (mod 4) we have P/ x2-ax+ a2

Hence P/y2+4m2

In other words y2+ 4m2 ≡ 0 (mod 4) or

y2 ≡ -4m2 (mod P) for some prime P of the form P ≡ -1(mod 4),

but, P being of the form P ≡ -1(mod 4), P-m by the hypothesis.


−4m² −1 p−1 4l−1−1
∴( ) = ( p ) = (-1) = (-1) ≡ -1
p 2 2

Since P ≡ -1(mod 4) => P = 4l-1 for some l.

i.e, the congruence.

U2 ≡ -4m2 (mod P) has not solution.

But this is a # to the fact y2 ≡ -4m2 (mod P)

Hence there doesn’t exist any so l+n for (1) in into’s this proves the theorem.

𝑟
Example: Let P be an odd prime of the form 4k+1.Then prove that ∑𝑝−1
𝑟=1 𝑟(𝑝) = 0

Observe that 1≤ r ≤ P-1

 1-P ≤- r ≤- 1 => 1 ≤ P-r ≤ P-1


i.e., if r ranges over the set {1,2,-,P-1} then P-r also ranges over the same set.
𝑟 𝑝−𝑟 −𝑟
∴ ∑𝑝−1 𝑝−1
𝑟=1 𝑟(𝑝) =∑𝑟=1 (𝑝 − 𝑟)( ) = ∑𝑝−1
𝑟=1 (𝑝 − 𝑟)( 𝑝 )
𝑝

−𝑟 −𝑟
= ∑𝑝−1 𝑝−1
𝑟=1 (𝑝)( 𝑝 ) - ∑𝑟=1 (𝑟)( 𝑝 )

p−1 𝑟 p−1 𝑟
=∑𝑝−1
𝑟=1 𝑝 (−1) (𝑝) - ∑𝑝−1
𝑟=1 𝑟 (−1) (𝑝 )
2 2

𝑟 𝑟 p−1
= ∑𝑝−1 𝑝−1
𝑟=1 (𝑝) - ∑𝑟=1 (𝑟)(𝑝) (∵(-1) = (-1)2k=1)
2

𝑟 𝑟
∴2∑𝑝−1 𝑝−1
𝑟=1 (𝑟)(𝑝) =p ∑𝑟=1 (𝑝)
p−1
If P is an odd prime, W.K.T exactly, no’s in the set {1,2,-,1,P-1} are quadratic residues
2
p−1
mod P and the rest no’s are quadratic non residue modulo P.
2

𝑟
Hence ∑𝑝−1
𝑟=1 (𝑝) = 0

𝑟 𝑟
∴2∑𝑝−1 𝑝−1
𝑟=1 (𝑟)(𝑝) = 0, so that ∑𝑟=1 (𝑟)(𝑝) = 0, in case p≡ 1(mod 4)

Geometric representation of partitions:

Introduction: we now look into another branch of number theory called additive number
theory.

Partition: Given any positive int. ‘n’ can be written as sum of integers from a given set A
where A= {a, a2,......}, ai’s are special numbers such as primes, squares, cubes etc., each
representation of ‘n’ as sum of elements of A is called partition of ‘n’. The arithmetical
function A(n) gives the number of partitions of n.

Gold Bach conjecture: Every int. n ≥ is sum of 2 –odd primes.

Eg: 6=3+3, 8=5+3, 20=5+5

If A = {Pi / Pi is an odd prime }, for n ∈ 1N A(n) gives the number of solutions of the
equation n=P1+P2 where P1, P2 ∈ A

Eg: 10=5+5 = 7+3

A(10) = 2 (∵10 is written in 2-ways)

Representing by squares:

Any given integer k ≥ 2 can be written as n= x 12 + x22 +……. xk2 . rk(n) gives the number of
solutions for the equation n= x12 + x22 +……. xk2 where xi2 can be positive or negative or
zero and the order of sum is taken in account.

For k=2,4,6,8, rk(n) is expressed in terms of divisor function r2(n) = 4{d1(n) – d3(n)} [ where
d1(n) ≡ 1(mod 4) and d3(n) ≡ 3(mod 4)] where d1(n) and d3(n) are number of divisors of ‘n’
congruent to 1 and 3 mod 4 respectively.

Example : for n=5


We have 5 = 22 + 12 =12 + 22 = (-2)2 + (-1)2 = (-1)2 +(-2)2 = (-2)2 + 12 + 12 + (-2)2 = (-1)2 + 22
= 22 + (-1)2

∴ rk(5)=8

For k= 4 we have

8 σ(n) if n is odd
r4(n) = ∑d/n d {24 ∑d d if n is even
;d is odd
n

For k ≥ 5

rk(n) is expressed as rk(n) = ρk(n) + Rk(n) where ρk(n) is principle term and is given by the
series
k
∏ (n)k/2 q G(h;q) k −2ᴨinh
ρk(n) = 2
k .∑∞
q=1 ∑n=1( ) .e
Γ( ) q q
2


(Γ(n) = ∫0 𝑒-x xn-1 dx)

The series for ρk(n) is called the singular series and the numbers G(h;q) are quadratic gauss
q −2ᴨinhr²
sum given by G(h;q) = ∑k=1 e . Rk(n) is the remainder term.
q

Note: rk(n) is also defined as the coefficient of xn in the power series expansion of the kth
power of the series ϑ = 1+2∑∞n=1 x

Waring’s Problem: Given a positive integer k there is an integer S (depending only on k)


∋,the equation n= x1k + x2k +……. xsk …..(1) has solution for every n≥1. The
partition from A(n) gives the number of solutions of equation (1) and the problem is to decide
if ∃ an S∋, A(n) ≥ 1 for all n.

