5
$\begingroup$

In our ongoing work to speed up symbolic summation and other similar algorithms in Sagemath, we notice that naive implementations of Gosper and Wilf-Zeilberger (a.k.a. WZ) algorithms are usually quite slow.

What is the state of the art here? E.g. is there software available that can derive Clausen $_4 F_3$ identity (see (V) on p.43 of the A=B book) quickly (few seconds on a laptop, say?) using the the WZ algorithm?

$\endgroup$
6

1 Answer 1

4
$\begingroup$

The group at RISC has been working aggressively towards developing improved algorithms for many computer algebra problems, including the Zeilberger's. So, it is a good place to ask.

Meantime, I just wanted to say this regarding Clausen $_4 F_3$ identity that the OP mentioned. If you read on in the book A=B, on page 127 there is a rewrite of the identity and a procedure outlined as well. I just checked it myself using a much earlier version from Zeilberger, called EKHAD, and executed in Maple. The verification of Clausen is instant!

$\endgroup$
2
  • $\begingroup$ p.127 gives $R(n,k)$. Do you mean to say that actually computing $R(n,k)$ is instant, or that carrying on if $R(n,k)$ is already known is instant? $\endgroup$ Jan 27, 2017 at 21:50
  • $\begingroup$ I mean computing $R(n,k)$ is instant and hence proving the identity is instant, if you know how the WZ machinary works. $\endgroup$ Jan 27, 2017 at 22:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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