Hello,
I have a pretty simple 1D simulation. I can solve it with COMSOL just fine, and the results are good, however when I take derivatives (even just the first derivative), I notice there can be quite a bit of error.
I've tried increasing mesh size, as well as using higher order geometry shapes up to quintic, unfortunately neither of these strategies seem to improve the results.
In order to rectify this, I've been taking the results generated by COMSOL, and then use those as boundary conditions in MATLABs bvp5c routine, resolve the ODE and then use the derivatives generated by MATLAB. For some reason the derivatives generated by bvp5c are good, while those generated by COMSOL aren't so good.
Any suggestions?
Thanks,
Mark
I have a pretty simple 1D simulation. I can solve it with COMSOL just fine, and the results are good, however when I take derivatives (even just the first derivative), I notice there can be quite a bit of error.
I've tried increasing mesh size, as well as using higher order geometry shapes up to quintic, unfortunately neither of these strategies seem to improve the results.
In order to rectify this, I've been taking the results generated by COMSOL, and then use those as boundary conditions in MATLABs bvp5c routine, resolve the ODE and then use the derivatives generated by MATLAB. For some reason the derivatives generated by bvp5c are good, while those generated by COMSOL aren't so good.
Any suggestions?
Thanks,
Mark