Non-zero coefficients

Consider the polynomial

f ( x ) = ( x 20 + x 13 + 1 ) 33 . f(x) = (x^{20} + x^{13} + 1)^{33}.

When fully expanded, how many terms have a non-zero coefficient?

Details and assumptions

As an explicit example, since ( x 2 + 1 ) 3 = x 6 + 3 x 4 + 3 x 2 + 1 (x^2 + 1)^3 = x^6 + 3x^4 + 3x^2 + 1 , there are 4 terms with a non-zero coefficient.

You may use the fact that ( 33 2 ) = 528 , ( 34 2 ) = 561 { 33 \choose 2 } = 528, {34 \choose 2 } = 561 and ( 35 2 ) = 595 { 35 \choose 2 } = 595 .


The answer is 490.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

5 solutions

Matt McNabb
Sep 3, 2013

Terms in the expansion of ( f ( x ) (f(x) look like: ( x 20 ) i ( x 13 ) j 1 k (x^{20})^i \cdot (x^{13})^j \cdot 1^k where i + j + k = 33 i+j+k = 33 .

So we need to find all the possible distinct values of 20 i + 13 j 20i + 13j subject to the constraints 0 i + j 33 0 \le i+j \le 33 and 0 i , j 0 \le i,j .

Now, if we restrict j j to the range [ 0 , 19 ] [0, 19] then we still include all possible values. This is because if there is some N = 20 i + 13 j N = 20i + 13j with j 20 j \ge 20 , then the value N = 20 ( i + 13 ) + 13 ( j 20 ) N = 20(i+13) + 13(j-20) is in the restricted range already, and i + 13 + j 20 = i + j 7 i+13+j-20 = i+j-7 satisfies the constraint 0 i + j 33 0 \le i+j \le 33 .

Since 13 13 and 20 20 are coprime, we do not have any duplicate values of N in this range either.

Now, if j = 0 j=0 then i [ 0 , 33 ] i \in [0, 33] , i.e. 34 possibilities.

And if j = 1 j=1 then i [ 0 , 32 ] i \in [0,32] , i.e. 33 possibilities.

Carry on to j = 19 j=19 implying i [ 0 , 14 ] i \in [0,14] which is 15 possibilities.

So the sum we want is 34 + 33 + . . . . + 15 = 490 34 + 33 + .... + 15 = \boxed{490}

Jeffery Yu
Sep 1, 2013

The first thing to note is that all nonzero terms upon expansion (even before combining like terms, if you were to distribute everything out) will have positive coefficients, so we do not have to worry about two like terms simplifying to 0. Essentially, the question is asking for the number of distinct exponents that can be formed upon expansion (we will ignore terms with a coefficient of 0).

Thus, we will consider only the exponents and not the coefficients. Upon expansion (you may wish to imagine using the Multinomial Theorem if you cannot easily understand the following), all exponents will have the form 20 a + 13 b + 0 c 20a + 13b + 0c , where a + b + c 33 a + b + c \le 33 and a , b , c W a, b, c \in \mathbb{W} . We have 0 c 0c to account for the constant in the factored form, but for our purposes, it is redundant, so we reduce the problem to finding the number of distinct values that can be formed from the expression 20 a + 13 b 20a + 13b where a + b 33 a + b \le 33 .

To do this, we will overcount and then subtract the overcounting. If we just consider the number of nonnegative integer pairs ( a , b ) (a, b) while a + b 33 a + b \le 33 , we get ( 35 2 ) = 595 \binom{35}{2} = 595 . If this is not clear, consider each case (eg: a + b = 0 a + b = 0 has 1 1 solution, up to a + b = 33 a + b = 33 has 34 34 solutions) and add them up.

Now we add the restriction of the expression 20 a + 13 b 20a + 13b being distinct for each ( a , b ) (a, b) . Since 20 20 and 13 13 are relatively prime, the first case of overcounting deals with the fact that 20 13 = 13 20 20 \cdot 13 = 13 \cdot 20 , which occurs from the pairs ( 13 , 0 ) (13, 0) and ( 0 , 20 ) (0, 20) . In fact, all cases of overcounting will occur from the pairs ( 13 n + m , k ) (13n+m, k) and ( m , 20 n + k ) (m, 20n+k) , where k , m , n N k, m,n \in \mathbb{N} . However, it is clear that for any n > 1 n > 1 , we have a + b > 33 a + b > 33 , so we only need to consider the case where n = 1 n = 1 . We see that we must have the condition m + k 13 m + k \le 13 , and there are ( 15 2 ) = 105 \binom{15}{2} = 105 ways for this to be satisfied.

Therefore, there will be 595 105 = 490 595 - 105 = \boxed{490} terms with a nonzero coefficient.

I think it should be a + b + c = 33 a + b + c = 33 , then because c 0 c \geq 0 we get a + b 33 a + b \leq 33 ???

Kim Phú Ngô - 7 years, 9 months ago
Nhat Le
Sep 4, 2013

We can write f ( x ) f(x) as the product of 33 33 brackets: ( x 20 + x 13 + 1 ) ( x 20 + x 13 + 1 ) ( x 20 + x 13 + 1 ) . . . . ( x 20 + x 13 + 1 ) (x^{20}+x^{13}+1)(x^{20}+x^{13}+1)(x^{20}+x^{13}+1)....(x^{20}+x^{13}+1)

A term in the expansion of f ( x ) f(x) is obtained by choosing one term from each of the 33 33 brackets and multiplying these terms together. Let's say the term x 20 x^{20} is chosen a a times and the term x 13 x^{13} is chosen b b times. The power of the term obtained will be 20 a + 13 b 20a+13b .

