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 The information provided may be out of date.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.
Parametric Sweep gives different results with separate calculations
Posted Jul 11, 2014, 5:00 p.m. EDT Parameters, Variables, & Functions, Studies & Solvers Version 4.3a 1 Reply
Please login with a confirmed email address before reporting spam
Hello,
Strange things happened when I use Parametric sweep function today. Initially, I solved a model by setting a parameter as a fixed number ( say "a"). Then I swept this parameters with different values (say "a", "b" , and 'c"). Both calculations converged but when I checked the results, the one from separate calculation ("a" only), is different from the one (with same parameter value "a") obtained from parametric sweep.
I checked several times and redid the modeling for couple of times, but do know the reasons.
Someone also encountered similar problems? Which results should I trust and what might be the reason?
Please advice,
Thanks!
Strange things happened when I use Parametric sweep function today. Initially, I solved a model by setting a parameter as a fixed number ( say "a"). Then I swept this parameters with different values (say "a", "b" , and 'c"). Both calculations converged but when I checked the results, the one from separate calculation ("a" only), is different from the one (with same parameter value "a") obtained from parametric sweep.
I checked several times and redid the modeling for couple of times, but do know the reasons.
Someone also encountered similar problems? Which results should I trust and what might be the reason?
Please advice,
Thanks!
1 Reply Last Post Jul 14, 2014, 7:06 a.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Dear Neverland,
A parametric sweep will typically reuse the previous result as initial conditions for the next parameter. This means that a different solving path is taken.
Next to this, a non-linear problem might have multiple solutions which are all valid. Depending on the initial conditions you can find a different result. (for instance different flow profiles can be found in a CFD model)
Maybe you can elaborate a bit on what type of difference we are looking at. Do you have a different magnitude of the solution (by how much), or do you have a different shape of the solution?
If the magnitude is highly different, you might want to check the results with a lower relative tolerance (say 1e-5 vs. 1e-3) and recheck the results.
If the shape is different, I typically rely more on the sweep results as this often follows a natural path.
Note that if you do not want to reuse the previous results, uncheck "use parametric solver" in the parametric sweep.
Best regards,
Frank
A parametric sweep will typically reuse the previous result as initial conditions for the next parameter. This means that a different solving path is taken.
Next to this, a non-linear problem might have multiple solutions which are all valid. Depending on the initial conditions you can find a different result. (for instance different flow profiles can be found in a CFD model)
Maybe you can elaborate a bit on what type of difference we are looking at. Do you have a different magnitude of the solution (by how much), or do you have a different shape of the solution?
If the magnitude is highly different, you might want to check the results with a lower relative tolerance (say 1e-5 vs. 1e-3) and recheck the results.
If the shape is different, I typically rely more on the sweep results as this often follows a natural path.
Note that if you do not want to reuse the previous results, uncheck "use parametric solver" in the parametric sweep.
Best regards,
Frank