FastFieldSolvers Forum
FastFieldSolvers Forum
Home | Profile | Register | Active Topics | Members | Search | FAQ
Username:
Password:
Save Password
 All Forums
 FastFieldSolvers
 FasterCap and FastCap2
 asymmetric cap matrix
 New Topic  Reply to Topic
 Printer Friendly
Author  Topic Next Topic  

Graeme

USA
4 Posts

Posted - Aug 12 2019 :  15:07:02  Show Profile  Reply with Quote
Fastcap converges but the matrix is strongly asymmetric. Compare the right column to the bottom row
1.941 -0.872 -0.030 0.001 0.001 0.000 -0.540
-0.884 3.291 -0.921 -0.032 0.002 0.002 -0.928
-0.032 -0.924 3.288 -0.903 -0.027 -0.000 -0.911
-0.000 -0.027 -0.904 3.289 -0.923 -0.032 -0.914
0.001 0.003 -0.033 -0.919 3.284 -0.882 -0.924
0.000 0.002 -0.000 -0.029 -0.871 1.939 -0.542
-0.886 -1.413 -1.398 -1.398 -1.408 -0.889 10.000
I have several "non-negative off-diagonal element" warnings but they are small and
"Weighted Frobenius norm of the difference between capacitance (auto option): 0.00633859" seems good.

Enrico

402 Posts

Posted - Aug 12 2019 :  23:13:23  Show Profile  Reply with Quote
Without further information I cannot help you. Are there dielectrics in the input file? Can you share the minimal input file still presenting this effect?

In general, FasterCap does not force symmetry of the resulting matrix (FastCap, in comparison, always averages over symmetric elements).
There are some cases where inaccurate input files may lead to asymmetries, in other cases this may be caused by relatively large permittivity ratios when simulating conductors embedded in dielectrics.

Best Regards,
Enrico

Go to Top of Page

Graeme

USA
4 Posts

Posted - Aug 13 2019 :  12:16:27  Show Profile  Reply with Quote
I tried removing conductors to make a minimal example but it stopped converging, a common occurrence with me. Perhaps I'm missing something? How do I attach a zip file?
Go to Top of Page

Enrico

402 Posts

Posted - Aug 13 2019 :  16:13:38  Show Profile  Reply with Quote
You can not - due to spam preventing measures. If the file is short enough, as is should be, you can copy&paste it in the message.
Or, you can send the file to me at the address you can find on the 'contacts' page.

Best Regards,
Enrico
Go to Top of Page
   Topic Next Topic  
 New Topic  Reply to Topic
 Printer Friendly
Jump To:
FastFieldSolvers Forum © 2018 FastFieldSolvers S.R.L. Go To Top Of Page
Powered By: Snitz Forums 2000 Version 3.4.06