I have a code I'm working on that involves solving a 1D Schrodinger equation using a Crank-Nicolson time step. The code is written in NumPy/SciPy, and I was doing a bit of profiling and discovered that the bulk of the simulation time is taken up with calls to the linear solver (spsolve in this case). I'm using a simple discretization such that my matrices are symmetric and tridiagonal, and I was wondering if anyone had tried doing things like implementing the Thomas algorithm directly and if this was able to beat spsolve.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.