This problem is named after English Mathematician E.Waring who started in 1770 that every
into n≥1 is the sum of 4 squares, of 9 cubes, of 19 fourth powers etc. If S exists for a given k
then there is a least value of S is denoted by g(k).

Unrestricted Partition Function p(n):

For n ∈ N , P(n) is number of solutions of n=ai1+ai2+….. where the number of summons is


unrestricted repetition is allowed and order is immaterial then p(n) is called unrestricted
partition function or partition function. The summands are called parts.

Example: for n=5

Then 5=3+2 =1+4 =0+5=1+2+2 =1+1+1+1+1=1+1+3=1+1+1+2

∴ P(5) = 7
For n=4

Then 4=4+0 =3+1=2+2= 1+3 = 1+1+1+1

∴ P(4) = 5

Representation of partition:

The way of representing partition geometrically by using a display of lattice points id called a
graph.

Eg: 5=3+2 can be represented geometrically as

∴ 5= 2+2+1 --> number of parts

Theorem (14): the number of partition of ‘n’ into m parts is equal to the number of partitions
of ‘n’ into parts, the largest of which is m

We prove this theorem geometrically

Consider a partition of 15 given by 6+3+3+2+1. Which can be represented by lattice points as

Reading the lattice points vertically we get another representation of 15 as 5+4+3+1+1+1.


These 2 partitions are said to be conjugate to each other. The largest part of 1st representation
i.e, 6.

Number of parts in the other i.e, 5+4+3+1+1+1 (6parts)

Generating Function For Partitions:

If f(n) is an arithmetic function, then the sum F(x) of the power series ∑ f(n)xn is called the
generating function of f(n). i.e, F(x)= ∑ f(n)xn is called the generating function of f(n).

Theorem (15): [Eulers theorem for the generating function of P(n)] for │x│<1 we have
1
∏∞
k=1 = ∑∞
n=0 P(n)x , where P(0) =1
n
1−xᴷ

1 1 1 1
Consider ∏∞
k=1 1−xᴷ =(1−x)(1−x²)(1−x³)…….

= (1-x)-1(1-x2)-1(1-x3)-1……

= (1+x+x2+x3+….)(1+x2+x4+x6+….)(1+x3+x6+….)

= 1+∑∞
k=1 a(k)x
k
T.P: a(k)=P(k)

Consider xk term in 1st series, x2k2 term in 2nd series where each ki≥0 then the product
xk1.x2k2.x3k3….xmkm = xk (say)

= > xk1+2k2+3k3+….mkm =xk

= > k= k1+2k2+3k3+…..mkm

= > (1+1+….. k1times)+(2+2+….k2times)+…….+(m+m+…..kmtimes)

Thus each partition of k gives one such term xk also, each term xk comes from a
corresponding partition of k

∴ a(k) = number of partitions of k = P(k)


1
∴ ∏∞ ∞ ∞ ∞
k=1 1−x = 1+ ∑k=1 a(k) = P(0) +∑k=1 P(k)x = ∑k=0 P(k)x
xᴷ k k

Second proof:
1
Define Fm(x) = ∏m
k=1 1−xᴷ

1
F(x) = ∏∞
k=1 1−xᴷ = lim Fm(x) for 0≤x≤1
m→∞

Now, ∑ xᴷ is absolutely convergent if 0≤x≤1

= > ∏m
k=1(1 − xᴷ) is also absolute convergent

1
= > ∏∞
k=1 1−xᴷ is also absolute convergent.

∴ Fm(x) is absolute convergent.


1 1 1
Now, Fm+1(x) = ∏m+1 m
k=1 1−xᴷ = (∏k=1 1−xᴷ)(1−xᵐˉ¹)

1
= Fm(x) (1−xᵐ⁺¹) ≥ Fm(x)

∴ Fm+1(x) ≥ Fm(x)

= > { Fm+1(x)} is an increasing series

∴ for each fixed x,

Fm(x) ≤ Fm+1(x) ≤ Fm+2(x) ………..≤ F(x) ; 0≤x≤1

∴ Fm(x) ≤ F(x)

(by previous proof) , Fm(x) = 1+∑∞


k=1 Pm(k)x where pm(k) is number of partitions of ‘k’ into
k

parts not exceeding ‘m’


If m≥k then pm(k) = p(k)

In other words, lim pm(k) = p(k)


m→∞

Fm(x) = 1+∑∞
k=1 Pm(k)x
k

= pm(0) + 1+∑∞
k=1 Pm(k)x
k

=∑∞
k=0 P(k)x
k

=∑∞ ∞
k=0 Pm(k)x + ∑k=m+1 Pm(k)x
k k

=∑∞ ∞
k=0 P(k)x + ∑k=m+1 Pm(k)x
k k

(∵ for m≥ k i.e, k≤m; Pm(k)=P(k))

