-
Notifications
You must be signed in to change notification settings - Fork 86
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
Adds (S)SOR Preconditioner #1633
base: develop
Are you sure you want to change the base?
Conversation
499f37f
to
f5b8d2a
Compare
97d4d54
to
f8cbc05
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM in general.
I am thinking of how to add the test for GaussSeidel in a cheap way or just do not test it, which calls SOR with relaxation 1
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM. The config test from registry is not yet there
only reference kernels are available
- documentation - tests - don't build upper solver if not symmetric Co-authored-by: Yu-Hsiang M. Tsai <yhmtsai@gmail.com>
This PR adds the core and reference implementation of a (S)SOR preconditioner. It uses triangular solves to apply the preconditioner. The preconditioner is either a single lower triangular solver, or a lower triangular solve followed by an upper triangular one. Thus the preconditioner is represented as a
gko::Combination
. TheSor
class itself is only a linop factory, similar to theLu
factorization. It can be parameterized by therelaxation_factor
(0 < w < 2
), symmetric or not (sym version has the two triangular solves), and the solver factories for the upper and lower triangular matrices.This also adds a Gauss-Seidel preconditioner as a special case of SOR.
Todo: