Absorbing boundary conditions in different velocity domains still result in reflections.

[meng deng]

  Thanks all. I need to simulate seismic wave propagation in media with different wave velocities, and step-24 is a great starting point. If seismic waves propagate through different velocity media,  I only modify the mass_matrix and boundary_matrix. However, there are still reflections occurring at the boundaries of the computational domain.  Is the modification correct?

[Wolfgang Bangerth]

Meng:
I believe what you see is a consequence of what this sentence in step-24
says:
In general, this is a hard problem: Good absorbing boundary conditions
are nonlinear and/or numerically very expensive. We therefore opt for
a simple first order approximation to absorbing boundary conditions
[...]

The method used in step-24 absorbs waves that hit the boundary
perpendicularly, but it lets some fraction of waves through that hit at
an oblique angle. If that is unacceptable in your application, you need
to use better absorbing boundary conditions.

Best
WB