NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI · #2188 most downloaded on PyPI
The official type stubs for SciPy
Last release 14 days ago
20 Sep 2026
Ships fairly regularly
a new release about every 3 weeks
Nearly every release is documented
notes for 38 of 38 stable releases
2 versions withdrawn
withdrawn after publishing
2 years old
44 releases · first in 2024
One column per month.
improve return dtypes of deprecated distance functions by @jorenham in #2342
This release targets SciPy 1.18.1 and supports Python 3.12–3.15, NumPy 2.0–2.7, and optype 0.15–0.19.
scipy.clusterscipy.constantsscipy.fftscipy.fftpackscipy.interpolatescipy.linalgorthogonal_procrustes: narrow return dtype by @jorenham in #2322qr_{delete, insert, update} : improve return dtypes by @Aniketsy in #2134khatri_rao: improve return dtypes by @jorenham in #2345scipy.ndimage{minimum,maximum}*: narrow return types when index is given by @jorenham in #2315watershed_ift: improve return dtypes by @jorenham in #2347scipy.optimizescipy.signalStateSpace.{A,B,C,D}: improve getter dtype by @jorenham in #2341hilbert: improve return dtype for the fallback overload by @jorenham in #2346{*_zpk, hilbert2}: gradual fallback overload dtypes by @jorenham in #2353scipy.sparsescipy.spatialscipy.specialsinc: propagate non-scalar dtypes and shape-type by @jorenham in #2326diric: narrow return dtype by @jorenham in #2327logsumexp: improve return dtype when return_sign=True by @jorenham in #2349scipy.stats{trim1,trimboth}: improve return dtypes by @jorenham in #2340sem: improve return dtypes by @jorenham in #2343Covariance.{whiten, colorize}: improve return dtypes and shape-types by @jorenham in #2344*entropy and mannwhitneyu and estimated_cdf by @jorenham in #2352rankdata: narrow fallback overload dtype by @jorenham in #2357multivariate_t: shape-typing support by @jorenham in #2381multivariate_hypergeom: shape-typing support by @jorenham in #2382normal_inverse_gamma: shape-typing support by @jorenham in #2384vonmises_fisher: shape-typing support by @jorenham in #2385uniform_direction: shape-typing support by @jorenham in #2386dirichlet_multinomial: shape-typing support by @jorenham in #2388scipy.integratescipy.interpolatescipy.linalgscipy.optimizescipy.signalscipy.specialld->d ufuncs by @jorenham in #2305zeta: propagate float32 and complex64 dtypes by @jorenham in #2317softplus: fix promotion rules by @jorenham in #2318float32 when other argument is float64 by @jorenham in #2320scipy.statsmedian_abs_deviation: propagate float32 dtypes by @jorenham in #2306pearsonr: widen pvalue dtype for non-f64 inputs by @jorenham in #2309binned_statistic*: widen edge output dtypes by @jorenham in #2311probplot: propagate dtype to 2nd return array by @jorenham in #2312float32 by @jorenham in #2313tukey_hsd: result statistic can be float32 by @jorenham in #2314median_abs_deviation: fix center param type rejecting np.mean by @jorenham in #2319mstats.pearsonr: fix float32 return dtype by @jorenham in #2328multivariate_normal: shape-typing fixes by @jorenham in #2383Akkudoktor-EOS/EOS to the list of downstream projects by @jorenham in #2372integrate._quadrature: prefer onp.ArrayND over onp.Array by @jorenham in #2336ndimage.{minimum,maximum}: use Any instead of Incomplete as scalar fallback by @jorenham in #2358_lib: annotate test module attribute by @jorenham in #2378sparse.linalg.*: fix type of internal test module attributes in deprecated submodules by @jorenham in #2379Counts are authored pull requests; + marks a first-time contributor.
Full Changelog: v1.18.1.0...v1.18.1.1
Update future SciPy (2.0) versions in deprecation messages by @jorenham in #2074
This scipy-stubs release comes with many improvements and even more fixes. It targets SciPy 1.18.1 and supports Python 3.12–3.15, NumPy 2.0–2.7, and optype 0.15–0.19.