≡ ∑m
k=0 P(k)x
k

∴ ∑m
k=0 P(k)x ≤ Fm(x) ≤ F(x) ………(1)
k

∴ ∑m ∞
k=0 P(k)x is convergent ∵ seq of partial sum of the series ∑k=0 P(k)x is convergent
k k

= > ∑∞
k=0 P(k)x is convergent
k

∵ Pm(k) ≤ P(k)

∑∞ ∞
k=0 Pm(k)x ≤ ∑k=0 P(k)x ≤ F(x) (∵ from ..(1))
k k

∴ for each fixed x, ∑∞


k=0 Pm(k)x converges uniformly
k

Let m-> ∞. F(x) = lim ∑∞


k=0 Pm(k)x
k
m→∞

= ∑∞ ∞
k=0 lim Pm(k)x = ∑k=0 P(k)x for 0≤x<1
k k
m→∞

By analytic continuation, the identity holds even in the disk │x│<1

Note:
1
(i) The generating function ∏∞
m 1−x²ᵐˉ¹ gives the number of partitions of m into odd number of
parts.
1
(ii) The generating function ∏∞
m gives the number of partitions of m into even number of
1−x²ᵐ
parts.

(iii) ∏∞
m(1 + xᵐ) is the generating function that enumerates the number of partition of n into
unequal parts.
Generating function number of partition of n into parts which are:

1
∏∞
m odd
1−x²ᵐˉ¹

1
∏∞
m even
1−x²ᵐ

1
∏∞
m squares
1−xᵐ²

∏∞
m(1 + xᵐ) unequal

1
∏P (P is a prime) primes
1−xᴾ

∏∞
m(1 + x²ᵐˉ¹) odd and unequal

∏∞
m(1 + x²ᵐ) even and unequal

∏∞
m(1 + xᵐ²) distinct square

∏P(1 + xᴾ) ( P is a prime) distinct primes

Euler’s Pentagonal number theorem:

The reciprocal of the generating function of P(n), ∏∞ ∞


m(1 + xᵐ) = ∑n=−∞(−1)ⁿx
w(n)
, which is
known as Euler’s Pentagonal number theorem.

Theorem (16): ∏∞ m(1 + xᵐ) = 1+∑m=1{ Pe(n) - Po(n)}x ,where Pe(n) is the number partition
n

of n into an even number of unequal parts and Po(n) is the number of partition of n into an
odd number of unequal parts.


Write ∏∞
m(1 + xᵐ) = 1+∑m=1 a(n)x
n

Where a(n) is a partition function.

The coefficient a(n) is +1 if xn is a product of even number of terms and -1 otherwise.

a(n) = Pe(n) - Po(n) for each ‘n’

where Pe(n) is the number of partitions of ‘n’ into even number of unequal parts and Po(n) is
the number of partitions of ‘n’ into odd number of unequal parts.

Pentagonal numbers:

Pe(n) = Po(n) for all n except those numbers belonging to a special set called pentagonal
numbers. The pentagonal numbers are 1, 5, 12, 22, ….. which are the sum of the terms in the
arithmetic progression 1, 4, 7, 10, …..3n+1,……

1, 4, 7, 10,….

1=1

1+4=5

1+4+7=12

1+4+7+12=22

Pentagonal figure :

N=5
N=12

N=22

If w(n) denotes the sum of 1st n terms in the progression then w(n) = 1+4+7+….(3n-
2)+(3n+1)

W(n)= ∑n−1 n−1 n−1


k=0(3k + 1)= 3∑k=0 k + ∑k=0 1

3(n−1)n 3n²−3n+2n 3n²−n


= +n= =
2 2 2

3(−n)2 −(−n) 3n²+n


Now, w(-n) = =
2 2

W(n) and w(-n) are called as pentagonal numbers.

For n=1; w(1)=1

For n=2; w(2)=5

For n=3; w(3)=12

For n=4; w(4)=22

Euler’s Pentagonal number theorem:

Theorem (16): if │x│<1 we have; ∏∞


m=1(1 − xᵐ) = 1-x-x +x +x -x -x +………
2 5 7 12 15

= 1+ ∑∞ n w(n)
n=1(−1) (x +xw(-n)) =
∑∞
n=−∞ (−1) xn w(n)

Let 0≤x≤1

Define Po=So=1, for n≥1

Let Pn= ∏nr=1(1 − xʳ)

Sn = 1+ ∑∞ n w(n)
n=1(−1) {x +xw(-n)}
Claim: │Sn-Pn│≤ nxn+1

Define Fn = ∑nr=0(−1)r(pn/pr)xrn+g(r)
r(r+1)
Where g(r) = 2

F1 = ∑1r=0(−1)r(pn/pr)xrn+g(r)
p₁ p₁
= pₒ -p₁ x1+g(1) (g(1)= 1)

= p1-x2 (pn = ∏nr=1(1 − xʳ) = > p1=1-x)

= 1-x-x2

S1 = 1-1{xw(1)+xw(-1)}

= 1-x-x2

∴ F1=S1

Consider Fn-Fn-1 = ∑nr=0(−1)r(pn/pr)xrn+g(r) - ∑n−1 r


r=0 (−1) (pn-1/pr)x
r(n-1)+g(r)

Pn
= (-1)nPnxn2+g(n)+ ∑nr=0(−1)r(pn/pr)xrn+g(r) - ∑n−1 r
r=0 (−1) (pn-1/pr)x
r(n-1)+g(r)

