Using findiff to solve the stationary Schrödinger equation about findiff

[maroba]

The matrix representation of a FinDiff operator is a sparse matrix, and so is your kinetic term in the Hamiltonian. However, when you add the potential by saying +u.reshape(-1), you are turning the whole Hamiltonian into a dense matrix representation (although most values are zero). This eats up all your memory.

What you need to do: declare your potential also as a sparse matrix. Someone has written an article at Medium.com for this use case.

[yyanghy]

Thank you so much for your reply!