linalg.blas & linalg.lapack complete type coveragespatial.tsearch: update deprecation target version to match upstream by @jorenham in #2160sparse._sputils.get_sum_dtype (private): update return type for SciPy 1.18.1 by @jorenham in #2267linalg.bandwidth: restore int return types for SciPy 1.18.1 by @jorenham in #2268stats.binomtest: restore int types of returned k and n for SciPy 1.18.1 by @jorenham in #2269linalg.pinv[h]: improved shape-typing and dtype propagation by @jorenham in #1997linalg.matrix_balance: improved shape-typing and dtype propagation by @jorenham in #1998linalg.matmul_toeplitz: improved shape-typing and dtype propagation by @jorenham in #1999linalg: improved matrix function shape-typing and dtype propagation by @jorenham in #2000linalg._batched_linalg: add missing private function stubs by @jorenham in #2021interpolate._interpnd: stub _ndim_coords_from_arrays by @jorenham in #2024sparse.csgraph._flow: stub private functions by @jorenham in #2027_lib._ccallback_c: ctypes declaration stubs by @jorenham in #2028stats: improve ttest_1samp by @jorenham in #2044stats: improve kstest by @jorenham in #2045stats: improve ks_1samp by @jorenham in #2046stats: improve ranksums by @jorenham in #2047stats: improve ttest_ind_from_stats by @jorenham in #2048stats: improve pointbiserialr by @jorenham in #2050stats: improve friedmanchisquare by @jorenham in #2051stats: improve alexandergovern by @jorenham in #2052stats: improved binned statistic function by @jorenham in #2054linalg: improve eigvals and eigvalsh by @jorenham in #2055linalg: improve hessenberg by @jorenham in #2056linalg: improve eigvals_banded by @jorenham in #2057linalg: improve eigvalsh_tridiagonal by @jorenham in #2058linalg: improve eig_banded by @jorenham in #2059linalg: improve eigh_tridiagonal by @jorenham in #2060linalg: improve eigh by @jorenham in #2061linalg: improve eig by @jorenham in #2062linalg: improve qr by @jorenham in #2063linalg: improve norm by @jorenham in #2064linalg: improve rq by @jorenham in #2065special.factorial: improved return dtype accuracy by @jorenham in #2080special.factorial2: improved return dtype accuracy by @jorenham in #2081special.factorialk: improved return dtype accuracy by @jorenham in #2082linalg: improve return dtypes of qz and ordqz by @jorenham in #2089linalg : improve return dtypes of ldl by @Aniketsy in #2029linalg.solve_sylvester: improve return dtypes by @jorenham in #2091linalg.solve_{continuous,discrete}_lyapunov: improve return dtypes by @jorenham in #2092linalg.solve_{continuous,discrete}_are: improve return dtypes by @jorenham in #2094sparse.csgraph.laplacian: improve return dtypes by @jorenham in #2095stats.mstats.tmean: improve shape-typing and return dtypes by @jorenham in #2098stats.mstats.spearmanr: improve shape-typing and return dtypes by @jorenham in #2099stats.mstats.normaltest: improve shape-typing and return dtypes by @jorenham in #2100stats.mstats.moment: improve shape-typing and return dtypes by @jorenham in #2101stats.mstats.{variation,skew,kurtosis}: improve shape-typing and return dtypes by @jorenham in #2102stats.mstats.trim: improve shape-typing and return dtypes by @jorenham in #2103stats.mstats.trimmed_mean: improve shape-typing and return dtypes by @jorenham in #2104stats.mstats.describe: improve shape-typing and return dtypes by @jorenham in #2105stats.mstats.linregress: improve shape-typing and return dtypes by @jorenham in #2106stats.mstats.{theil,siegel,sen_seasonal_}slopes: improve result types by @jorenham in #2107stats.mstats.hdquantiles: improve shape-typing and return dtypes by @jorenham in #2117stats.mstats.ks_2samp: improve shape-typing and return dtypes by @jorenham in #2118stats.mstats.friedmanchisquare: improve input signature by @jorenham in #2120stats.mstats.trimboth: improve shape-typing and return dtypes by @jorenham in #2123stats.mstats.hdmedian: improve shape-typing and return dtypes by @jorenham in #2124stats.mstats.trimmed_mean_ci: improve shape-typing and return dtypes by @jorenham in #2125stats.mstats.mjci: improve shape-typing and return dtypes by @jorenham in #2126stats.mstats.mquantiles_cimj: improve shape-typing and return dtypes by @jorenham in #2127stats.mstats.median_cihs: improve shape-typing and return dtypes by @jorenham in #2128stats.mstats.idealfourths: improve shape-typing and return dtypes by @jorenham in #2129stats.mstats.compare_medians_ms: improve shape-typing and return dtypes by @jorenham in #2130stats.mstats.hdquantiles_sd: improve shape-typing by @jorenham in #2131stats.mstats.trimtail: improve shape-typing and return dtypes by @jorenham in #2132stats.mstats.kstest: improve shape-typing and return dtypes by @jorenham in #2133stats.mstats.winsorize: improve shape-typing and return dtypes by @jorenham in #2135stats.mstats.rankdata: improve shape-typing and return dtypes by @jorenham in #2136stats.mstats.trimmed_std: improve shape-typing and return dtypes by @jorenham in #2138stats.mstats.trimmed_stde: improve shape-typing and return dtypes by @jorenham in #2140stats.mstats.tvar: improved input signature and return type by @jorenham in #2141stats.mstats.tsem: improve return type by @jorenham in #2142stats.mstats.{trima,trimr}: improve shape-typing and return dtypes by @jorenham in #2143stats.mstats.ks_1samp: improve shape-typing and return dtypes by @jorenham in #2144stats.mstats.brunnermunzel: also accept complex input by @jorenham in #2145stats.mstats.{tmin,tmax}: improve shape-typing and return dtypes by @jorenham in #2146stats.mstats.sem: improve shape-typing and return dtypes by @jorenham in #2148stats.mstats.plotting_positions: improve shape-typing by @jorenham in #2149stats.mstats.obrientransform: improve shape-typing by @jorenham in #2150stats.mstats.scoreatpercentile: improve shape-typing and return dtypes by @jorenham in #2151stats.mstats.msign: improve shape-typing and return dtypes by @jorenham in #2152stats.mstats.find_repeats: improve shape-typing by @jorenham in #2153stats.mstats.argstoarray: improve shape-typing by @jorenham in #2154spatial.KDTree.query: shape-dependent return types by @jorenham in #2159stats.sampling.RatioUniforms: fix pdf callable and rvs shape-typing by @jorenham in #2166interpolate.interp1d (legacy): generic dtype parameter by @jorenham in #2171ndimage.label: narrow return dtype when output=None (default) by @jorenham in #2173stats: improve .rvs() return type of simple continuous distributions by @jorenham in #2174stats.wilcoxon: improve return types by @jorenham in #2175linalg.polar: improve return dtypes by @jorenham in #2196stats.ecdf: longdouble support by @jorenham in #2197stats.yeojohnson: longdouble support by @jorenham in #2198linalg._decomp_qr: improve return dtypes by @jorenham in #2202sparse.kron: batching support by @jorenham in #2203sparse: fmt-dependent random_array/random return types by @jorenham in #2208integrate.solve_ivp: gradual t_events and y_events union type by @jorenham in #2213optimize.fsolve: mark full_output=True infodict njev key as not required by @jorenham in #2223stats.bws_test: improved shape-typing and dtype propagation by @jorenham in #2238stats.fit: narrow FitResult.params tuple element types by @jorenham in #2245linalg.lapack: complete type coverage by @jorenham in #2262linalg.blas: complete type coverage by @jorenham in #2263linalg.get_lapack_funcs: overloads for single/multiple names by @jorenham in #2270optimize.minimize: add more method-dependent result keys by @jorenham in #2272optimize.root: add more method-dependent result keys by @jorenham in #2273differentiate: improve return type for vectorized functions by @jorenham in #2278sparse.csgraph.maximum_bipartite_matching: narrow the return dtype by @jorenham in #2283sparse.linalg.LaplacianNd.tosparse: improved return type by @jorenham in #2284spatial.distance.seuclidean: narrow return type to np.float64 by @jorenham in #2289cluster.hierarchy.dendrogram: "leaves" is never None by @jorenham in #2290linalg: basic solvers propagate f32 and c64 by @jorenham in #2297io: fix netcdf_variable.getValue return type by @jorenham in #1990io.MatFile5Reader: fix .read_file_header().__header__ and .get_variables() return types by @jorenham in #1991stats.sampling.FastGeneratorInversion: fix .qrvs() return type for size=(1,) by @jorenham in #1992stats.multivariate_normal: fix [log]{pdf,.cdf} shape-typing by @jorenham in #1994sparse: fix .multiply method by @jorenham in #1995ndimage: labeled_comprehension ignores out_dtype when index=None; reject labels=None with non-Note truncated.