The problem thus reduces to the following. Given two non-negative integers a a and b b such that a + b 33 a+b\leq 33 , how many distinct values can 20 a + 13 b 20a+13b take?

We will solve this by first counting the total number of pairs of a a and b b , then subtracting the number of times two pairs give repeated values of 20 a + 13 b 20a +13b .

We start by counting the number of pairs of non-negative integers a a and b b such that a + b 33 a+b \leq33 . When a = 0 a=0 , b b can range from 0 0 to 33 33 , giving 34 34 possible pairs. When a = 1 a=1 , b b can range from 0 0 to 32 32 , giving 33 33 possible pairs.When a = 2 a=2 , b b can range from 0 0 to 31 31 , giving 32 32 possible pairs. Continuing in this pattern, we will reach the case when a = 33 a=33 , b b can only be 0 0 , giving 1 possible pair. Thus the total number of pairs is 34 + 33 + 32 + . . . + 1 = ( 35 2 ) = 595 34+33+32+...+1= \binom{35}{2}=595

Now we count the cases where 20 a + 13 b 20a+13b are not distinct. Consider the case when 20 a + 13 b = 20 a + 13 b 20a+13b=20a'+13b' , with a > a a' \gt a (and therefore b < b b' \lt b . This happens if and only if a = a + 13 t a' = a+13t and b = b 20 t b' =b-20t for some positive integer t t . We have 0 b = b 20 t 33 20 t 0 \leq b'=b-20t \leq 33-20t so 33 20 t 0 33-20t \geq 0 , giving t 1 t \leq 1 . Since t t is positive, t t can only be 1 1 .

Thus we see that a = a + 13 a'=a+13 and b = b 20 b'=b-20 . We then proceed to count the number of duplicates. For convenience, we denote ( a , b ) ( a , b ) (a,b) \equiv (a',b') to show that this is a pair of duplicates.

When a = 0 a=0 we have the following duplicates: ( 0 , 20 ) ( 13 , 0 ) , ( 0 , 21 ) ( 13 , 1 ) , ( 0 , 22 ) ( 13 , 2 ) , . . . ( 0 , 33 ) ( 13 , 13 ) (0,20) \equiv (13,0), (0,21) \equiv (13,1), (0,22) \equiv (13,2),...(0,33) \equiv (13,13) Altogether there are 14 14 pairs

Continuing in the same way, we see that when a = 1 a=1 there are 13 13 pairs, when a = 2 a=2 there are 12 12 pairs,... until when a = 13 a=13 there is 1 1 pair.

Thus the total number of duplicates is 14 + 13 + 12 + . . . + 1 = ( 15 2 ) = 105 14+13+12+...+1 = \binom{15}{2} = 105

The answer we need is thus equal to 595 105 = 490 595-105=\fbox{490}

Brilliant solution...

Pravesh Trivedi - 7 years, 9 months ago

I thought it should be 504 because you should let one of each non-distinct solution to be a part of the solution. 490 means that non of the distinct solution is a part of the solution. 490 + 14 = 504

Billy Sugiarto - 7 years, 9 months ago

Log in to reply

595 595 solutions include the duplicates counted twice, so we need to subtract once to get the number of distinct values. You don't have to add back.

Nhat Le - 7 years, 9 months ago
Daniel Chiu
Sep 1, 2013

We can find the total number of terms since we can each of the 33 "powers" into a "20" box, a "13" box, or a "0" box. Basically, we find the number of pairs of integers ( a , b ) (a,b) such that a + b 33 a+b\le 33 . This is ( 35 2 ) = 595 \dbinom{35}{2}=595 Now, we must account for overcounting. Since gcd ( 13 , 20 ) = 1 \gcd(13,20)=1 , the first number that can be expressed as a multiple of 13 plus a multiple of 20 in more than one way is 13 20 13\cdot 20 . 20 13 = 13 20 20\cdot 13=13\cdot 20 A second collision would occur if we have 13 m 20 n 13m\cdot 20n , but since we only have 13 more powers to distribute on the left side, we cannot have m , n > 1 m,n>1 .

Now, we have 13 more powers to distribute among both sides. The number of ways to do this, similarly to how we distributed the 33 powers initially, is ( 15 2 ) = 105 \dbinom{15}{2}=105 The answer is 595 105 = 490 595-105=\boxed{490} .

Arjen Vreugdenhil
Apr 19, 2016

The exponents of the non-zero terms are precisely the set { 20 a + 13 b + 0 c a + b + c = 33 } = { 20 a + 13 b a + b 33 } . \{20a + 13b + 0c | a+b+c = 33\} = \{20a + 13b | a + b \leq 33\}. There are 34 35 / 2 = 595 34\cdot 35/2 = 595 such pairs ( a , b ) (a,b) ; however, the values of 20 a + 13 b 20a + 13b are not all unique.

Any pair ( a , b ) (a, b) with b 20 b \geq 20 generates the same exponent as the pair ( a + 13 , b 20 ) (a+13, b-20) . Therefore we limit ourselves to the cases with b < 20 b < 20 . SInce 13 and 20 are coprime, all values of 20 a + 13 b 20a + 13b are now guaranteed to be unique.

Count therefore all pairs ( a , b ) (a,b) with 0 a 0 \leq a , 0 b 19 0 \leq b \leq 19 , and a + b 33 a + b \leq 33 ; that gives b = 0 1 9 ( 34 b ) = 20 34 b = 0 1 9 b = 20 34 34 35 2 = 490 . \sum_{b = 0}^19 (34-b) = 20\cdot 34 - \sum_{b=0}^19 b = 20\cdot 34 - \frac{34\cdot 35}2 = \boxed{490}.

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...