multi-polynomial subs memory leak

[Vincent Delecroix]

Dear all,

Substitution on multivariate polynomials is leaking memory
(https://trac.sagemath.org/ticket/27261). This can actually
crash sage if used a lot (with a memory overflow error from
singular). The code of subs and __call__ use some low-level
calls to the singular library and this is likely to be the
origin of the problem. Unless somebody knows the code or
want to dig in, I propose to switch to naive symbolic
evaluation. This is infinitely slower but at least does not
leak memory. Please have a look.

To my mind, it would make sense to add this to the 9.3
release.

Best
Vincent

[Markus Wageringel]

vdelecroix schrieb am Freitag, 7. Mai 2021 um 10:09:59 UTC+2:

I propose to switch to naive symbolic
evaluation. This is infinitely slower but at least does not
leak memory.

Recently, I have looked into performance problems of multivariate power series. One of the bottlenecks is the use of subs() of polynomials, so I am worried that the proposed change has an unforeseen impact on performance in other places, especially if this is rushed into 9.3 at this point. This is a trade-off between running out of memory and running out of time in computations. 

[Dima Pasechnik]

Singular's  subst() command itself
appears to be very leaky, I don't know why we expect to be able to plug these leaks in libSingular.

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/efeba33b-a44c-4b75-b6d9-45c5c0c63e49n%40googlegroups.com.