I'm having some problems trying to resolve one exercise from The art and craft of problem solving by Paul Zeitz. What this problem asks you is to prove that $F(x)$ is equal to 1 for all $x$, where $$F(x) = (1-x)(1+x)(1-x^3)(1+x^2)(1-x^5)(1+x^3)(1-x^7)(1+x^4)\cdots$$ This would demonstrate that the number of unequal partitions and odd partitions of an integer is the same since the generating function $U(x)$ for the number of unequal partitions of an integer $n$ is $$U(x) = \sum_{n=0}^\infty u_nx^n = (1+x)(1+x^2)(1+x^3)\cdots$$ and the generating function $V(n)$ for the number of odd partitions of an integer $n$ is $$V(x) = \sum_{n=0}^\infty v_nx^n = (1+x+x^2+x^3+\cdots)(1+x^3+x^6+x^9+\cdots)(1+x^5+x^{10}+x^{15}+\cdots)\cdots = \frac{1}{(1-x)(1-x^3)(1-x^5)\cdots}$$ and so $F(x)=\frac{U(x)}{V(x)}$
The way the exercise suggests to carry out the demonstration is to prove that $F(x)$ is invariant under the substitution $x\mapsto x^2$. This is easy to chech $$F(x^2) = \prod_{k=1}^\infty (1-(x^2)^{2k-1})(1+(x^2)^k)=\prod_{k=1}^\infty (1-x^{2k-1})(1+x^{2k-1})(1+x^{2k}) = F(x)$$
Now the idea is to keep iterating the substitution and I guess that the solution is that for values of $x$ such that $|x| < 1$ all the $x$ terms go to $0$, but I see two problems with this:
- This wouldn't work for values of $x$ greater than $1$ or less than $-1$. We don't care about these values because the function does not converge when evaluated with them? Why?
- After iterating the substitution $n$ times we get $$F(x) = F(x^{2^n}) = \prod_{k=1}^\infty (1-x^{2^n(2k-1)})(1+x^{2^nk})$$ and although the $x$ terms go to $0$ for the previously mentioned values of $x$ as $n$ goes to infinity, it is still an infinite product and thus an indeterminate form $1^\infty$ (I think) which I haven't been able to resolve.
We already know from OPs derivation \begin{align*} F(x)=F\left(x^2\right)=\cdots=F\left(x^{2^q}\right)\qquad\qquad q\in\mathbb{N}_0\tag{1} \end{align*}
We see the coefficients of odd powers of $x$ in $F(x)$ are zero. Let's have a look at \begin{align*} &a_0+a_1x+a_2x^2+\color{blue}{a_3x^3}+\cdots\\ &\qquad=a_0+a_1x^2+a_2x^4+\color{blue}{a_3x^6}+\cdots\\ &\qquad=a_0+a_1x^4+a_2x^8+\color{blue}{a_3x^{12}}+\cdots\\ &\qquad=a_0+a_1x^8+a_2x^{16}+\color{blue}{a_3x^{24}}+\cdots\\ &\qquad\ \ \vdots \end{align*}
We observe by iteratively using $F(x)=F\left(x^2\right)=F\left(x^4\right)=\cdots$ that since for instance $a_3=0$, also all coefficients \begin{align*} a_6&=[x^6]F(x)=[x^6]F\left(x^2\right)=[x^3]F(x)=0\\ a_{12}&=[x^{12}]F(x)=[x^{12}]F\left(x^4\right)=[x^3]F(x)=0\\ a_{24}&=[x^{24}]F(x)=[x^{24}]F\left(x^8\right)=[x^3]F(x)=0\\ &\ \ \vdots \end{align*} and in general all coefficients of the form \begin{align*} a_{3\cdot2^q}=[x^{3\cdot 2^q}]F(x)=[x^{3\cdot 2^q}]F(x^{2^q})=[x^3]F(x)=0\qquad\qquad q\geq 0 \end{align*} are zero. Here we use the coefficient of operator $[x^m]$ to denote the coefficient of $x^m$ of a series.