This scipy-stubs release fixes many bugs and comes with several improvements. It targets SciPy 1.18.0 and supports Python 3.12-3.14, NumPy 2.0-2.7, an
This scipy-stubs release fixes many bugs and comes with several improvements. It targets SciPy 1.18.0 and supports Python 3.12-3.14, NumPy 2.0-2.7, and optype 0.15-0.19.
linalg.[cumulative_]simpson: update dtype promotion behavior (changed in 1.18) by @jorenham in #1956stats.mannwhitneyu: fix dtype promotion and .pvalue dtype propagation by @jorenham in #1976sparse.linalg.svds: narrow dtype of the returned singular values array by @jorenham in #1879signal: improve shape-typing for peak-finding functions by @jorenham in #1891stats: emulate property overloading on the new distribution parameters by @jorenham in #1920ndimage: fourier_* should reject float16 intput by @Aniketsy in #1929interpolate: shape-typed generic polynomial interpolation classes by @jorenham in #1966stats.directional_stats: fix and improve shape-typing and dtype propagation by @jorenham in #1969sparse: fix & improve sparse array/matrix division / addition by @jorenham in #1970optimize.NonlinearConstraint: parametrized lb/ub and keep_feasible by @jorenham in #1971linalg.eig: fix output dtypes and output sizes by @jorenham in #1874io.loadmat: fix incorrect value type of the returned dict by @jorenham in #1875stats: fix discrete distribution .rvs() output dtype by @jorenham in #1876stats: fix axis default for {skew,kurtosis,normal}test by @jorenham in #1877stats: fix axis default for tmean by @jorenham in #1878sparse.linalg.svds: add missing overloads for return_singular_vectors by @jorenham in #1879spatial.distance.jaccard: accept floating weights by @jorenham in #1880interpolate.NdPPoly.integrate: accept any-length tuple for ranges by @jorenham in #1883signal.{c,q}spline2d: fix output shape-types (1d -> 2d) by @jorenham in #1884optimize.{basinhopping,dual_annealing}: fix result .message type (str -> list[str]) by @jorenham in #1887optimize.shgo: fix result xl and funl types by @jorenham in #1888interpolate.RegularGridInterpolator: fix dtype of grid arrays (always float64) by @jorenham in #1889signal.peak_widths: fix dtype of 2nd output array by @jorenham in #1890special: fix h{1,2}vp return type for Nd input by @jorenham in #1892cluster.hierarchy.dendogram: fix result dict {i,d}coord list element types (int -> float) by @jorenham in #1893linalg.solve: fix output shape-type for 1d b by @jorenham in #1894stats.expectile: add missing overloads for axis and keepdims by @jorenham in #1895stats.quantile: fix axis / keepdims overloads by @jorenham in #1896optimize.lsq_linear: fix result attr typo (const -> cost) by @jorenham in #1898interpolate.interp1d: accept complex y by @jorenham in #1900odr: add missing iwork attribute to Output by @Aniketsy in #1899interpolate.[bi]splev: accept scalars by @jorenham in #1901stats.{cum,rel}freq: fix type of result binsize and extrapoints by @jorenham in #1902odr: fix typo inc_condnum to inv_condnum in _odrpack.pyi by @Aniketsy in #1904linalg.block_diag: accept 2d input by @jorenham in #1905stats.mstats: fix count_tied_groups return value dtype by @Aniketsy in #1903interpolate.splint: fix output types for tck: BSpline and full_output=True with 2d tck by @jorenham in #1906interpolate: univariate/bivariate spline shape-typing fixes by @jorenham in #1907spatial.geometric_slerp: 1d output for scalar t by @jorenham in #1908signal: fix z dtype for IIR design output="zpk" by @jorenham in #1909signal: fix residue[z] / invres[z] float64/complex128 overloads by @jorenham in #1910stats: fix several confidence_interval() field type by @jorenham in #1911sparse: fix lil_array.