torch.sparse.spsolve

torch.sparse.spsolve(input, other, *, left=True) → Tensor

Computes the solution of a square system of linear equations with a unique solution. Its purpose is similar to torch.linalg.solve(), except that the system is defined by a sparse CSR matrix with layout sparse_csr.

Parameters

• input (Tensor) – a sparse CSR matrix of shape (n, n) representing the coefficients of the linear system.
• other (Tensor) – a dense matrix of shape (n, ) representing the right-hand side of the linear system.
• left (bool, optional) – whether to solve the system for input @ out = other (default) or out @ input = other. Only left=True is supported.