[ Pn=∏nk=1(1 − xᴷ) = ∏n−1


k−1(1 − xᴷ)(1 − xⁿ)= Pn-1(1 − xⁿ)]

Pn−1
= (-1)nxn²+g(n) +∑n−1
r=0 (−1)ʳ (1 − xⁿ) xrn+g(r) - ∑n−1 r
r=0 (−1) (pn-1/pr)x
r(n-1)+g(r)
Pr

Pn−1
=(-1)nxn²+g(n) +∑n−1
r=0 (−1)ʳ xrn+g(r) - ∑n−1 r
r=0 (−1) (pn-1/pr)x
r(n-1)+g(r)
- ∑n−1 r
r=0 (−1) (pn-1/pr)x
r(n-
Pr
1)+g(r)

Pn−1
=(-1)nxn²+g(n) +∑n−1
r=0 (−1)ʳ xr(n-1)+g(r)(xr-1) - ∑n−1 r
r=0 (−1) (pn-1/pr)x
r(n-1)+g(r)
Pr

xʳˉ¹ xʳˉ¹ −(1−xʳ) −1


( Pr = ∏r = = pr−1
k=1 (1−xᴷ) ∏r−1 (1−xᴷ)(1−xʳ)
k=1

−1 Pn−1
= (-1)nxn²+g(n) + ∑n−1 r
r=1 (−1) Pn-1x
rn+g(r-1)
(Pr−1) - ∑nr=0(−1)ʳˉ¹ Pr−1 xrn+g(r-1)

Pn−1 Pn−1 Pn−1


= (-1)nxn²+g(n) + ∑n−1 r-1
r=1 (−1) Pe−1 x
rn+g(r-1)
- ∑n−1
r=1 (−1)ʳˉ¹ Pr−1 x
rn+g(r-1)
– (-1)r-1Pn−1 xn²+g(n-1)

= (-1)nxn²+g(n) + (-1)nxn²+g(n-1)
n(n+1) 3n²+n
[ n2+g(n) = n2+ = = w(-n)
2 2

n(n−1) 3n²−n
n2+g(n-1) = n2+ = = w(n) ]
2 2

= (-1)n wx(-n) +(-1)n xw(n) = (-1)n { xw(n) + xw(-n) }


= Sn – Sn-1

∴ Fn-Fn-1 = Sn-Sn-1

=> Fn-Sn= Fn-1- Sn-2 = Fn-3-Sn-3 ………= F1-S1 = 0

∴ Fn=Sn
Pn
Fn = ∑nr=0(−1)r(pn/pr)xrn+g(r) = Pn+ ∑nr=1(−1)ʳ Pr xrn+g(r)

Pn
Fn-Pn = ∑nr=1(−1)ʳ Pr xrn+g(r)

Pn
│Fn-Pn│≤ ∑nr=1(−1)ʳ│ Pr │xrn+g(r)

∵ 0≤x≤1 = > xrn+g(r) ≤ xn+1


Pn ∏n (1−xᴷ)n
= ∏k=1
r (1−xᴷ) = ∏k=r+1(1 − xᴷ) ≤ 1
Pr k=1

∴ │Fn-Pn│ ≤ ∑nr=1 xn+1 = nxn+1

∵ Fn=Sn

= > │Sn-Pn│ ≤ nxn+1

As n -> ∞ ; nxn+1 -> 0

∴ lim Sn = lim Pn
n→∞ n→∞

But lim Pn = lim ∏nr=1(1-xʳ) = ∏nr=1(1-xʳ)


n→∞ n→∞

lim Sn = lim {1+ ∑nn=1(−1)n{xw(n)+xw(-n)}}


n→∞ n→∞

= 1+ ∑∞ n w(n)
n=1(−1) {x +xw(-n)}

∴ ∏∞ ∞
m=1(1 − xᵐ) = 1+ ∑n=1(−1) (x
n w(n)
+xw(-n)) = ∑∞ n w(n)
n=−∞ (−1) x = 1-x-x2+x5+x7-x12-
x15+………

F. Franklin’s Combinatorial proof Euler pentagonal number theorem:

We have ∏∞ ∞ n
𝑚=1(1 − 𝑥ᵐ) =1+ ∑𝑚=1(pe(n)-Po(n))x .

Where Pe (n) us the no. of partitions of n into an even number of unequal parts and Po (n) is
the number of partitions of n into an odd no. of unequal parts.
Claim : unless n is a pentagonal number there is a 1-1 correspondence b/w partitions of n into
an odd & even no. of unequal parts i. e Pe (n) = Po (n) if n is not a pentagonal no.

We shall prone this claim by the use of graph us of partitions. We 1st define few term related
to graphs of partitions.

(i) we say that the graph (of a partition) is in standard form if the parts are arranged
in decreasing order.

(ii) the longest line segment connecting points in the last row is called the based of the
base of the graph & the no. of lattices points on the base is denoted by b.; b≥1
always.
(iii) the longest 45o line segment joining the last point in the 1st row with the other
points in the graph is called the slope and the no. of lattice points on the slope is
denoted by S ; S=4, with these basic definition, we define two operation A&B on
the graph.
The operation A moves the point on the base 80 that they lie on a line parallel to
the slope.
Operation B moves the points on the slope so that they lie on a line parallel to the
base.

