|Re: [eigen] Best Solver for Large but Sparse Linear System|
[ Thread Index |
| More lists.tuxfamily.org/eigen Archives
- To: eigen@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [eigen] Best Solver for Large but Sparse Linear System
- From: Benoit Bardet <benoit.bardet@xxxxxxxxx>
- Date: Tue, 3 Sep 2019 17:12:35 +0200
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=mime-version:references:in-reply-to:from:date:message-id:subject:to; bh=Zit6LYZ0AbDyyFSe/XI4cmWn/wLbSnamQA+hW3eDThE=; b=r3YjzI7G15jtNPLu7o5OlOMZu29crCf7j1tigLCI8eH7z7o+4ktXmJGMeOvQyy8B9B mhZp0sEHB32Of/sq1k0BAi/CbjHDGC2sS7FJpeUmeMKuYrdB6pNLHVHX8bDVf25zEC64 hXo5JtoM2tXwpEXBBdTaWGozpZ4ZhQXwtXCR+Q7klfXkFksRpC62z8d0AKAOSFJrEDg4 nCa1nb03/gowuKwXfoCs/I4CPiRb6lGgSiwXijpQJJBw3iL6IvdUT4OPpkDf3Psdxf/O j6UkdUpEvCFgqr+2zRADYrv52WQHIfMf6s9u6Ux4NhcfztVsFRe1mQcHnAl/SM0HdBI+ umpg==
It depends on your matrix condition factor (not sure of the name in english).
A way I like for bad condition number is IncompleteLUT preconditionner with GMRES. Try different fill factor for ILUT.
For better matrix, UMFPACK is very efficient. I'm not sure it is in eigen. Maybe with Diagonal preconditionner.
You can use CG as well with Diag precond.
But do the benchmark on a real matrix, not a random one. And benchmark in time and not in iterations.
I used GMRES for 8e6 by 8e6 matrix with about 10 non Zero as well per line (Finite Element with bad meshes). Be careful with memory.
I have a linear system with a matrix size of about 2E+06 x 2E+06. The maximum number of non-zero matrix elements in each row is 10. Which solver would you recommend for this system?
The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.