MathLinks Forum LaTeX Help AoPS Classes Books Classroom MathLinks Contest Math Resources
The time now is Tue Feb 09, 2010 7:32 pm
All times are UTC + 2
View posts since last visit
View unanswered posts
View previous topicView next topic
 
2 Posts • Page 1 of 1
Author Message
Peter
Birch & Swinnerton Dyer
Birch & Swinnerton Dyer


Offline
Joined: 05 May 2004
Posts: 5214
Location: Ghent
Belgium

To rate posts you must be logged in
#1
E 16
MM, Problem 1392, George Andrews

Prove that for any prime p in the interval \left]n, \frac{4n}{3}\right], p divides \sum^{n}_{j=0}{{n}\choose{j}}^{4}.

PostPosted: Fri May 25, 2007 2:24 am
darij grinberg
Birch & Swinnerton Dyer
Birch & Swinnerton Dyer


Offline
Joined: 10 Feb 2004
Posts: 5806
Location: Karlsruhe / Munich

To rate posts you must be logged in
#2
Re: E 16
MM, Problem 1392, George Andrews

EDIT: An edited and extended (Theorem 1 generalized even further, Theorem 5 extended and proved anew) version of the below solution can be found here:
http://projectpen.files.wordpress.com/2009/04/pen-e16.pdf
or here:
http://www.cip.ifi.lmu.de/~grinberg/pene16.pdf

Peter wrote:
Prove that for any prime p in the interval \left]n, \frac {4n}{3}\right], p divides
\sum^{n}_{j = 0}{{n}\choose{j}}^{4}.


Nice and instructive problem. Let's generalize:

Theorem 1. If n and k are positive integers and p is a prime such that \frac {2k - 1}{2k}\left(p - 1\right) < n < p, then p\mid\sum_{j = 0}^{n}\binom{n}{j}^{2k}.

Before we prove this, we first show some basic facts about binomial coefficients and remainders modulo primes.

Definition. The binomial coefficient \binom{x}{u} is defined for all reals x and for all integers u as follows: \binom{x}{u} = \frac {x\cdot\left(x - 1\right)\cdot ...\cdot\left(x - u + 1\right)}{u!} if u is a positive integer, \binom{x}{0} = 1, and \binom{x}{u} = 0 if u is a negative integer.

Theorem 2, the upper negation identity. If n is a real, and r is an integer, then \binom{ - n}{r} = \left( - 1\right)^{r}\binom{n + r - 1}{r}.

Proof of Theorem 2. We will only consider the case of r being positive (the other cases are trivial). Then, using the definition of binomial coefficients,

\binom{ - n}{r} = \frac {\left( - n\right)\cdot\left( - n - 1\right)\cdot ...\cdot\left( - n - r + 1\right)}{r!} = \left( - 1...
= \left( - 1\right)^{r}\cdot\frac {\left(n + r - 1\right)\cdot ...\cdot\left(n + 1\right)\cdot n}{r!} = \left( - 1\right)^{r}....

This proves Theorem 2.

Theorem 3. If p is a prime, if u and v are two integers such that u\equiv v\mod p, and if k is an integer such that k < p, then \binom{u}{k}\equiv\binom{v}{k}\mod p.

Proof of Theorem 3. If k\leq 0, then \binom{u}{k} = \binom{v}{k}, so that Theorem 3 is trivial. Thus, it remains to consider the case k > 0 only. In this case, k! is coprime with p (since k! = 1\cdot 2\cdot ...\cdot k, and all numbers 1, 2, ..., k are coprime with p, since p is a prime and k < p).

Now, u\equiv v\mod p yields

k!\cdot\binom{u}{k} = k!\cdot\frac {u\cdot\left(u - 1\right)\cdot ...\cdot\left(u - k + 1\right)}{k!} = u\cdot\left(u - 1\rig...
\equiv v\cdot\left(v - 1\right)\cdot ...\cdot\left(v - k + 1\right) = k!\cdot\frac {v\cdot\left(v - 1\right)\cdot ...\cdot\le...
= k!\cdot\binom{v}{k}\mod p.

Since k! is coprime with p, we can divide this congruence by k!, and thus we get \binom{u}{k}\equiv\binom{v}{k}\mod p. Hence, Theorem 3 is proven.

Finally, a basic property of binomial coefficients:

Theorem 4. For every nonnegative integer n and any integer k, we have \binom{n}{k} = \binom{n}{n - k}.

This is known, but it is important not to forget the condition that n is nonnegative (it is necessary!).

Now to something completely different:

Theorem 5. If p is a prime, and f is a polynomial of degree < p - 1 whose coefficients are all integers, then \sum_{j = 0}^{p - 1}f\left(j\right)\equiv 0\mod p.

Proof of Theorem 5. Since f is a polynomial of degree < p - 1 whose coefficients are all integers, it can be written in the form f\left(X\right) = \sum_{i = 0}^{p - 2}f_{i}X^{i} with f_{i} being an integer (not necessarily nonzero) for every i\in\left\{0;\;1;\;...;\;p - 3;\;p - 2\right\}.

Hence,

\sum_{j = 0}^{p - 1}f\left(j\right) = \sum_{j = 0}^{p - 1}\sum_{i = 0}^{p - 2}f_{i}j^{i} = \sum_{i = 0}^{p - 2}f_{i}\cdot\sum....