We call an operation permissible if it P reserves the standard for of the graph, i.e. if the new
graph again has unequal parts arranged in decreasing order.

If A is permissible we got a new partition of ‘n’ into unequal parts but the no. of parts is 1
less than before.

If B is permissible we get a new partition of n into unequal parts but 1 greater than before

∴ If for every partition of n, exactly 1 of A or B is permissible

There will be a 1-1 correspondence b/w partition of ‘n’ into odd & even no. of unequal parts
so that Pe (n) = Po (n) for such ‘n’.

We shall now see under what condition exactly one of A or B is permissible.

So see this we consider 3 cases.

Case (i) : b < s

If b<s then b≤s-1 so operation A is permissible but B is not permissible since destroys the
standard form.

Case (ii) : b = s

If b=s, operation B is not permissible since if results in a new graph that is now in standard
form. Operation A is permissible except when the base & slope intersect as illustrated in the
following figure in which case the new graph is not in standard form.
Case (iii) : b>s

If b>s, operator A is not permissible where as B is permissible. Except when b=s+1 and the
base and slope intersect as illustrated in following figure in case the new graph contain 1
equal parts.

∴ Exactly one of the operations A or B is permissible with the 2 exceptions noted in case (ii)
& (iii) consider the 1st exceptional case, shown in figure (1) & suppose these are k rows in
graph then b=k so that the no ‘n’ is given by
3k²−k
N= k+(k+1)+(k+2)+……….+(2k-1) = 2

For this partition of n we have an = w(k)


Extra partition into even parts if K is even & an extra k partition into odd parts if K is

Odd so Pe(n) – Po(n) = (-1)k

In the 2nd exceptional case mentioned in the case (iii) a shown in fig. (2), there is an
additional lattice Pt in each row so.
3k²−k 3k²+k
N= +k = = w(-K)
2 2

Recursion Formula and Bounds For p(n):

Theorem (17): euler recursion formula for P(n):

Let P(0) and define P(n) to be 0 if n<0. Then for n≥1 we have P(n) – P(n-1)-P(n-2)+P(n-
5)+P(n-7)+… =0

Or

P(n) = ∑∞ k+1
k=1(−1) {P(n-(W(k))+P(n-W(-k))}, where W(k), W(-k) are euler pentagonal
numbers.

By euler’s identity we have


1
∏∞
m=1 = ∑∞
m=0 P(m)x ……. (1)
m
1−xᵐ

where P(0) =1 and │x│ < 1

From Eulers pentagonal theorem,

∏∞ m ∞ k W(k)
m=1(1 −x ) = 1+ ∑k=1(-1) {x +xW(-k)} …….(2)
1
(1)×(2) = > ∏∞ ∞ ∞
m=1( (1−xᵐ). (1 − xᵐ)) = ∑m=0 P(m)x (1+ ∑k=1(-1) {x
m k W(k)
+xW(-k)}

= > 1= ∑∞ ∞ ∞
m=0 P(m)x + ∑m=0 ∑k=1 P (m)xᵐ (−1)ᴷ{x
m W(k)
+xW(-k)}

= > 1 =∑∞ ∞ ∞
m=0 P(m)x + ∑m=0 ∑k=1 P (m)xᵐ (−1)ᴷx
m W(k)
+ ∑∞ ∞
m=0 ∑k=1 P (m)xᵐ (−1)ᴷx
W(-k)

= > 1 =∑∞ ∞ ∞
m=0 P(m)x + ∑m=0 ∑k=1(−1)ᴷP (m)xᵐ x
m W(k)
+ ∑∞ ∞
m=0 ∑k=1(−1)ᴷP (m)xᵐ x
W(-k)

= > 1 =∑∞ ∞ ∞
m=0 P(m)x + ∑n ∑k=1(−1)ᴷP (n − W(k))x + ∑n ∑k=1(−1)ᴷP (n − W(−k))x
m n n
( n=m+w(k) = > m=n-W(k))

Equating the coefficients of xn on both sides

= > 0= P(n) + ∑∞ ∞
k=1(−1)ᴷP (n − W(k)) + ∑k=1(−1)ᴷP (n − W(−k))

= > P(n) = - ∑∞
k=1(−1)ᴷ [ P(n-W(k)) + P(n-W(-k))]

= > P(n) = - ∑∞
k=1(−1)ᴷ⁺¹ [ P(n-W(k)) + P(n-W(-k))]

= > P(n) –P(n-1) – P(n-2) +P(n-5) +P(n-7) +…. =0

UPPER BOUND FOR P(n) :

1
Theorem (18): If n≥1 we have F(x) =∏∞ ∞ ∞
n=1 1−xⁿ = 1+ ∑n=1 P(n)xⁿ =∑n=0 P(n)xⁿ

1
Restrict ‘x’ in the interval 0<x<1 F(x) =∏∞ ∞
n=1 1−xⁿ = 1+ ∑n=0 P(n)xⁿ > P(n)x
n

Apply log on both sides


1
= > log F(x) = log (∏∞ n
n=1 1−xⁿ) > log (P(n)x )

