scipy/log-of-scipy-det
| Rule | log-of-scipy-det |
|---|---|
| Pack | SciPy |
| Severity | warning |
| Fix | unsafe, with xpd check --fix --unsafe-fixes |
| Library versions | every version |
linalg.det overflows or underflows for a large matrix, so its log is inf or -inf. Use np.linalg.slogdet(a)[1].
logdet = np.log(linalg.det(cov))logdet = np.linalg.slogdet(cov)[1]What the rule reports
Section titled “What the rule reports”The rule reports np.log(linalg.det(a)), np.log(np.abs(linalg.det(a))) and
np.log(abs(linalg.det(a))), where linalg is scipy.linalg or det was imported from it. On two
covariance matrices the old form gave -inf and inf, where slogdet lost less than one digit, as
the table shows. np.log(np.linalg.det(a)) belongs to the NumPy rule
log-of-det.
| Covariance matrix | np.log(linalg.det(a)) |
Digits that slogdet lost |
|---|---|---|
| 120 features at scale 1e-3 | -inf | 0.54 |
| 80 features at scale 1e4 | inf | 0.27 |
How to fix it
Section titled “How to fix it”xpd check --fix --unsafe-fixes applies the fixes of this rule. Check the result, because an unsafe fix can change what the code does.
The unsafe fix writes np.linalg.slogdet(a)[1].
Message
Section titled “Message”xpd prints this message with the issue. A name that starts with $ stands for code from your file.
$D($A) overflows to inf or underflows to 0 for a large matrix, so its log is inf or -inf. On covariance matrices of 80 and 120 features that lost all 16 digits where slogdet lost less than 1. Write $NP.linalg.slogdet($A)[1].Browse the other rules of the SciPy pack.