Open mloubout opened 8 years ago
Excellent, I just had a look and managed to make this compute a bit faster by removing the symbolic stuff and replacing it with pre-compiled NumPy/SciPy methods (see PR #3). This should now make it fast enough for integration with the Opesci-FD test suite. @mloubout, could you please describe how you would expect the respective error for an FD implementation to behave with varying spacial resolutions and different stencil orders?
The 2D analytical solution for a constant velocity model is ready. I putted the expression of the solution and the code, if you need more details, let me know and I'll try to .