1
= > log F(x) = ∏∞
n=1 log(1−xⁿ) > log P(n)+log x
n

= > log F(x) > log P(n)+log xn

= > log F(x) > log F(x)+nlog x

= > log P(n) < log F(x)-nlog x

= > log P(n) < log F(x)+nlog x-1


1
= > log P(n) < log F(x)+nlog x ……………. (1)

We have log F(x) = ∑∞ ∞ ∞


n=1 log(1 − xⁿ) = ∑n=1(−1)log (1-x ) = -∑n=1 log(1 − xⁿ)
-1 n

−x² x³
[log (1-x) = -x- - - ……..
2 4

−x² x³
Log (1+x) = +x+ + + ……..]
2 4

(xⁿ)² (xⁿ)³
= > - ∑∞
n { -(x ) – (
n
– - …….}
2 3

(xⁿ)² (xⁿ)³ (xⁿ)²


= > - ∑∞ n
n −{ (x ) + ( + + …….} = ∑∞ ∞
n=1 ∑m=1
2 3 m
1
= > ∑∞ ∞
m=1 m ∑n=1 xᵐⁿ

1
= > ∑∞
m=1 m {x +x +x +……}
m 2m 3m

1 xᵐ
= ∑∞
m=1 m 1=xᵐ ………….(1)

1−xᵐ
We have 1+x+x2+……..+xm-1 = 1+xᵐ

a(1−rⁿ)
(sum to n terms in geometric progression = )
1−r

∵ 0<x<1
1−xᵐ
mxm-1 < <m
1−x

[ for m=2, x=0.1


1−(0.1)²
We get 2(0.1)2-1 < <2
1−0.1

= 0.2 < 1+0.1 < 2

=0.2 <1.1 <2 ]


mxᵐ(1−x)
=> < 1-xm < m(1-x)
x

m(1−x) 1−xᵐ m(1−x)


< <
x xᵐ xᵐ

x xᵐ xᵐ
> 1−xᵐ > m(1−x)
m(1−x)

xᵐ xᵐ x
< m(1−x) < m(1−x)
m(1−x)

xᵐ xᵐ x
=> < m(1−xᵐ) < m(1−x)
m²(1−x)

xᵐ xᵐ
= > m(1−x) < m²(1−x)

1 xᵐ
∴ from (1), log F(x) = ∑∞
m==1 m{ (1−xᵐ)}

x
< ∑∞
m=1 m²(1−x)

x 1 x Π²
= 1−x ∑∞
m=1 m² = 1−x ( 6 )

π² 1−x
= 6t where t= x

π²
= > log F(x) < 6t …………(2)
For t > 0 , log (1+t) <t
1−x 1
But 1+t = 1+ =x
x

1
= > log (1+t) = log (x) < t

1
= > log (x) < t …………….(3)

1
Substitute (3) in (I) = > Log p(n) < log F(x) + nlog x

π²
= > log p(n) < 6t + nt ……. (by (2))

π²
Let f(t) = 6t + nt

π²
f´(t) = -6t + n

π²
f´(t) =0 = > -6t + n = 0

π²
= > 6t = n

π² π
= > t² = 6t = > t =
√6n

2π² π²
f´´(t) = + 0 = 3t³
2t³

π
= > f´´(t) │at t= >0
√6n

π
= > f(t) is minimum at t=
√6n

π
From (4) log p(n) < minimum ( + nt)
√6n

π² π nπ π√ n π n
= 6× + = + = 2𝑒 √6
√6n √6n √6 √6n

1 2
= k√n where k = 2𝑒√6 = 𝑒(3)0.5

∴ log p(n) < k√n


2
P(n) < e^ k√n where k= π(3)0.5

Jacobi’s Triple Product Identity:

Theorem (19): For complex x and y with │x│ < 1, z≠0, we have
∏∞ 2ⁿ 2 1 2 2 1 2 ∞
n=1(1 − x )(1 + x ⁿˉ 𝑧 )(1 + x ⁿˉ zˉ ) ∏m=−∞ xᵐ² z …… (1)
2m

2 1 2 2 1 2
Since │x│ < 1, the product ∏∞ 2 ∞ ∞
n=1(1 − x ⁿ) , ∏n=1(1 + x ⁿˉ z ) , ∏n=1(1 + x ⁿˉ zˉ ) and
∑∞m=−∞ xᵐ z
² 2m
are absolutely convergent.

Also, these products and the series converge uniformly on compact subsets c-{0} for each
fixed x with │x│ < 1

∴ each term in equation (1) is an analytic function of z and z≠0


2 1 2
For z≠0 , define a function f(z) = ∏∞
n=1(1 + x ⁿˉ z )(1+x²ⁿˉ¹zˉ²)

Claim: xz2F(xz) = F(z)


2 1 2
Consider F(xz) = ∏∞
n=1(1 + x ⁿˉ x² z )(1+x²ⁿˉ¹xˉ²zˉ²)

= ∏∞ 2 2
n=1(1 + x ⁿ⁺¹z )(1+x²ⁿˉ³zˉ²)

= ∏∞ 2 2 ∞
n=1(1 + x ⁿ⁺¹z ) ∏n=1(1+x² ⁿˉ³zˉ²)

= ∏∞ 2 2 ∞
m=2(1 + x ᵐˉ¹z ) ∏r=0(1+x²ʳˉ¹zˉ²)