{__iter__, getrow} output types by @jorenham in #1912sparse: fix sparse multiplication output formats by @jorenham in #1913sparse: fix sparse __pow__/power output formats by @jorenham in #1914sparse: fix spmatrix.getcol output formats by @jorenham in #1915sparse: fix kron output formats when format=None by @jorenham in #1916sparse.linalg: fix inv/expm/matrix_power output formats and dtypes by @jorenham in #1917sparse.linalg: fix spsolve sparse format propagation source by @jorenham in #1918stats.Binomial: remove spurious second type parameter by @jorenham in #1919stats: fix dtype of sommersd and median_test result class .table by @jorenham in #1921stats: fix bootstrap/permutation_test off-by-one rank by @jorenham in #1922special.legendre_p_all: fix output rank for scalar input by @jorenham in #1923stats: fix ttest_{1samp,rel} result .df dtype by @jorenham in #1925linalg.sqrtm: real input can return complex output by @jorenham in #1927stats.wilcoxon: add missing zstatistic result class attribute by @jorenham in #1930special.lqmn: scalar z should return 2d arrays by @jorenham in #1931optimize.minimize: fix returned jac type for method="trust-constr" by @jorenham in #1932stats: fix multivariate .rvs method signatures by @jorenham in #1934stats: fix [inv]wishart.{mean,var,mode} by @jorenham in #1935spatial: fix {ConvexHull,Delaunay}.vertices by @jorenham in #1936signal.detrend: fix output dtypes for float16 and complex64 input by @jorenham in #1937spatal.HalfspaceIntersection: fix dual_vertices dtype by @jorenham in #1938linalg.pinv: fix returned rank type (int -> np.int_) by @jorenham in #1939sparse.csgraph.shortest_path: fix return shape-type by @jorenham in #1941signal.place_poles: fix return dtypes by @jorenham in #1945signal: fix lsim/step/impulse shape-type by @jorenham in #1946integrate: fix fixed_quad and quad_vec return type for >1d integrands by @jorenham in #1950linalg.ordqz: fix 3rd return dtype by @jorenham in #1951datasets: restrict face(gray=1) to 3-D overload to match gray is True by @Aniketsy in #1949stats.sobol_indices: fix shape-type for single-output func by @jorenham in #1952optimize: correct nnls residual type to float, not np.float64 by @Aniketsy in #1953signal: gauss_spline should propagate float16/float32, not promote to float64 by @Aniketsy in #1954linalg.lstsq: fix residues shape-types and rank dtype by @jorenham in #1955interpolate.interp1d: fix fill_value type by @jorenham in #1957optimize.least_squares: fix result class activate_mask dtype by @jorenham in #1958differentiate.hessian: fix result class nfev dtype by @jorenham in #1959sparse.linalg.SuperLU: fix perm_{r,c} dtypes by @jorenham in #1960optimize.rosen*: fix result dtype and batching support by @jorenham in #1961special.polygamma: fix shape-typing for scalar input by @jorenham in #1962special: fix swapped ufunc types and ntypes for 1->2 ufuncs by @jorenham in #1963ndimage: fix labeled_comprehension return type for index=None/scalar by @Aniketsy in #1964stats: fix false_discovery_control(0.05) output should be scalar, not Nd by @Aniketsy in #1965interpolate: shape-typed generic polynomial interpolation classes by @jorenham in #1966optimize.fixed_point: scalar method="del2" input should return 0d array by @jorenham in #1968stats.directional_stats: fix and improve shape-typing and dtype propagation by @jorenham in #1969sparse: fix & improve sparse array/matrix division / addition by @jorenham in #1970optimize.NonlinearConstraint: parametrized lb/ub and keep_feasible by @jorenham in #1971sparse: fix shape-typing issues for .toarray, .nonzero, and .mean by @jorenham in #1972sparse.linalg.aslinearoperator: fix integer dtype propagation by @jorenham in #1973io.FortranFile.read_record: fix dtype and dtype tuple propagation by @jorenham in #1974sparse.csgraph: fix return dtype propagation by @jorenham in #1975stats.mannwhitneyu: fix dtype promotion and .pvalue dtype propagation by @jorenham in #1976stats.odds_ratio: fix .confidence_interval() field type by @jorenham in #1977stats.binned_statistic[_2d]: fix .binnumber shape-type and gradual .statistic scalar-type by @jorenham in #1978ndimage.find_objects: fix return type (include None in value type) by @jorenham in #1979sparse: add tests to increase scipy.sparse coverage by @Aniketsy in #1640signal: add test coverage for whittaker_henderson by @Aniketsy in #1967scipy-stubs/*.{py,pyi} by @jorenham in #1881ty and zuban errors in scripts/ by @jorenham in #1942scripts/ in CI by Note truncated.
linalg.bandwidth : sync with upstream (backwards incompatible type change) by @jorenham in #1624
This is the first scipy-stubs release that supports SciPy 1.18 (>=1.18.0rc2, <1.19) and supports Python 3.12-3.14, NumPy 2.0-2.7, and optype 0.14-0.18. If you encounter any regressions or other issues, please report them at https://github.com/scipy/scipy-stubs/issues.