But for every i\in\left\{0;\;1;\;...;\;p - 3;\;p - 2\right\}, we have \sum_{j = 0}^{p - 1}j^{i}\equiv 0\mod p (in fact, for i = 0, this is clear since \sum_{j = 0}^{p - 1}j^{0} = \sum_{j = 0}^{p - 1}1 = p, and for i\geq 1, this is equivalent to \sum_{j = 1}^{p}j^{i}\equiv 0\mod p (since \sum_{j = 0}^{p - 1}j^{i}\equiv \sum_{j = 1}^{p}j^{i}\mod p, because 0^{i}\equiv p^{i}\mod p), what follows from http://www.problem-solving.be/pen/viewtopic.php?t=676 part a) because p - 1\nmid i). Thus,

\sum_{j = 0}^{p - 1}f\left(j\right) = \sum_{i = 0}^{p - 2}f_{i}\cdot\sum_{j = 0}^{p - 1}j^{i}\equiv\sum_{i = 0}^{p - 2}f_{i}\...,

and Theorem 5 is proven.

Proof of Theorem 1. We have p - n - 1\geq 0, since n < p yields n + 1\leq p.

Also, \frac {2k - 1}{2k}\left(p - 1\right) < n yields \left(2k - 1\right)\left(p - 1\right) < 2kn, so that 2k\left(p - 1\right) - \left(p - 1\right) < 2kn, thus 2k\left(p - 1\right) - 2kn < p - 1, or, equivalently, 2k\left(p - n - 1\right) < p - 1.

We have to prove that p\mid\sum_{j = 0}^{n}\binom{n}{j}^{2k}. This rewrites as \sum_{j = 0}^{n}\binom{n}{j}^{2k}\equiv 0\mod p. This is equivalent to \sum_{j = 0}^{p - 1}\binom{n}{j}^{2k}\equiv 0\mod p (in fact, since n < p, the two sums \sum_{j = 0}^{p - 1}\binom{n}{j}^{2k} and \sum_{j = 0}^{n}\binom{n}{j}^{2k} differ only in some terms \binom{n}{j}^{2k} with indices j in the interval \left]n;\;p - 1\right], and these terms are zero, so we have \sum_{j = 0}^{p - 1}\binom{n}{j}^{2k} = \sum_{j = 0}^{n}\binom{n}{j}^{2k}).

For every integer j with 0\leq j < p, we have

\binom{n}{j} = \binom{ - \left( - n\right)}{j} = \left( - 1\right)^{j}\binom{\left( - n\right) + j - 1}{j} (after Theorem 2)
\equiv\left( - 1\right)^{j}\binom{p - n + j - 1}{j} (by Theorem 3, since \left( - n\right) + j - 1\equiv p - n + j - 1\mod p and j < p)
= \left( - 1\right)^{j}\binom{p - n + j - 1}{\left(p - n + j - 1\right) - j} (by Theorem 4, since p - n + j - 1 is nonnegative, since p - n - 1\geq 0 and j\geq 0)
= \left( - 1\right)^{j}\binom{p - n + j - 1}{p - n - 1}\mod p,

so that

\binom{n}{j}^{2k}\equiv\left(\left( - 1\right)^{j}\binom{p - n + j - 1}{p - n - 1}\right)^{2k} = \binom{p - n + j - 1}{p - n ... (since 2k is even).

Therefore,

(1) \left(p - n - 1\right)!^{2k}\cdot\sum_{j = 0}^{p - 1}\binom{n}{j}^{2k}\equiv\left(p - n - 1\right)!^{2k}\cdot\sum_{j = 0}^{p ...
= \sum_{j = 0}^{p - 1}\left(\left(p - n - 1\right)!\cdot\binom{p - n + j - 1}{p - n - 1}\right)^{2k}
= \sum_{j = 0}^{p - 1}\left(\left(p - n - 1\right)!\cdot\frac {\prod_{u = 0}^{\left(p - n - 1\right) - 1}\left(\left(p - n + ...
= \sum_{j = 0}^{p - 1}\left(\prod_{u = 0}^{\left(p - n - 1\right) - 1}\left(\left(p - n + j - 1\right) - u\right)\right)^{2k}....

Now,

\prod_{u = 0}^{\left(p - n - 1\right) - 1}\left(\left(p - n + X - 1\right) - u\right)

is a polynomial of the variable X whose degree is p - n - 1. Thus,

\left(\prod_{u = 0}^{\left(p - n - 1\right) - 1}\left(\left(p - n + X - 1\right) - u\right)\right)^{2k}

is a polynomial of the variable X whose degree is 2k\left(p - n - 1\right). Since 2k\left(p - n - 1\right) < p - 1, it is therefore a polynomial of degree < p - 1. Since the coefficients of this polynomial are all integers, Theorem 5 thus yields

\sum_{j = 0}^{p - 1}\left(\prod_{u = 0}^{\left(p - n - 1\right) - 1}\left(\left(p - n + j - 1\right) - u\right)\right)^{2k}\e....

Using (1), this becomes

\left(p - n - 1\right)!^{2k}\cdot\sum_{j = 0}^{p - 1}\binom{n}{j}^{2k}\equiv 0\mod p.

Now, \left(p - n - 1\right)! is coprime with p (since \left(p - n - 1\right)! = 1\cdot 2\cdot ...\cdot\left(p - n - 1\right), and all numbers 1, 2, ..., p - n - 1 are coprime with p because p is a prime and p - n - 1 < p). Hence, we can divide this congruence by \left(p - n - 1\right)!^{2k}, and obtain \sum_{j = 0}^{p - 1}\binom{n}{j}^{2k}\equiv 0\mod p. As we said, this proves Theorem 1.

Darij

PostPosted: Fri May 25, 2007 2:24 am
Display posts from previous:   Sort by:   
2 Posts • Page 1 of 1
View previous topicView next topic
Jump to:  

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum
You cannot post calendar events in this forum

Created and Maintained by Valentin Vornicu - (c) AoPS Inc. 2004-2008