1+xz² 1+xz² 1+xz² 1+xz²


Now, xz2F(xz) =1+xˉ¹zˉ² F(xz) [1+xˉ¹zˉ² = 1 = xz²1+x²z² = xz² ]
1+
xz²

1+xz²
= 1+xˉ¹zˉ² ∏∞ 2 2 ∞
m=2(1 + x ᵐˉ¹z ) ∏r=0(1+x²ʳˉ¹zˉ²)

= ∏∞ 2 2 ∞
n=1(1 + x ᵐˉ¹z²)( 1 + xz ) ∏r=0(1+x²ʳˉ¹zˉ²)

If m=1, (1 + x2 ᵐˉ1 z²) = (1+ xz2)

If r=0 , (1+x²ʳˉ¹zˉ²) = 1+x-1z-2

= ∏∞ 2 2 ∞
m=1(1 + x ᵐˉ¹z ) ∏r=1(1+x²ʳˉ¹zˉ²)

= ∏∞ 2 2
n=1(1 + x ⁿˉ¹z )(1+x²ⁿˉ¹zˉ²)

= F(z) = > xz2F(xz) = F(z) ……(2)

Let G(z) = F(z) . ∏∞ 2


n=1(1 − x ⁿ)

∴ xz2 G(xz) = xz2F(xz) ∏∞ 2


n=1(1 − x ⁿ)

= F(z) ∏∞ 2
n=1(1 − x ⁿ) …… by (2)

= > xz2G(xz) = G(z) ………….(3)

Consider G(-z) = F(-z) ∏∞ 2


n=1(1 − x ⁿ)

= F(z) ∏∞ 2
n=1(1 − x ⁿ) g(z)
[ ∵ F(-z) = ∏∞ 2 2
n=1(1 + x ⁿˉ¹(−z) )(1+x²ⁿˉ¹(-z)ˉ²)

= ∏∞ 2 2
n=1(1 + x ⁿˉ¹z )(1+x²ⁿˉ¹zˉ²) ]

∴ G(z) is even function and also analytic ∀ z≠0 by lagrangian expansion,

We get G(z) = ∑∞
m=−∞ am(z)
2m

Since xz2G(xz) = G(z)

= > xz2∑∞
m=−∞ am(xz)
2m
= ∑∞
m=−∞ am(z)
2m

= xz2∑∞
m=−∞ amx²ᵐz
2m
= ∑∞
m=−∞ amz
2m

Comparing the coefficient of z2m on both sides, we get

To get z2m+2 =z2m

2m+2 = 2m

M= m-1

Then 2(m-1) +2 = 2m

Coefficient of z2m is x.am-1.x2(m-1) = x2m-2+1.am-1 = x2m-1 am-1 in LHS

= > x2m-1 am-1 = am

= > am = am-1 x2m-1 = x2m-1 am-1

= x2m-1 x2m-3 am-1

= x2m-1 x2m-3 x2m-5 am-3

= x2m-1 x2m-3……..x3x1 a0

= x(2m-1)+(2m-3)+……3+1 a0

= > am = xm² a0 ∀ m≥0

1+3+……+2m-2 --- > arithmetic progression

Nth term tn = a+(n-1)d

(2m-1) = 1+(n-1)2
2m-1 = 1+ 2n-2

2m = 2n = > m=n
n
Sum to nth term = 2 (2a+(n-1)d)
m
= (2.1 +(m-1)2)
2

= m(1+1+m-1) = m2

∴ G(z) = ∑∞
m=−∞ amxᵐ z
² 2m

Since G(z) and a0 are dependent on x,

We have Gx(z) = ∑∞
m=−∞ amxᵐ z
² 2m
……(B)

T.P: a0 (x) -> 1 as x->0


πi
Let z = e 4

Gx(z) = F(z) ∏∞ 2n
n=1(1-x )

= ∏∞ 2n ∞
n=1(1-x ) ∏n=1(1+x z ) (1+x2n-1 z-2) …………..(A)
2n-1 2

πi πi πi
Gx(e 4 ) = ∏∞ 2n ∞
n=1(1-x ) ∏n=1(1+x
2n-1
(e 4 )2) (1+x2n-1 (e 4 )-2)
πi πi
= ∏∞ 2n ∞
n=1(1-x ) ∏n=1(1+x
2n-1
(e 4 )) (1+x2n-1 (e 4 )-1)

= ∏∞ 2n ∞
n=1(1-x ) ∏n=1(1+x
2n-1
i) (1+x2n-1 (-i))

= ∏∞ 2n ∞
n=1(1-x ) ∏n=1(1+ix
2n-1
) (1-ix2n-1)

= ∏∞ 2n ∞ 2
n=1(1-x ) ∏n=1(1-i (x
2n-1 2
))

= ∏∞ 2n ∞
n=1(1-x ) ∏n=1(1+ x
4n-2
)

∵ every even number is of the form 4n or 4n-2


4ⁿ
∏∞ 2n ∞
n=1(1-x ) = ∏n=1(1- x )(1+ x
4n-2
) …… (*)
πi
4ⁿ
∴ Gx(e 4 ) = ∏∞
n=1(1- x )(1- x
4n-2
) (1+ x4n-2)
4ⁿ
= ∏∞
n=1(1- x )(1- x
8n-4
)
8ⁿ
= ∏∞
n=1(1- x )(1- x
8n-4
)(1-x8n-4) ….. (4) (by (*))

