Earlier, I posted this MO question to which Max Alekseyev found counter-examples. I realized that there was some error in my computational programming with Maple. Oh, well $\dots$ I have scaled down the problem.

Recall: Let $\nu_2(x)$ denote the $2$-adic valuation of $x$ and $s(x)$ stand for the number of $1$’s in the $2$-ary (binary) expansion of $x$.

Consider the sequence $w_0=1$ and $w_{n+1}=\sum_{i=0}^nw_i^2w_{n-i}^2$.

QUESTION. Is the following true? If $C_n=\frac1{n+1}\binom{2n}n$, then $\nu_2(w_n)=(C_n\mod2)+2s(n+1)-3$. Equivalently, $$\nu_2(w_n)=\begin{cases} 2s(n+1)-2 \qquad\text{if $n=2^k-1$} \\ 2s(n+1)-3 \qquad\text{otherwise}. \end{cases}$$


This is still not true. The smallest counterexample is given by $n=30$ with $\nu_2(w_{30}) = 10$ and $2s(31)-3=7$. This can be easily verified by computing $w_n$ modulo $2^{11}$.


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.