On Mon, Apr 12, 2021 at 9:45 PM Zizhou Huang <zizhou@nyu.edu> wrote:
Hi Matthew,
Thanks for your advice!
I did some basic tests with the Navier-Stokes solver inside the Firedrake. For the lid-driven cavity problem on a NxNxN grid, it takes 23sec for N=5, 594sec for N=10, and 7424sec for N=20. For N=40 it almost takes forever, but it's really not a dense mesh in practice. I had the same experience with my own implementation of fem, using Pardiso to solve the linear system directly, and I've been finding efficient linear solvers to get rid of this problem, so I'm asking if the Firedrake team has any better options for the linear solver.
What solver did you use and how many iterates did it take? Also, posting the logs makes things easier. Thanks Matt
Best, Zizhou
On Mon, Apr 12, 2021 at 11:21 PM Matthew Knepley <knepley@gmail.com> wrote:
On Mon, Apr 12, 2021 at 11:00 AM Zizhou Huang <zizhou@nyu.edu> wrote:
To one whom it may concern,
I'm a PhD student at NYU. I'm working on a project on solving the Navier-Stokes equation with FEM, so I'm really interested in the amazing demo of solving the Navier-Stokes equation on the website. ( https://www.firedrakeproject.org/demos/navier_stokes.py.html)
I modified the code to solve the 3d lid-driven cavity problem with a uniform grid of size 40x40x40, and found that it takes extremely long time and large memory. I had the same issue with my code of similar sizes, which uses Pardiso to solve the linear system. I'm wondering if fem is indeed unsolvable for a large problem like this? Do you have any experience to make it faster by changing the parameters in the linear solver?
When you want to understand how a solver is scaling, start with a small problem, and look at the performance with -log_view. Then start increasing the problem size, and look at how the number of iterates, subiterates, and event times scale. Once you have that data, getting help is much easier.
Thanks,
Matt
Looking forward to hearing from you.
Best, Zizhou _______________________________________________ firedrake mailing list firedrake@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/firedrake
-- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener
https://www.cse.buffalo.edu/~knepley/ <http://www.cse.buffalo.edu/~knepley/>
-- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener https://www.cse.buffalo.edu/~knepley/ <http://www.cse.buffalo.edu/~knepley/>