Starting from the famous infinite product $$ (1+z)^2(1-z^2)(1+z^3)^2(1-z^4)(1+z^5)^2(1-z^6)\cdots=1+2z+2z^4+2z^9+2z^{16}+\dots $$ it is easy to show by induction that $$ \prod_{k\geqslant1}\left((1-z^k)(1-z^{2k-1})^N(1+z^{2k})^{N+1}(1+z^{2k-1})^{N+3}\right)=\sum_{n=-\infty}^{+\infty}z^{n^2} $$ for any $N\geqslant0$.

Can one give sense to passing to the limit w. r. t. $N\to\infty$ and if yes what does one obtain?

A by-question: it seems that this also holds for all negative $N$ but I somehow cannot prove it, is it true?


As it frequently happens with me after seeing it accurately formulated and typed after a while I saw a solution. Unfortunately it is disappointingly simple in that no limiting process of any kind is actually needed.

The point is that there is another very well known infinite product identity $$ P(t):=\prod_{k\geqslant 1}\left((1-t^{2k-1})(1+t^k)\right)=1; $$ and the lhs in the needed equality is $$ P(z^2)^N\prod_{k\geqslant1}\left((1-z^k)(1+z^{2k})(1+z^{2k-1})^3\right), $$ so $P(z^2)^N$ is $1$ while the terms in the product can be easily regrouped to get $$ \prod_{k\geqslant1}\left((1-z^{2k-1})(1-z^{2k})(1+z^{2k})(1+z^{2k-1})^3\right) =\prod_{k\geqslant1}\left((1-z^{4k-2})(1-z^{4k})(1+z^{2k-1})^2\right) =\prod_{k\geqslant1}\left((1-z^{2k})(1+z^{2k-1})^2\right) $$ which by the very first equality the question starts with is $\sum_{n\in\mathbb Z}z^{n^2}$.

Thus indeed this works for any $N$ (not just any integer but any number at all).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.