scipy/solve-in-loop
| Rule | solve-in-loop |
|---|---|
| Pack | SciPy |
| Severity | warning |
| Fix | none, edit by hand |
| Library versions | every version |
spsolve or solve inside a loop factors the same matrix again on every pass. Factor it once before the loop with splu, then call lu.solve in the loop.
xs = [spsolve(A, b) for b in sides]lu = splu(A.tocsc())xs = [lu.solve(b) for b in sides]What the rule reports
Section titled “What the rule reports”The rule reports scipy.sparse.linalg.spsolve(A, b) and scipy.linalg.solve(A, b) inside a
comprehension or a loop that doesn't assign A or take it as its loop name. Each call factors
A again, and factoring once was faster:
| Matrix | Right sides | Factoring once was faster by |
|---|---|---|
| sparse, 10,000 rows | 100 | 17 to 20 times |
| dense, 1,000 by 1,000 | 20 | 18 to 23 times |
How to fix it
Section titled “How to fix it”xpd has no automatic fix, so make the change by hand.
Message
Section titled “Message”xpd prints this message with the issue. A name that starts with $ stands for code from your file.
spsolve factors $A again on every pass of this loop. Factor it once before the loop with lu = splu($A.tocsc()) from scipy.sparse.linalg, then call lu.solve(b) in the loop, which ran 17 to 20 times faster for 100 right sides.Browse the other rules of the SciPy pack.