Menu

Discussion Forum

New Discussion
Filter

Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

About convergence

Posted Sep 17, 2012, 12:25 p.m. EDT Results & Visualization Version 4.3 4 Replies

Yuanhai Su
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Now I meet an unfamiliar issue. For the solver I had set the relative tolerance as 0.001 and used the direct solution. As the simulation processed, this value was lower than 0.001, but the simulation could not stop until the Maximum number of iteration reached the setting value (50). Finally it showed that :"Failed to find a solution. Maximum number of Newton iterations reached. Returned solution is not converged.". I can't understand that since the relative tolerance was lower than the setting value why It could not converge. Who knows this answer and help me. Thank you very much!!!
4 Replies Last Post Sep 19, 2012, 10:41 a.m. EDT
Yuanhai Su
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Sep 18, 2012, 3:21 a.m. EDT
It is so strange! Previously, if the relative tolerance is lower than the fixed value, the simulaiton will converge and stop. But now......Does anyone know it why? Thank you.
It is so strange! Previously, if the relative tolerance is lower than the fixed value, the simulaiton will converge and stop. But now......Does anyone know it why? Thank you.
Andrew Prudil Nuclear Materials
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Sep 18, 2012, 3:55 p.m. EDT
Are you sure you are not thinking of absolute tollerances instead of the relative tollerances?

It wouldn't make sense to say something is defined because the absolute error is less than the relative error tollerance. Think about dependent variables less than one. For example, if I am trying to calculate a concentration which is about 1e-6 and my relative error is 0.001 then I need to be converged within 1e-9 not 1e-3.

Another possiblity is that you have set a tolerance factor to a value less than 1 in your solver.
Are you sure you are not thinking of absolute tollerances instead of the relative tollerances? It wouldn't make sense to say something is defined because the absolute error is less than the relative error tollerance. Think about dependent variables less than one. For example, if I am trying to calculate a concentration which is about 1e-6 and my relative error is 0.001 then I need to be converged within 1e-9 not 1e-3. Another possiblity is that you have set a tolerance factor to a value less than 1 in your solver.
Yuanhai Su
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Sep 18, 2012, 5:55 p.m. EDT
Thank you for your response! I am very sure my setting was the relative tolerance and the tolerance factor was 1. I tried again and again, but it still could not converge. Even the previous simulation which was certainly converged can not be converged now. It is so strange....... In addition, the ErrEst was certainly much lower than the fixed tolerance.......but finally showed the simulation could not converge.








Are you sure you are not thinking of absolute tollerances instead of the relative tollerances?

It wouldn't make sense to say something is defined because the absolute error is less than the relative error tollerance. Think about dependent variables less than one. For example, if I am trying to calculate a concentration which is about 1e-6 and my relative error is 0.001 then I need to be converged within 1e-9 not 1e-3.

Another possiblity is that you have set a tolerance factor to a value less than 1 in your solver.

Thank you for your response! I am very sure my setting was the relative tolerance and the tolerance factor was 1. I tried again and again, but it still could not converge. Even the previous simulation which was certainly converged can not be converged now. It is so strange....... In addition, the ErrEst was certainly much lower than the fixed tolerance.......but finally showed the simulation could not converge. [QUOTE] Are you sure you are not thinking of absolute tollerances instead of the relative tollerances? It wouldn't make sense to say something is defined because the absolute error is less than the relative error tollerance. Think about dependent variables less than one. For example, if I am trying to calculate a concentration which is about 1e-6 and my relative error is 0.001 then I need to be converged within 1e-9 not 1e-3. Another possiblity is that you have set a tolerance factor to a value less than 1 in your solver. [/QUOTE]
Yuanhai Su
Send Private Message Flag post as spam

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Sep 19, 2012, 10:41 a.m. EDT
Now I had solved this issue. This is totally because of version! I updated the comsol tool, and then it became right.











Are you sure you are not thinking of absolute tollerances instead of the relative tollerances?

It wouldn't make sense to say something is defined because the absolute error is less than the relative error tollerance. Think about dependent variables less than one. For example, if I am trying to calculate a concentration which is about 1e-6 and my relative error is 0.001 then I need to be converged within 1e-9 not 1e-3.

Another possiblity is that you have set a tolerance factor to a value less than 1 in your solver.


Now I had solved this issue. This is totally because of version! I updated the comsol tool, and then it became right. [QUOTE] Are you sure you are not thinking of absolute tollerances instead of the relative tollerances? It wouldn't make sense to say something is defined because the absolute error is less than the relative error tollerance. Think about dependent variables less than one. For example, if I am trying to calculate a concentration which is about 1e-6 and my relative error is 0.001 then I need to be converged within 1e-9 not 1e-3. Another possiblity is that you have set a tolerance factor to a value less than 1 in your solver. [/QUOTE]

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.

Suggested Content