Hello, is there anyone using low Mach number preconditioning (whatever) with the Thompson/Lele treatment of boundary conditions ??? If so, please reply, I ll explain the problem I got. Regards, Dario

Hello,

I have to admit that I don't know what the Thompson/Lele treatment is. Currently I am working on a (mainly) one-dimensional compressible solver and for outflow boundaries I use a formulation based on the characteristics.

In 1D this is nice and simple ...

If used with a preconditioner, however, I get a somewhat strange behaviour on the outflow boundaries. Currently most of the solver consists of "down-scaled" code from a 3D program. Hence there are various sources of errors and the boundary problem might be a trivial one as well.

In general, since you change the characteristics with the preconditioner, I would assume that the boundary needs some attention paid to it as well.

I don't know if that is similar to your problem or not.

cheers, Oliver

MAny thanks for the message. This is exactly my problem. The preconditioned system has different eigenvalues and eigenvectors, thus, as all the papers on the topic report, bc based on characteristic should be modified, but what they do not report...is that the preconditioned system DOESN'T HAVE CHARACTERISTICS (I would really be happy if somebody showed me I am wrong on this ), i.e., the system, cannot be diagonalized. Since the diagonalization is essential in the Thompson/Lele treatment of BC, I can't figure out how to modify it. Anyway, I think the issue is not trivial. Regards,