It is possible to define a position dependent variable in COMSOL by some sort of logical conditional expression, like: 8*(y10). However, in practical 2D example, there is some mismatch (in plot) between the variable that is introduced and the actual definition of variable. Strictly speaking, some of the values for different X, right on the edge (e.g. y=10 in above example) does not introduce well and there is irregularity on the border along the x axis. Right on edge, several values along the x direction are considered to be 0 while it must be 8. I tried to change the mesh but the position of abnormallies along the x axis are changed. This makes me suspicious about the precision of such type of definition.
Does anybody have a suggestion for improving the quality of such position dependent variable?
Dear Markus
Thank you very much for paying attention to my question. Unfortunately, changing the mesh and adding "eps" did not solve this problem. I thought maybe, I can solve this problem by using "step function" but I don't know how to make the a step function in 2D.
Dear Markus
Thank you very much for paying attention to my question. Unfortunately, changing the mesh and adding "eps" did not solve this problem. I thought maybe, I can solve this problem by using "step function" but I don't know how to make the a step function in 2D.
Dear Markus
I am very thankful for your suggestion. it is working if I consider smoothing, while the result will keep having abnormally if I exclude smoothing.
I wish you good luck in your research
- Badges
- Science topic