_lib: SciPy 1.18 changes by @jorenham in #1616cluster: sync with upstream by @jorenham in #1618differentiate.derivative: sync with upstream by @jorenham in #1619integrate: sync with upstream by @jorenham in #1621interpolate: sync with upstream by @jorenham in #1622io: sync with upstream by @jorenham in #1623linalg.bandwidth: sync with upstream (backwards incompatible type change) by @jorenham in #1624linalg: sync with upstream by @jorenham in #1626optimize: sync with upstream by @jorenham in #1627signal: sync with upstream by @jorenham in #1628sparse.linalg.LinearOperator: new rdot method by @jorenham in #1629sparse.linalg.LinearOperator: new xp=None constructor parameters by @jorenham in #1630sparse: sync with upstream by @jorenham in #1631spatial.transform: upstream changes by @jorenham in #1632special._ufunc_tools: new private module by @jorenham in #1633interpolate: private _regrid module stubs by @jorenham in #1638stats: new estimated_cdf function by @jorenham in #1642stats: sync with upstream (private API) by @jorenham in #1645stats.ContinuousDistribution.lmoment: new method by @jorenham in #1646stats: new axis, keepdims, and nan_policy keyword parameters by @jorenham in #1647UP040 ruff errors by @jorenham in #1650UP046 ruff errors by @jorenham in #1651UP047 ruff errors by @jorenham in #1654spatial: deprecate minkowski_distance[_p] and distance_matrix by @jorenham in #1656spatial: deprecate tsearch by @jorenham in #1658interpolate: deprecate pade, lagrange, approximate_taylor_polynomial by @jorenham in #1659io: deprecate spmatrix parameter default by @jorenham in #1660stats.contingency.crosstab: return coo_array instead of coo_matrix by @jorenham in #1663stats.obrientransform: change return type to tuple by @jorenham in #1664stats.mannwhitneyu: new zstatistic return attribute by @jorenham in #1665sparse.linalg.minres: support dtype complex by @jorenham in #1666spatial.KDTree.sparse_distance_matrix: sparse array return types by @jorenham in #1667spatial.distance.{minkowski,euclidian,sqeuclidean}: N-d support by @jorenham in #1668sparse.linalg.LinearOperator: N-d support by @jorenham in #1669stats.ansari: accept PermutationMethod instance for method by @jorenham in #1674spatial.transform.Rotation.approx_equal: return np.bool if both rotations are single. by @jorenham in #1675linalg.{blas,lapack}: add HAS_{I}LP64 constants by @jorenham in #1676interpolate.NearestNDInterpolator.tree: narrow attribute type to KDTree by @jorenham in #1677special: new private ufuncs by @jorenham in #1678_lib._array_api: new internal helper functions by @jorenham in #1679linalg.cho_factor: change 2nd return type to bool array by @jorenham in #1680interpolate.splint: change 2nd return type to None (full_output=True) by @jorenham in #1681sparse.{bsr,dok}_{array,matrix}.count_nonzero: support for axis by @jorenham in #1682stats.binomtest: batching support (undocumented) by @jorenham in #1683_lib: remove np_long and np_ulong re-exports by @jorenham in #1709stats: Narrow RatioUniforms.rvs return type to array of np.float64 by @loicdiridollou in #1685stats.make_distribution: shape-typing support by @jorenham in #1696io.matlab._mio: accept FileLike by @myzhang1029 in #1716signal: fft_mode-dependent ShortTimeFFT.istft sctype by @jorenham in #1731stats: vonmises_fisher kappa is float, not int by @feinedsquirrel in #1655sparse.csgraph: add missing overload for connected_components(return_labels=False) by @jorenham in #1672stats.make_distribution: discrete distribution support by @jorenham in #1692stats.make_distribution: shape-typing support by @jorenham in #1696stats: fix kappa type and rvs size in vonmises_fisher and multivariate_normal stubs by @Aniketsy in #1714signal: fft_mode-dependent ShortTimeFFT.istft sctype by @jorenham in #1731optimize: missing minimizer_kwargs items for global optimizers by @jorenham in #1732optimize: fix global optimizer func type when jac=True by @jorenham in #1735stats: add test for qmc.update_discrepancy by @Aniketsy in #1643stats: add test for sampling.RatioUniforms by @Aniketsy in #1644sparse: add tests for scipy.sparse.csgraph by @Aniketsy in #1641--python-version=3.12 by @jorenham in #1617fft: ignore stubtest errors for bundled _duccfft by @jorenham in #1620np.bool_ with np.bool by @Aniketsy in #1634_lib: private _fpumode module stubs by @jorenham in #1639cvxpy and pytensor to the project selector by @jorenham in #1661optype by @jorenham in #1701scripts/check_version_literals now that stubtest also does this by @jorenham in #1710pyrefly to 1.1.0 by @jorenham in #1724spatial.transform: override RigidTransform.__reduce__ by @jorenham in #1725pyrefly coverage check in CI by @jorenham in #1727Full Changelog: v1.17.1.5...v1.18.0.0
This release targets SciPy 1.17.1 and supports Python 3.11-3.14, NumPy 1.26-2.4, and optype 0.14-0.18.
