Compute $a_{1996}$ for $\prod_{k=1}^{1996} (1+kx^{3^k})$

  • Context: MHB 
  • Thread starter Thread starter anemone
  • Start date Start date
Click For Summary

Discussion Overview

The discussion revolves around computing the coefficient \( a_{1996} \) in the expansion of the product \( \prod_{k=1}^{1996} (1+kx^{3^k}) \). The focus is on the mathematical reasoning and potential errors in the solution process.

Discussion Character

  • Mathematical reasoning, Technical explanation, Debate/contested

Main Points Raised

  • One participant expresses the goal of computing \( a_{1996} \) from the product expansion.
  • Another participant notes a possible off-by-one error in their solution but asserts that the method is valid.
  • A later post presents an amended solution, indicating a revision of the initial approach.
  • Further amendments are mentioned, but details are not provided in the thread.
  • One participant acknowledges the contributions of another, highlighting a correction made regarding a subtraction error.

Areas of Agreement / Disagreement

There is no clear consensus on the correctness of the solutions presented, as some participants acknowledge potential errors while others provide amended solutions. The discussion remains unresolved regarding the final value of \( a_{1996} \).

Contextual Notes

Participants have noted possible errors in their calculations, including an off-by-one error and a subtraction error, but the specific implications of these errors on the final result are not fully explored.

anemone
Gold Member
MHB
POTW Director
Messages
3,851
Reaction score
115
Let $\displaystyle \prod_{k=1}^{1996} (1+kx^{3^k})=1+a_1x^{n_1}+a_2x^{n_2}+\cdots+a_mx^{n_m}$, where $a_1,\,a_2,\,\cdots$ are nonzero and $n_1<n_2<\cdots<n_m$.

Compute $a_{1996}$.
 
Mathematics news on Phys.org
Note: solution not verified, possible off-by-one error, but it definitely works.

We express $a_n$ as a recurrence by induction on $k$. First consider the base case $k = 1$, which gives us $a_0 = a_1 = 1$. Suppose that for some $k - 1 \in \mathbb{N}$, we have nonzero polynomial coefficients $a_0, a_1, \cdots, a_{n - 1}$ for some $n$, then it follows that:
$$(1 + k x^{3^k}) \sum_{i = 0}^{n - 1} a_i x^{n_i} = \sum_{i = 0}^{n - 1} a_i x^{n_i} + k \sum_{i = 0}^{n - 1} a_i x^{3^k + n_i}$$
We observe that $3^k$ is larger than all exponents $n_i$ which are at most $3^k - 1$, being a sum of powers of $3$ less than $k$, and so the coefficients $a_0$ to $a_{n - 1}$ remain unchanged by the multiplication by $1 + k x^{3^k}$. Furthermore, as a result the polynomial for $k$ must then have $2n$ nonzero terms (twice as many). These two facts together allows us to derive $k$ from $n$ as $k = \lceil \log_2(n + 1) \rceil$ and hence express $a_n$ as a simple recurrence:
$$a_n = k a_{n - 2^{k - 1}} ~ ~ ~ \text{where} ~ k = \lceil \log_2(n + 1) \rceil, a_0 = a_1 = 1$$
So that:
$$a_{1996} = 11 a_{1996 - 1024} = 11 a_{972}$$
$$a_{972} = 10 a_{972 - 512} = 10 a_{460}$$
$$a_{460} = 9 a_{460 - 256} = 9 a_{204}$$
$$a_{204} = 8 a_{204 - 128} = 8 a_{76}$$
$$a_{76} = 7 a_{76 - 64} = 7 a_8$$
$$a_{8} = 4 a_{8 - 8} = 4 a_0$$
And so we conclude that:
$$a_{1996} = 11 \times 10 \times 9 \times 8 \times 7 \times 4 \times a_0 = 221760$$
 
Amended solution:

