I'm programing in Matlab an elliptic mesh generator. But I've got problems when I try to cluster the iso\eta lines to the internal boundary. Some people told me about using force functions like:

P=A*sign(xi-xio)*exp(-B*(eta-etao))

to cluster the lines on some point (xio,etao). I DO not want to cluster any iso\xi line. So that, the value of xo has no sense for me. I'm only intending to cluster the iso\eta lines around the surface body (the internal boundary).

What function must I use instead of that given above?

Thanks in advance.

check the book section Thompson's online-book about grid generation. It has clustering technique

It was just what I had done when I got into troubles. But it is not enough clear. I need some experience of people who have tried to compute something similar.

Hi, Doctor Blade

- Your point (xio, etao), not to be clustered, just can not be used. and no lines will not be clustered in the point. Just Dont use it !!!

- Your control function is similar with Thomson's one, I think. Thomson's are very good example, however, so many altinatives can be used... In my work, the full boundary control method is very robust one in structured grid. Of course, so many of the other functions are possible....

- If you want to know about them then e-mail me....

-Good Luck !!!