This release targets SciPy 1.17.1 and supports Python 3.11-3.14, NumPy 1.26-2.4, and optype 0.14-0.18.
signal: improved LTI system class constructors by @jorenham in #1563signal: improved sepfir2d dtype overloads and shape-typing by @jorenham in #1564stats: multinomial shape-typing by @jorenham in #1566stats: multivariate_normal shape-typing by @jorenham in #1568stats: Add missing 1D diagonal support for matrix_normal rowcov/colcov by @Aniketsy in #1569sparse.csgraph: add missing callable return type for laplacian with form="function" by @Aniketsy in #1583special: fix [{assoc,sph}_]legendre[_all] and sph_harm_y[_all] by @jorenham in #1606signal: improve type-checker compatibility of LTI class constructors by @jorenham in #1556special: improve softplus compatibility with older numpy versions by @jorenham in #1582signal: add tests for cspline1d, cspline1d_eval, cspline2d by @fbourgey in #1548signal: add tests for gauss_spline, spline_filter, qspline1d, qspline1d_eval, qspline2d, symiirorder1, symiirorder2 by @fbourgey in #1551signal: add tests for BadCoefficients, StateSpace, TransferFunction, ZerosPolesGain by @fbourgey in #1552signal: add tests for get_window, place_poles, sepfir2d by @fbourgey in #1554stats: add test for binned_statistic by @Aniketsy in #1565stats: add tests for scipy.stats.mstats.tmax, tmean, and tmin by @Aniketsy in #1570sparse : add tests for csgraph_traversal by @Aniketsy in #1573special: add more tests by @fbourgey in #1577special: more tests by @fbourgey in #1581special: tests for multigammaln, geterr, seterr, errstate, ellip_harm, ellip_normal, legendre_p by @fbourgey in #1584special: tests for legendre_p_all , assoc_legendre_p , assoc_legendre_p_all , sph_legendre_p , sph_legendre_p_all, sph_harm_y , sph_harm_y_all by @fbourgey in #1590scipy.stats.chi2_contingency in test coverage script by @Aniketsy in #1567lefthook.yml -> .lefthook.yml by @jorenham in #1607update_dprint workflow by @jorenham in #1609Full Changelog: v1.17.1.4...v1.17.1.5
This release targets SciPy 1.17.1 and supports Python 3.11-3.14, NumPy 1.26-2.4, and optype 0.14-0.18.
This release targets SciPy 1.17.1 and supports Python 3.11-3.14, NumPy 1.26-2.4, and optype 0.14-0.18.
approx_fprime and newton to accept functions returning n-D arrays in scipy.optimizescipy.stats and FFT shifting functions in scipy.fftsignal: stub the private _signal_api module by @jorenham in #1492signal: stub the private _delegators module by @jorenham in #1493stats: improved ks_2samp shape-typing by @jorenham in #1494stats: improved brunnermunzel shape-typing by @jorenham in #1497stats: improved f_oneway shape-typing by @jorenham in #1499stats: improved kruskal shape-typing by @jorenham in #1500stats: improved sigmaclip dtypes by @jorenham in #1501fftpack: improve return dtypes for real transforms by @Deshan-5 in #1509signal: improve lp2{lp,hp,bp,bs}[_zpk] and bilinear_zpk by @Aniketsy in #1523stats.dunnett: support for longdouble dtypes by @Aniketsy in #1528fft: fftshift and ifftshift shape-typing support by @Aniketsy in #1532_lib._uarray._backend: fix wrap_single_convertor[_instance] overloads by @jorenham in #1481optimize: add vector-valued approx_fprime overload by @fbourgey in #1530fft: [i]fftshift dtype preservation for int and bool inputs by @Aniketsy in #1533optimize: generalize newton overloads to ND arrays by @fbourgey in #1534optype type aliases with simpler annotations by @jorenham in #1490optype.numpy.ToBool annotations by @jorenham in #1491optimize.Bounds: improve type-checker compatibility by @jorenham in #1507actions/checkout by @jorenham in #1536uvx command by @jorenham in #1541UV_LOCKED and disable UV_PRERELEASE by @jorenham in #1542--pedantic audits in update_dprint.yml by @jorenham in #1544Full Changelog: v1.17.1.3...v1.17.1.4
This release targets SciPy 1.17.1 and supports Python 3.11-3.14, NumPy 1.26-2.4, and optype 0.14-0.18.
