scipy/removed-keyword
| Rule | removed-keyword |
|---|---|
| Pack | SciPy |
| Severity | error |
| Fix | unsafe, with xpd check --fix --unsafe-fixes |
| Library versions | every version |
These keywords and argument values were removed from SciPy and raise TypeError on SciPy 1.18.1. Use the replacement that the message names, such as rtol= for tol=.
x, info = cg(A, b, tol=1e-8)w = linalg.eigh(a, eigvals=(0, 2), eigvals_only=True)dist, idx = tree.query(points, k=3, n_jobs=-1)x, info = cg(A, b, rtol=1e-8)w = linalg.eigh(a, subset_by_index=(0, 2), eigvals_only=True)dist, idx = tree.query(points, k=3, workers=-1)What the rule reports
Section titled “What the rule reports”The rule reports keywords and argument values that SciPy removed and that raise on SciPy 1.18.1.
| Function | Removed form | Fix |
|---|---|---|
cg, gmres and the other iterative solvers of scipy.sparse.linalg |
tol=, atol='legacy', a fourth positional argument |
rtol= for tol= |
gmres |
restrt= |
restart= |
linalg.solve |
sym_pos= |
assume_a='pos' for sym_pos=True |
linalg.eigh, linalg.eigvalsh |
eigvals=, turbo= |
subset_by_index= for eigvals= |
linalg.pinv, linalg.pinvh |
cond=, rcond= |
none |
integrate.simpson |
even= |
deletes it |
integrate.cumulative_trapezoid |
an initial other than 0 |
none |
the query methods of KDTree and cKDTree |
n_jobs= |
workers= |
stats.kendalltau |
initial_lexsort= |
deletes it |
stats.iqr, stats.median_abs_deviation |
scale='raw' |
scale=1.0 |
optimize.fmin_l_bfgs_b |
iprint=, disp= |
deletes them |
spatial.distance.cdist, spatial.distance.pdist |
the metric 'wminkowski' |
none |
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 safe fix writes rtol= for tol= on the sparse.linalg solvers, restart= for gmres restrt=, assume_a='pos' for linalg.solve sym_pos=True, subset_by_index= for eigh and eigvalsh eigvals=, workers= for n_jobs= on the cKDTree query methods, and scale=1.0 for iqr scale='raw', and it deletes kendalltau initial_lexsort= and fmin_l_bfgs_b iprint= and disp=. On the last SciPy that takes the old form, from 1.8.1 to 1.17.1, each new form gives a result with an equal pickle, and fmin_l_bfgs_b prints nothing either way. Unsafe: it deletes simpson even=, because even='avg' gives another result, and it writes workers= for n_jobs= on KDTree and scale=1.0 for median_abs_deviation scale='raw', because neither ever took the old form. The other forms have no fix, because the call already passes the new keyword, or the message names two rewrites, or the value the new form needs is not in the file.
Message
Section titled “Message”xpd prints this message with the issue. A name that starts with $ stands for code from your file.
$F no longer takes tol= and raises TypeError. Write rtol=, the relative tolerance, and set atol= if the old code relied on an absolute one.Browse the other rules of the SciPy pack.