I think this last example is related to trac #27536 <https://trac.sagemath.org/ticket/27536> (conversion of mathematical constant such as pi to RDF leaks memory). There is still a memory leak (but smaller, I think) if pi^i is replaced with pi, but I don't see a memory leak if pi^i is replaced with RDF.pi(). I also don't see a leak with RDF.pi()^i, but I quickly get a RecursionError ("maximum recursion depth exceeded"). I know almost nothing about memory leaks, but I just wanted to point out that I noticed some related issues that might possibly provide an explanation (and certainly seem to be relevant to a complete solution).
On Thursday, July 23, 2020 at 3:39:09 PM UTC-6, Spencer Dembner wrote: > > It may be related, but I don't think it's exactly the same issue. For > example, the following, involving no square root, still leaks memory: > > for i in [1,..,20000]: > if i%100 == 0: > print(i); > print(get_memory_usage()); > C = continued_fraction(pi^i); > C.denominator(100); > > > > > > > > > On Thursday, July 23, 2020 at 4:27:46 PM UTC-5, Dave Morris wrote: >> >> Is this the issue that was reported in Trac #27185 >> <https://trac.sagemath.org/ticket/27185> (defect: sqrt memory leak)? >> >> On Thursday, July 23, 2020 at 3:12:07 PM UTC-6, Spencer Dembner wrote: >>> >>> Yup- what you wrote gives me the same result as far as climbing memory >>> usage. >>> >>> On Thursday, July 23, 2020 at 4:07:23 PM UTC-5, vdelecroix wrote: >>>> >>>> Thanks for your report. >>>> >>>> Actually, it does not seem to have much to do with continued >>>> fractions but rather with the symbolic ring >>>> >>>> sage[4]: for i in [2500000,..,3000000]: >>>> .......: if i%1000 == 0: >>>> .......: print(i); >>>> .......: print(get_memory_usage()) >>>> .......: _ = RIF(sqrt(i)) >>>> >>>> Le 23/07/2020 à 18:30, Spencer Dembner a écrit : >>>> > When using continued_fraction to compute denominators of continued >>>> fraction >>>> > convergents, I'm encountering what seems to be a memory leak. I'm >>>> running >>>> > SageMath 9.0 on Windows 10 64-bit. If I run the following, >>>> > >>>> > for i in [2500000,..,2600000]: >>>> > if i%1000 == 0: >>>> > print(i); >>>> > print(get_memory_usage()); >>>> > C = continued_fraction(sqrt(i)); >>>> > C.denominator(100); >>>> > >>>> > then I see memory usage steadily climbing as I iterate through the >>>> loop. On >>>> > the other hand, if I initialize sqrt(i) as an algebraic number, >>>> memory >>>> > usage is essentially stable: >>>> > >>>> > for i in [2500000,..,2600000]: >>>> > if i%1000 == 0: >>>> > print(i); >>>> > print(get_memory_usage()); >>>> > if sqrt(i) not in QQ: >>>> > K.<sqrti> = QuadraticField(i); >>>> > C = continued_fraction(sqrti); >>>> > C.denominator(100); >>>> > >>>> >>> -- 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+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/99c77db7-5eea-43e2-aa23-2f10dadfc8feo%40googlegroups.com.