This release targets SciPy 1.17.1 and supports Python 3.11-3.14, NumPy 1.26-2.4, and optype 0.14-0.18.
scipy.sparse arrays and matrices for certain slicing and multi-indexing operations (see #1454)scipy.io, scipy.linalg, scipy.ndimage, scipy.signal, and scipy.stats.ndimage: improved morphology functions by @jorenham in #1458linalg.interpolative: improved shape-typing and dtypes by @jorenham in #1459linalg: improved svdvals and subspace_angles shape-types and dtypes by @jorenham in #1460ndimage: improved interpolation functions by @jorenham in #1461stats: improve {g,h,p}mean by adding 3*26 overloads by @jorenham in #1463stats: improved iqr dtypes by @jorenham in #1464stats: improved zmap dtypes and shape-typing by @jorenham in #1465io: improved MMFile.{read,write} signatures by @jorenham in #1467signal: minor improvements to some filter design functions by @jorenham in #1475signal: improved LTI conversion functions dtypes (_lti_conversion) by @jorenham in #1477sparse: fix sparse array/matrix sliced index assignment by @jorenham in #1456sparse: fix sparse array/matrix mixed multi-index assignment by @jorenham in #1457stats: test Covariance by @jorenham in #1468stats: test chatterjeexi and spearmanrho by @jorenham in #1469stats: test remaining resampling functions and classes by @jorenham in #1470stats: test dirichlet_multinomial and matrix_t by @jorenham in #1471stats: test rv_histogram by @jorenham in #1472stats: test new_distributions by @jorenham in #1473stats: test abs and Mixture by @jorenham in #1474signal: test the remaining filter design functions (_filter_design) by @jorenham in #1476signal: test detrend, residue[z], and invres[z] by @jorenham in #1478signal: test remaining waveform functions by @jorenham in #1479#pyrefly ignore by @jorenham in #1455Full Changelog: v1.17.1.2...v1.17.1.3
This release targets SciPy 1.17.1 and supports Python 3.11-3.14, NumPy 1.26-2.4, and optype 0.14-0.18.
This release targets SciPy 1.17.1 and supports Python 3.11-3.14, NumPy 1.26-2.4, and optype 0.14-0.18.
scipy.spatial, scipy.linalg, and scipy.stats.scipy.sparse array indexing issue (see #1412)dimensions=None in scipy.io.netcdf_file (see #1423)spatial: improved Qhull classes by @jorenham in #1420spatial: improved procrustes shape-typing by @jorenham in #1421linalg: improved Cholesky function shape-types and dtypes by @jorenham in #1426linalg: improved Schur function shape-types and dtypes by @jorenham in #1428linalg: improved LU function shape-types and dtypes by @jorenham in #1431spatial.distance: improved shape-typing and dtypes by @jorenham in #1443stats.contingency: improved shape-typing and dtypes by @jorenham in #1444stats: improve many functions (in _stats_py) by @jorenham in #1448sparse: fix sparse array indexing by @jorenham in #1413spatial: fix geometric_slerp return shape-type by @jorenham in #1418io: accept dimensions=None in netcdf_file.createDimension by @jorenham in #1424io: include None in the netcdf_file.dimensions value type by @jorenham in #1425spatial: increase type-test coverage (_kdtree) by @jorenham in #1417spatial: test plotting functions by @jorenham in #1419spatial: test SphericalVoronoi by @jorenham in #1422interpolate: test make_*_spline by @jorenham in #1433interpolate: test NdBSpline by @jorenham in #1434interpolate: test legacy B-spline functions by @jorenham in #1435interpolate: test spline creation functions by @jorenham in #1436interpolate: test *BivariateSpline by @jorenham in #1437interpolate: test approximate_taylor_polynomial by @jorenham in #1438interpolate: test interpn by @jorenham in #1439interpolate: test Rbf by @jorenham in #1440interpolate: fix _fitpack_py test imports by @jorenham in #1441interpolate: test the remaining classes and functions (_interpolate module) by @jorenham in #1442ndimage: remove redundant typevar default in _measurements by @jorenham in #1447Full Changelog: v1.17.1.1...v1.17.1.2
Your coding agent can read these notes before it upgrades. Set up the MCP server →