1

Hello I would like to solve this : $$ \lim_{x \rightarrow 1^-} \sum_{n=0}^{+\infty}(-1)^{n}x^{n^2}=\dfrac{1}{2} $$ So my idea was to use the triple product of Jacobi wich is the following : $$\sum_{n=-\infty}^{\infty}z^nq^{n^2}=\prod_{n\geq 0}(1-q^{2n+2})(1+zq^{2n+1})(1+z^{-1}q^{2n+1})$$ But I have some problems like the negative infinity and others things ...

Could someone help me ?

2 Answers2

2

Let us set $x=e^{-z}$. The original limit can be computed through a convolution with an approximate identity: $$ \lim_{z\to 0^+}\sum_{n\geq 0}(-1)^n e^{-n^2 z} = \lim_{m\to +\infty}\int_{0}^{+\infty}\sum_{n\geq 0} (-1)^n e^{-n^2 z} m^2 e^{-m^2 z}\,dz = \lim_{m\to +\infty}\sum_{n\geq 0}\frac{m^2 (-1)^n}{m^2+n^2}$$ On the other hand, by applying $\frac{d}{dz}\log(\cdot)$ to the Weierstrass product for the hyperbolic sine function we get an explicit form for the last series involved: $$ \lim_{z\to 0^+}\sum_{n\geq 0}(-1)^n e^{-n^2 z} = \lim_{m\to +\infty}\left(\frac{1}{2}+\frac{\pi m}{\sinh(\pi m)}\right)=\color{red}{\frac{1}{2}}$$ so the given limit is simple to compute even without invoking Jacobi's triple product, the Poisson summation formula or the identity $r_2(n) = 4(\chi_4*1)(n)$.

Jack D'Aurizio
  • 353,855
1

With $z=-1$ and $q=x$, your formula becomes $$\sum_{n=-\infty}^{\infty}(-1)^nx^{n^2}=\prod_{n\geq 0}(1-x^{2n+2})(1-x^{2n+1})^2.$$ Since $(-1)^{-n}=(-1)^n$, the LHS is $\displaystyle2\sum_{n=0}^{+\infty}(-1)^{n}x^{n^2}-1$, giving $$\sum_{n=0}^{+\infty}(-1)^{n}x^{n^2}=\frac12\left(1+\prod_{n\geq 0}(1-x^{2n+2})(1-x^{2n+1})^2\right).$$ Of course, the product converges $\to0$ with $x\to1-$, and faster than any power of $1-x$. The Poisson sum formula would give the same result, naturally.

  • (+1) This is also pretty neat, indeed. – Jack D'Aurizio Feb 10 '18 at 15:15
  • This is easier to understand than the solutions based on Poisson formula or the transformation formulas for theta functions which are given in linked question. The only hard part is the Jacobi Triple Product which OP is willing to use. +1 – Paramanand Singh Feb 11 '18 at 09:22
  • It's so beautiful that I accept your accept your answer thanks a lot ! –  Feb 11 '18 at 17:40