We express $a_n$ as a recurrence by induction on $k$. First consider the base case $k = 1$, which gives us $a_0 = a_1 = 1$. Suppose that for some $k - 1 \in \mathbb{N}$, we have nonzero polynomial coefficients $a_0, a_1, \cdots, a_{n - 1}$ for some $n$, then it follows that:
$$(1 + k x^{3^k}) \sum_{i = 0}^{n - 1} a_i x^{n_i} = \sum_{i = 0}^{n - 1} a_i x^{n_i} + k \sum_{i = 0}^{n - 1} a_i x^{3^k + n_i}$$
We observe that $3^k$ is larger than all exponents $n_i$ which are at most $3^k - 1$, being a sum of powers of $3$ less than $k$, and so the coefficients $a_0$ to $a_{n - 1}$ remain unchanged by the multiplication by $1 + k x^{3^k}$. Furthermore, as a result the polynomial for $k$ must then have $2n$ nonzero terms (twice as many). These two facts together allows us to derive $k$ from $n$ as $k = \lceil \log_2(n + 1) \rceil$ and hence express $a_n$ as a simple recurrence:
$$a_n = k a_{n - 2^{k - 1}} ~ ~ ~ \text{where} ~ k = \lceil \log_2(n + 1) \rceil, a_0 = a_1 = 1$$
So that:
$$a_{1996} = 11 a_{1996 - 1024} = 11 a_{972}$$
$$a_{972} = 10 a_{972 - 512} = 10 a_{460}$$
$$a_{460} = 9 a_{460 - 256} = 9 a_{204}$$
$$a_{204} = 8 a_{204 - 128} = 8 a_{76}$$
$$a_{76} = 7 a_{76 - 64} = 7 a_{12}$$
$$a_{12} = 4 a_{12 - 8} = 4 a_4$$
$$a_{4} = 3 a_{4 - 4} = 3 a_0$$
And so we conclude that:
$$a_{1996} = 11 \times 10 \times 9 \times 8 \times 7 \times 4 \times 3 \times a_0 = 665280$$
 
Bacterius said:
Amended solution:

We express $a_n$ as a recurrence by induction on $k$. First consider the base case $k = 1$, which gives us $a_0 = a_1 = 1$. Suppose that for some $k - 1 \in \mathbb{N}$, we have nonzero polynomial coefficients $a_0, a_1, \cdots, a_{n - 1}$ for some $n$, then it follows that:
$$(1 + k x^{3^k}) \sum_{i = 0}^{n - 1} a_i x^{n_i} = \sum_{i = 0}^{n - 1} a_i x^{n_i} + k \sum_{i = 0}^{n - 1} a_i x^{3^k + n_i}$$
We observe that $3^k$ is larger than all exponents $n_i$ which are at most $3^k - 1$, being a sum of powers of $3$ less than $k$, and so the coefficients $a_0$ to $a_{n - 1}$ remain unchanged by the multiplication by $1 + k x^{3^k}$. Furthermore, as a result the polynomial for $k$ must then have $2n$ nonzero terms (twice as many). These two facts together allows us to derive $k$ from $n$ as $k = \lceil \log_2(n + 1) \rceil$ and hence express $a_n$ as a simple recurrence:
$$a_n = k a_{n - 2^{k - 1}} ~ ~ ~ \text{where} ~ k = \lceil \log_2(n + 1) \rceil, a_0 = a_1 = 1$$
So that:
$$a_{1996} = 11 a_{1996 - 1024} = 11 a_{972}$$
$$a_{972} = 10 a_{972 - 512} = 10 a_{460}$$
$$a_{460} = 9 a_{460 - 256} = 9 a_{204}$$
$$a_{204} = 8 a_{204 - 128} = 8 a_{76}$$
$$a_{76} = 7 a_{76 - 64} = 7 a_{12}$$
$$a_{12} = 4 a_{12 - 8} = 4 a_4$$
$$a_{4} = 3 a_{4 - 4} = 3 a_0$$
And so we conclude that:
$$a_{1996} = 11 \times 10 \times 9 \times 8 \times 7 \times 4 \times 3 \times a_0 = 665280$$

Thanks for participating, Bacterius! Yes, your amended solution is correct and the subtraction error was spotted by Opalg and so I want to thank Opalg because he has helped me numerous times in the past even if he was busy at the time.
 

Similar threads

  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 2 ·
Replies
2
Views
2K
  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 2 ·
Replies
2
Views
1K
  • · Replies 2 ·
Replies
2
Views
2K
Replies
5
Views
2K
  • · Replies 3 ·
Replies
3
Views
3K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 2 ·
Replies
2
Views
5K