Now from (A) Gx4 (i) = ∏∞ 4 2n ∞ 4 2n-1 2


n=1(1- (x ) )∏n=1(1- (x ) i )( 1- (x4)2n-1i-2)

= ∏∞ 8n
n=1(1-x )(1+x
8n-4
(-1))( 1+x8n-4(-1))
= ∏∞ 8n
n=1(1-x )(1-x
8n-4
)( 1-x8n-4)……………(5)

From (4) and (5) (∵ (4)=(5))


πi
= > Gx(e 4 )= Gx4 (i)………………(6)

From (B)
πi πi
Gx(e 4 )= ∑∞
m=−∞ am (x)xᵐ (e 4 )
² 2m

= ∑∞
m=−∞ am (x)xᵐ (i) ………(***)
² m

πi
{ (e 4 )m = (i)m

i3 = -I , i-3 = i = -i3

i3= -i3 … odd}


πi
{ Gx(e 4 )= Gx4 (i) then from (**)

Gx4 (i) = ∑∞
m=−∞ a0 (x)xᵐ (i)
² m

= ∑∞
n=−∞ a0 (x) x
(2n)² 2n
)i }
πi
Gx(e 4 )
∴ = ∑∞
m=−∞ 𝑥ᵐ (i)
² m
a0(x)

= ∑∞
n=−∞(x
(2n)² 2n
)i

If m is odd, im= -im

∴ odd terms get cancelled and remaining is even terms i.e, m=2n

= ∑∞ 4n² 2n
n=−∞(x )i

Gx(i)
= ∑∞ 4 n² 2 n
n=−∞(x ) (i ) = a0(x)

πi
Gx(e 4 ) Gx4(i)
= a0(x)4
a0(x)

∴ a0(x) = a0(x4) (∵ from (6))

Repeating this process for x4², x4³, ……

a0(x) = a0(x4ᴷ) for k=1,2,3….

Since │x│< 1, x4ᴷ  0 as k - > ∞

∴ a0(x) - > 1 as x -> 0


∴ G(z) = ∑∞ m² 2m
m=−∞ x z

m²+m
Theorem 20: if │x│< 1 then ∏∞ ∞
n=1(1 − xⁿ)³ = ∑m=−∞(−1) ᵐ x m=
2
m²+m
∑∞
m=−∞(−1) ᵐ (2m + 1)x 2

By jacobi’s triple product identity

∏∞ 2ⁿ 2 1 2 2 1 2 ∞
n=1(1 − x )(1 + x ⁿˉ 𝑧 )(1 + x ⁿˉ zˉ ) ∏m=−∞ xᵐ² z
2m

Let z2=-xz
2 1 2 1
Then ∏∞ 2ⁿ ∞
n=1(1 − x )(1 + x ⁿˉ (−𝑥𝑧))(1 + x ⁿˉ (−𝑥𝑧)ˉ¹) ∏m=−∞ xᵐ² (-xz)
m

= > ∏∞ 2ⁿ 2 1 ∞
n=1(1 − x )(1 − x ⁿ𝑧)(1 − 𝑥²ⁿˉ²zˉ ) = ∑m=−∞(-1) x
m m²+m m
z

= ∑∞ z + ∑−∞
m m²+m m
m=0(-1) x z ….. (1)
m m²+m m
m=−1(-1) x

Consider ∑−∞
m=−1(-1) x z = ∑∞
m m²+m m -m m²-m -m
m=1(-1) x z [(-m)2 = m2 ]

= ∑∞
m=0(-1)
-(m+1) (m+1)² - (m+1) -(m+1)
x z

= - ∑∞ -m m²+m -(m+1)
m=0(-1) x z

= - ∑∞ m m²+m -(m+1)
m=0(-1) x z

(1) = > ∏∞ 2n 2n
n=1(1-x )(1-x z)(1-x
2n-2 -1
z )

= ∑∞
m=0(-1) x z - ∑∞
m m²+m m m m²+m -(m+1)
m=0(-1) x z

= ∑∞ m m²+m
m=0(-1) x {zm - z-(m+1)} …….. (2)

We have ∏∞
n=1( 1-x z ) = (1-z-1) ∏∞
2n-2 -1 2 1
n=1(1 − x ⁿzˉ )

And zm-z-m-1 = (1-z-1)(1+z-1+z-1+…+z-2m)zm


zˉᵐˉ¹
zm(1- ) = zm(1-z-2m-1) = zm(1-(z-1)2m+1)
zᵐ

= zm(1-z-1)(1+z-1+(-z)2+….+(z-1)2m )

[ (ak –bk ) = (a-b) (ak-1+ak-2.b + ….+bk-1) ]

(2) = > ∏∞ 2n 2n -1 ∞ 2n -1
n=1(1-x )(1-x z)(1-z ) ∏n=1( 1-x z )
= ∑∞ m m²+m
m=0(−1) x (1-z-1)(1+z-1+z-2+…+z-2m) zm

Let x=x½ and z= 1


= > ∏∞ n n n
n=1(1-x )( 1-x )( 1-x )

m²+m
= ∑∞ m
m=0(-1) x (2m+1)
2

m²+m
= > ∏∞ n 3 ∞ m
n=1(1-x ) = ∑m=0(-1) x (2m+1)
2

Вам также может понравиться