-
Notifications
You must be signed in to change notification settings - Fork 13
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Issues with Computing Groebner Basis for High-Order Runge Kutta Systems with 8 Stages and Order 7 #62
Comments
Hi @jpcurbelo , thanks for sharing this interesting system! |
You mentioned that you also have smaller similar systems. If you expect those to have a "similar" groebner basis as this large one, feel free to share the small systems as well, it will perhaps help |
Hi @sumiya11. Thanks for your quick response. In fact, the system for RK(s=8, p=7) doesn't have a solution - that's what I expected to obtain from this. For example, RK(s=4, p=4) has a solution (multiple solutions, in fact) and I got this Grobner basis. On the other hand, RK(s=6, p=6) doesn't have a solution, and the Please, let me know if I didn't make myself clear enough :) Thanks again! |
Hey @sumiya11. I was wondering if you have had the time to look at this issue (a significant amount of memory RAM is required for specific problems). Please, let me know if you think there is something I could (try to) do on my end to somehow improve the performance of the jobs I'm running. Many thanks for your efforts! |
Hi @jpcurbelo. I have looked at the issue, the problem is that in F4 algorithm we don't limit the maximum number of critical pairs that can be handled at the same time, rather strangely though I am at it ( #64 ), but haven't figured out the best fix yet. Will try again this weekend. Sorry for this taking so long |
I think I messed up git, so the issue got closed automatically 😅 . |
Hi @jpcurbelo , so there is good news and bad news:
On my machine, while computing RK(s=8, p=7) with The commands The I will work on reducing the overall memory footprint of Groebner.jl this summer, but I think at the moment Groebner.jl unfortunately cannot tackle such systems |
As a sidenote, if you do not care about the basis monomial ordering, you can explicitly set it to # system is RK(s=6, p=6)
@time groebner(system);
# 258.834883 seconds (661.72 M allocations: 154.943 GiB, 5.61% gc time)
# Max. RAM: 9100.656 MiB
@time groebner(system, ordering=DegRevLex());
# 28.632552 seconds (7.53 M allocations: 2.620 GiB, 2.63% gc time)
# Max. RAM: 1676.414 MiB
@time groebner(system, ordering=DegRevLex(), maxpairs=1000);
# 12.556571 seconds (6.62 M allocations: 1.900 GiB, 4.33% gc time)
# Max. RAM: 1574.133 MiB |
I think there is a chance that incremental signature GB algorithms can deal with these problems (e.g., https://www.singular.uni-kl.de/Manual/4-0-3/sing_391.htm). I'll check if I can make it work via Singular.jl |
Hi @sumiya11! Many thanks. Your efforts are highly appreciated. I will take a look at the updates and materials you mentioned. I'll keep my finger crossed so you can find time to work on reducing the memory footprint. Once again, thank you so much. Your package has already helped me with other problems as well. |
@jpcurbelo |
Hi @pogudingleb. I don't think there is a simple way to describe how the systems are formed but you are right, there are key substitutions involved. |
I'm having trouble computing the Groebner basis for high-order Runge Kutta systems. Specifically, I'm working with a problem that has 8 stages and order 7, and the system of equations is given in this file. However, when I try to compute the Groebner basis for this problem, it seems to require more than 1Tb of RAM, which is causing issues.
For lower-order problems, it seems to work just fine. However, for the higher-order problem, I'm not sure if there's some workaround in the inputs that I might be missing, or if this behavior is to be expected for this type of problem. If anyone has any suggestions or insights into this issue, I would greatly appreciate it. Thank you!
The text was updated successfully, but these errors were encountered: