SciPy 1.15.0 版本说明

SciPy 1.15.0 是 6 个月辛勤工作的结晶。它包含了许多新功能、大量的错误修复、改进的测试覆盖率和更好的文档。此版本中有许多弃用和 API 更改，下面进行了记录。强烈建议所有用户升级到此版本，因为它修复了大量错误并进行了优化。在升级之前，我们建议用户检查自己的代码是否使用了已弃用的 SciPy 功能（为此，请使用 python -Wd 运行代码并检查 DeprecationWarning）。我们的开发重点将转移到 1.15.x 分支的错误修复版本，以及主分支的新功能添加。

此版本需要 Python 3.10-3.13 和 NumPy 1.23.5 或更高版本。

此版本的亮点

• 稀疏数组现在对一维和二维数组完全可用。我们建议所有新代码使用稀疏数组而不是稀疏矩阵，并建议开发人员开始将现有代码从稀疏矩阵迁移到稀疏数组：从 spmatrix 迁移到 sparray。 sparse.linalg 和 sparse.csgraph 都支持稀疏矩阵或稀疏数组，并且内部使用稀疏数组。

• 稀疏数组现在为 COO 格式的多维数组提供了基本支持，包括 add、subtract、reshape、transpose、matmul、dot、tensordot 等。未来版本将提供更多功能。

• 对自由线程 Python 3.13 的初步支持。

• SciPy scipy.stats 中的新概率分布功能可用于提高现有连续分布的速度和准确性，并执行新的概率计算。

• 多项新功能支持与符合 Python Array API 标准的输入进行向量化计算（请参阅下面的“Array API 标准支持”）

  • SciPy scipy.differentiate 是一个新的顶级子模块，用于精确估计黑盒函数的导数。

  • SciPy scipy.optimize.elementwise 包含用于单变量函数求根和最小化的新函数。

  • SciPy scipy.integrate 提供了新的函数 cubature、tanhsinh 和 nsum，分别用于多变量积分、单变量积分和单变量级数求和。

• SciPy scipy.interpolate.AAA 添加了 AAA 算法，用于实函数或复函数的重心有理逼近。

• SciPy scipy.special 添加了新函数，提供了改进的勒让德函数实现，具有更一致的接口。

新功能

scipy.differentiate 介绍

新的 SciPy scipy.differentiate 子包包含用于精确估计黑盒函数导数的函数。

• 使用 SciPy scipy.differentiate.derivative 来计算标量输入、标量输出函数的一阶导数。

• 使用 SciPy scipy.differentiate.jacobian 来计算向量输入、向量输出函数的任意偏一阶导数。

• 使用 SciPy scipy.differentiate.hessian 来计算向量输入、标量输出函数的任意偏二阶导数。

所有函数都使用高阶有限差分规则，并带有自适应（实数）步长。为了便于批量计算，这些函数是向量化的，并且除了 NumPy 之外，还支持多种符合 Array API 标准的数组库（请参阅下面的“Array API 标准支持”）。

scipy.integrate 改进

• 新的 SciPy scipy.integrate.cubature 函数支持多维积分，并且支持近似具有一个或多个无限积分限的积分。

• SciPy scipy.integrate.tanhsinh 现在已公开可用，允许使用 tanh-sinh-quadrature 评估收敛积分。

• SciPy scipy.integrate.nsum 计算有限和无限级数及其对数。

• SciPy scipy.integrate.lebedev_rule 计算球面上积分的横坐标和权重。

• QUADPACK Fortran77 包已移植到 C。

scipy.interpolate 改进

• SciPy scipy.interpolate.AAA 添加了 AAA 算法，用于实函数或复函数的重心有理逼近。

• SciPy scipy.interpolate.FloaterHormannInterpolator 添加了重心有理插值。

• 新函数 scipy.interpolate.make_splrep 和 scipy.interpolate.make_splprep 实现平滑样条的构建。算法内容等同于 FITPACK（splrep 和 splprep 函数，以及 *UnivariateSpline 类），并且用户 API 与 make_interp_spline 一致：这些函数接收数据数组并返回一个 scipy.interpolate.BSpline 实例。

• 新的生成器函数 scipy.interpolate.generate_knots 实现 FITPACK 策略，用于在给定平滑度参数 s 的情况下选择平滑样条的节点。该函数公开了 splrep 和 *UnivariateSpline 使用的节点选择的内部逻辑。

scipy.linalg 改进

• SciPy scipy.linalg.interpolative 的 Fortran77 代码已移植到 Cython。

• SciPy scipy.linalg.solve 支持 assume_a 参数的多个新值，从而能够更快地计算对角、三对角、带状和三角矩阵。此外，当 assume_a 未指定时，该函数现在会自动检测并利用对角、三对角和三角结构。

• SciPy scipy.linalg 矩阵创建函数（scipy.linalg.circulant、scipy.linalg.companion、scipy.linalg.convolution_matrix、scipy.linalg.fiedler、scipy.linalg.fiedler_companion 和 scipy.linalg.leslie）现在支持批量矩阵创建。

• SciPy scipy.linalg.funm 速度更快。

• SciPy scipy.linalg.orthogonal_procrustes 现在支持复数输入。

• 在 SciPy scipy.linalg.lapack 中添加了以下 LAPACK 例程的包装器：?lantr、?sytrs、?hetrs、?trcon 和 ?gtcon。

• SciPy scipy.linalg.expm 已用 C 重写。

• SciPy scipy.linalg.null_space 现在接受新的参数 overwrite_a、check_finite 和 lapack_driver。

• id_dist Fortran 代码已用 Cython 重写。

scipy.ndimage 改进

• 几个额外的滤波函数现在支持 axes 参数，该参数指定输入过滤要执行的轴。这些包括 correlate、convolve、generic_laplace、laplace、gaussian_laplace、derivative2、generic_gradient_magnitude、gaussian_gradient_magnitude 和 generic_filter。

• 二值和灰度形态学函数现在支持 axes 参数，该参数指定输入过滤要执行的轴。

• SciPy scipy.ndimage.rank_filter 的时间复杂度已从 n 提高到 log(n)。

scipy.optimize 改进

• 嵌入的 HiGHS 库已从 1.4.0 升级到 1.8.0，为求解器带来了准确性和性能的提升。

• MINPACK Fortran77 包已移植到 C。

• L-BFGS-B Fortran77 包已移植到 C。

• 新的 SciPy scipy.optimize.elementwise 命名空间包含用于单变量函数求根和最小化的函数 bracket_root、find_root、bracket_minimum 和 find_minimum。为了便于批量计算，这些函数是向量化的，并且除了 NumPy 之外，还支持多种符合 Array API 标准的数组库（请参阅下面的“Array API 标准支持”）。与现有函数（例如 SciPy scipy.optimize.root_scalar 和 scipy.optimize.minimize_scalar）相比，这些函数在与 NumPy 数组一起使用时可以提供超过 100 倍的速度提升，并且使用其他符合 Array API 标准的数组库（例如 CuPy）可以获得更大的性能提升。

• SciPy scipy.optimize.differential_evolution 现在支持更通用的 workers 用法，例如传递一个类似 map 的可调用对象。

• SciPy scipy.optimize.nnls 已用 Cython 重写。

• HessianUpdateStrategy 现在支持 __matmul__。

scipy.signal 改进

• 为 signal.chirp() 添加了复数值波形的功能。

• SciPy scipy.signal.lombscargle 增加了两个新参数 weights 和 floating_mean，允许对每个频率进行样本加权和独立移除未知 y 偏移。此外，normalize 参数包含一个新选项，用于返回幅度和相位的复数表示。

• 新的 SciPy scipy.signal.envelope 函数用于计算实值或复数值信号的包络。

scipy.sparse 改进

• 现在提供了一个 迁移指南，用于在您的代码/库中从 sparse.matrix 迁移到 sparse.array。

• 稀疏数组现在支持一维和二维数组的索引。因此，稀疏数组对一维和二维数组已完全可用。

• COO 格式的多维稀疏数组现在可以构造、重塑并用于基本算术运算。

• 新函数 sparse.linalg.is_sptriangular 和 sparse.linalg.spbandwidth 模仿现有的密集工具 linalg.is_triangular 和 linalg.bandwidth。

• sparse.linalg 和 sparse.csgraph 现在支持稀疏数组。请注意，您的索引数组必须是 32 位。我们正在努力支持 64 位。

• 嵌入的 ARPACK 库已升级到 3.9.1 版本。

• COO、CSR、CSC 和 LIL 格式现在为 count_nonzero 支持 axis 参数。

• 当使用不兼容的数据类型（例如 float16）初始化时，稀疏数组和稀疏矩阵现在可能会引发错误。

• min、max、argmin 和 argmax 现在通过新的 explicit 参数支持仅在非零元素上进行计算。

• 新的函数 get_index_dtype 和 safely_cast_index_arrays 可用于简化 sparse 中的索引数组转换。

scipy.spatial 改进

• Rotation.concatenate 现在接受裸 Rotation 对象，并将返回其副本。

scipy.special 改进

• 提供了新的函数，改进了勒让德函数实现，并提供了更一致的接口。有关更多信息，请参阅各自的文档字符串。

  • scipy.special.legendre_p、scipy.special.legendre_p_all

  • scipy.special.assoc_legendre_p、scipy.special.assoc_legendre_p_all

  • scipy.special.sph_harm_y、scipy.special.sph_harm_y_all

  • scipy.special.sph_legendre_p、scipy.special.sph_legendre_p_all，

• 阶乘函数 special.{factorial,factorial2,factorialk} 现在通过传递 extend='complex' 的关键字参数，提供对复数域的扩展。这是选择加入的，因为它会更改负数输入的（默认返回 0）以及某些整数（对于 factorial2 和 factorialk；有关更多详细信息，请查看各自的文档字符串）的值。

• SciPy scipy.special.zeta 现在在复平面上定义了黎曼 zeta 函数。

• SciPy scipy.special.softplus 计算 softplus 函数

• 球形贝塞尔函数（scipy.special.spherical_jn、scipy.special.spherical_yn、scipy.special.spherical_in 和 scipy.special.spherical_kn）现在支持具有实数类型的负参数。

• 当求和中的一个元素的大小远大于其他元素时，SciPy scipy.special.logsumexp 现在可以保持精度。

• 几个函数的准确性得到了提高。

  • SciPy scipy.special.ncfdtr、scipy.special.nctdtr 和 scipy.special.gdtrib 在整个定义域内得到了改进。

  • 对于 b=1、较小的 x 和较小的 a 的情况，SciPy scipy.special.hyperu 得到了改进。

  • 在 p=0.5 的参数附近，SciPy scipy.special.logit 得到了改进。

  • 当 x/y 溢出、下溢或接近 1 时，SciPy scipy.special.rel_entr 得到了改进。

• 对于 sqrt(2)/2 < |x| < 1 的情况，SciPy scipy.special.ndtr 效率更高。

scipy.stats 改进

• 为单变量连续分布的实现添加了新的概率分布基础设施。与以前的基础设施相比，它在速度、准确性、内存和接口方面具有多项优势。有关教程，请参阅 随机变量转换指南。

  • 使用 SciPy scipy.stats.make_distribution 将现有连续分布（例如 SciPy scipy.stats.norm）与新基础设施进行对接。这可以提高现有分布的速度和准确性，特别是那些具有未被分布特定公式覆盖的方法的分布。

  • SciPy scipy.stats.Normal 和 scipy.stats.Uniform 是预定义的类，分别表示正态分布和均匀分布。它们的接口可能比 make_distribution 生成的接口更快、更方便。

  • SciPy scipy.stats.Mixture 可用于表示混合分布。

• SciPy scipy.stats.Normal、scipy.stats.Uniform 以及由 scipy.stats.make_distribution 返回的类的实例支持多项数学变换。

  • SciPy scipy.stats.truncate 用于截断支持域。

  • SciPy scipy.stats.order_statistic 用于给定数量的独立同分布随机变量的顺序统计量。

  • SciPy scipy.stats.abs、scipy.stats.exp 和 scipy.stats.log。例如，scipy.stats.abs(Normal()) 服从折叠正态分布，而 scipy.stats.exp(Normal()) 服从对数正态分布。

• 新的 SciPy scipy.stats.lmoment 计算样本 l-矩和 l-矩比。值得注意的是，这些样本估计量是无偏的。

• SciPy scipy.stats.chatterjeexi 计算 Xi 相关系数，它可以检测非线性依赖。该函数还执行样本之间独立性的假设检验。

• SciPy scipy.stats.wilcoxon 对默认 method='auto' 的方法解析逻辑进行了改进。用户提供的 method 的其他值现在在所有情况下都会被尊重，并且 approx 参数已重命名为 asymptotic，以与其他类似函数保持一致。（为了向后兼容，仍允许使用 approx。）

• 有几项新的概率分布

  • SciPy scipy.stats.dpareto_lognorm 代表双帕累托对数正态分布。

  • SciPy scipy.stats.landau 代表兰道分布。

  • SciPy scipy.stats.normal_inverse_gamma 代表正态-逆伽马分布。

  • SciPy scipy.stats.poisson_binom 代表泊松二项分布。

• 使用 SciPy scipy.stats.alexandergovern 和 scipy.stats.combine_pvalues 进行批量计算速度更快。

• SciPy scipy.stats.chisquare 添加了一个参数 sum_check。默认情况下，当期望频率和观察频率的总和不相等时，函数会引发错误；将 sum_check=False 设置为 False 会禁用此检查，以方便除 Pearson 卡方检验以外的其他假设检验。

• 几个分布方法的准确性得到了提高，包括

  • SciPy scipy.stats.nct 方法 pdf

  • SciPy scipy.stats.crystalball 方法 sf

  • SciPy scipy.stats.geom 方法 rvs

  • SciPy scipy.stats.cauchy 方法 logpdf、pdf、ppf 和 isf

  • 不覆盖这些方法通用实现的分布的 logcdf 和/或 logsf 方法，包括 SciPy scipy.stats.beta、scipy.stats.betaprime、scipy.stats.cauchy、scipy.stats.chi、scipy.stats.chi2、scipy.stats.exponweib、scipy.stats.gamma、scipy.stats.gompertz、scipy.stats.halflogistic、scipy.stats.hypsecant、scipy.stats.invgamma、scipy.stats.laplace、scipy.stats.levy、scipy.stats.loggamma、scipy.stats.maxwell、scipy.stats.nakagami 和 scipy.stats.t。

• SciPy scipy.stats.qmc.PoissonDisk 现在接受下界和上界参数 l_bounds 和 u_bounds。

• SciPy scipy.stats.fisher_exact 现在支持形状不是 (2, 2) 的二维表。

对 CPython 3.13 的自由线程支持（初步）

SciPy 1.15 版本初步支持 CPython 3.13 的自由线程构建。这允许 SciPy 功能与 Python 线程并行执行（请参阅 threading 标准库模块）。此支持是通过修复纯 Python 和 C/C++/Cython/Fortran 扩展模块中的大量线程安全问题实现的。PyPI 上提供了此版本的 wheel 文件；运行时需要 NumPy >=2.1.3。请注意，为自由线程解释器构建需要 Cython 3.1.0 的最新预发行版或 nightly 版本。

对自由线程 Python 的支持并不意味着 SciPy 完全线程安全。有关更多详细信息，请参阅 SciPy 中的线程安全。

如果您对自由线程 Python 感兴趣，例如因为您有一个基于多进程的工作流，并且您希望使用 Python 线程运行它，我们鼓励您进行测试和实验。如果您遇到由于 SciPy 引起的问题，请打开一个 issue，并先检查该 bug 是否也出现在“普通”非自由线程 CPython 3.13 版本中。许多线程 bug 也可能出现在释放 GIL 的代码中；禁用 GIL 只会使遇到线程 bug 变得更容易。

Array API 标准支持

在最近的 SciPy 版本中，在现有的子包中添加了对 NumPy 以外的数组库的实验性支持。请考虑通过设置环境变量 SCIPY_ARRAY_API=1 并提供 PyTorch、JAX、ndonnx 或 CuPy 数组作为数组参数来测试这些功能。为 SciPy 1.15.0 添加支持的功能包括：

• SciPy scipy.differentiate（新子包）中的所有函数

• SciPy scipy.optimize.elementwise（新命名空间）中的所有函数

• SciPy scipy.optimize.rosen、scipy.optimize.rosen_der 和 scipy.optimize.rosen_hess

• scipy.special.logsumexp

• scipy.integrate.trapezoid

• SciPy scipy.integrate.tanhsinh（新公开函数）

• SciPy scipy.integrate.cubature（新函数）

• SciPy scipy.integrate.nsum（新函数）

• SciPy scipy.special.chdtr、scipy.special.betainc 和 scipy.special.betaincc

• scipy.stats.boxcox_llf

• scipy.stats.differential_entropy

• SciPy scipy.stats.zmap、scipy.stats.zscore 和 scipy.stats.gzscore

• SciPy scipy.stats.tmean、scipy.stats.tvar、scipy.stats.tstd、scipy.stats.tsem、scipy.stats.tmin 和 scipy.stats.tmax

• SciPy scipy.stats.gmean、scipy.stats.hmean 和 scipy.stats.pmean

• scipy.stats.combine_pvalues

• SciPy scipy.stats.ttest_ind、scipy.stats.ttest_rel

• scipy.stats.directional_stats

• SciPy scipy.ndimage 函数现在将委托给 cupyx.scipy.ndimage，对于其他后端，将通过主机上的 NumPy 数组进行传输。

已弃用功能和未来更改

• 函数 scipy.linalg.interpolative.rand 和 scipy.linalg.interpolative.seed 已被弃用，并将在 SciPy 1.17.0 中移除。

• SciPy scipy.spatial.distance.cosine 和 scipy.spatial.distance.correlation 中的复数输入已被弃用，并将在 SciPy 1.17.0 中引发错误。

• scipy.spatial.distance.kulczynski1 和 scipy.spatial.distance.sokalmichener 已被弃用，并将在 SciPy 1.17.0 中移除。

• scipy.stats.find_repeats 已被弃用，并将在 SciPy 1.17.0 中移除。请使用 numpy.unique/numpy.unique_counts 代替。

• scipy.linalg.kron 已弃用，建议使用 numpy.kron。

• 在 SciPy 1.17.0 中，在 SciPy scipy.signal 的卷积/相关函数（scipy.signal.correlate、scipy.signal.convolve 和 scipy.signal.choose_conv_method）和滤波函数（scipy.signal.lfilter、scipy.signal.sosfilt）中使用对象数组和 longdouble 数组已被弃用，并将在 SciPy 1.17.0 中移除。

• SciPy scipy.stats.linregress 弃用了单参数用法；必须将两个变量指定为单独的参数。

• scipy.stats.trapz 已弃用，建议使用 SciPy scipy.stats.trapezoid。

• scipy.special.lpn 已弃用，建议使用 SciPy scipy.special.legendre_p_all。

• scipy.special.lpmn 和 scipy.special.clpmn 已弃用，建议使用 SciPy scipy.special.assoc_legendre_p_all。

• scipy.special.sph_harm 已弃用，建议使用 SciPy scipy.special.sph_harm_y。

• 从 SciPy 1.17.0 开始，传递给 SciPy scipy.linalg.toeplitz、scipy.linalg.matmul_toeplitz 或 scipy.linalg.solve_toeplitz 的多维 r 和 c 数组将被视为 1-D 系数的批处理。

• SciPy scipy.stats.ttest_ind 的 random_state 和 permutations 参数已弃用。请改用 method 来执行置换检验。

已过期弃用

• SciPy scipy.signal 中的小波函数已被移除。这包括 daub、qmf、cascade、morlet、morlet2、ricker 和 cwt。用户应改用 pywavelets。

• scipy.signal.cmplx_sort 已被移除。

• scipy.integrate.quadrature 和 scipy.integrate.romberg 已被移除，建议使用 SciPy scipy.integrate.quad。

• scipy.stats.rvs_ratio_uniforms 已被移除，建议使用 SciPy scipy.stats.sampling.RatioUniforms。

• 当 exact=True 时，SciPy scipy.special.factorial 现在对非整数标量引发错误。

• 当 initial 的值为 0 和 None 以外的值时，SciPy scipy.integrate.cumulative_trapezoid 现在会引发错误。

• 在 SciPy scipy.interpolate.Akima1DInterpolator 和 scipy.interpolate.PchipInterpolator 中，复数数据类型现在会引发错误。

• special.btdtr 和 special.btdtri 已被移除。

• special.factorialk 中 exact= 关键字参数的默认值已从 True 更改为 False。

• scipy.misc 子模块中的所有函数都已被移除。

向后不兼容的更改

• interpolate.BSpline.integrate 的输出现在始终是 numpy 数组。以前，对于一维样条，输出是 python 浮点数或 0D 数组，具体取决于 extrapolate 参数的值。

• SciPy scipy.stats.wilcoxon 现在尊重用户提供的 method 参数。以前，即使指定了 method='exact'，函数在某些情况下也会退回到 method='approx'。

• scipy.integrate.AccuracyWarning 已被移除，因为发出该警告的函数（scipy.integrate.quadrature 和 scipy.integrate.romberg）已被移除。

其他变更

• 一个单独的配套类型存根包 scipy-stubs 将随着 1.15.0 版本一起提供。 安装说明.

• SciPy scipy.stats.bootstrap 现在如果输入数组的形状不一致，会发出 FutureWarning。将数组广播到相同的批处理形状（即，除了 axis 参数指定的维度之外的所有维度）以避免警告。将来会自动执行广播。

• SciPy 支持 SPEC-7，该规范提出了一个 rng 参数来以标准方式控制伪随机数生成（PRNG），取代 seed 和 random_sate 等旧参数。在许多情况下，使用 rng 将改变函数的行为，除非该参数已经是 numpy.random.Generator 的实例。

  • 自 SciPy 1.15.0 起生效

    • 已向以下函数添加了 rng 参数： scipy.cluster.vq.kmeans、scipy.cluster.vq.kmeans2、scipy.interpolate.BarycentricInterpolator、scipy.interpolate.barycentric_interpolate、scipy.linalg.clarkson_woodruff_transform、scipy.optimize.basinhopping、scipy.optimize.differential_evolution、scipy.optimize.dual_annealing、scipy.optimize.check_grad、scipy.optimize.quadratic_assignment、scipy.sparse.random、scipy.sparse.random_array、scipy.sparse.rand、scipy.sparse.linalg.svds、scipy.spatial.transform.Rotation.random、scipy.spatial.distance.directed_hausdorff、scipy.stats.goodness_of_fit、scipy.stats.BootstrapMethod、scipy.stats.PermutationMethod、scipy.stats.bootstrap、scipy.stats.permutation_test、scipy.stats.dunnett，以及所有 scipy.stats.qmc 中需要消耗随机数的类，以及 scipy.stats.sobol_indices。

    • 当通过关键字参数传递时，rng 参数将遵循 SPEC 7 标准行为：在 np.random.default_rng 中进行标准化后使用。

    • 当通过位置参数或旧关键字参数传递时，该参数的行为将保持不变（目前）。

  • 计划在 1.17.0 版本开始，旧的参数将开始发出警告，并在 1.19.0 版本更改默认行为。

  • 在所有情况下，用户可以通过将 np.random.Generator 的实例作为关键字参数 rng 传递，来避免未来的中断。有关详细信息，请参阅 SPEC-7。

• SciPy 构建不再为 Fortran 代码添加 -std=legacy，除非使用 Gfortran。这可以避免与新的 Flang 和 AMD Fortran 编译器出现问题。对于其他编译器，可能会出现新的构建警告——如果出现，请提交 issue。

• scipy.signal.sosfreqz 已重命名为 scipy.signal.freqz_sos。新代码应使用新名称。旧名称保留为向后兼容的别名。

• 已在以下模块中进行了与 Python 3.13t 相关的线程安全性改进测试： scipy.special、scipy.spatial、scipy.sparse、scipy.interpolate。

作者 (提交)

• endolith (4)

• h-vetinari (62)

• a-drenaline (1) +

• Afleloup (1) +

• Ahmad Alkadri (1) +

• Luiz Eduardo Amaral (3) +

• Virgile Andreani (3)

• Isaac Alonso Asensio (2) +

• Matteo Bachetti (1) +

• Arash Badie-Modiri (1) +

• Arnaud Baguet (1) +

• Soutrik Bandyopadhyay (1) +

• Ankit Barik (1) +

• Christoph Baumgarten (1)

• Nickolai Belakovski (3)

• Krishan Bhasin (1) +

• Jake Bowhay (89)

• Michael Bratsch (2) +

• Matthew Brett (1)

• Keith Briggs (1) +

• Olly Britton (145) +

• Dietrich Brunn (11)

• Clemens Brunner (1)

• Evgeni Burovski (185)

• Matthias Bussonnier (7)

• CJ Carey (32)

• Cesar Carrasco (4) +

• Hood Chatham (1)

• Aadya Chinubhai (1)

• Alessandro Chitarrini (1) +

• Thibault de Coincy (1) +

• Lucas Colley (217)

• Martin Diehl (1) +

• Djip007 (1) +

• Kevin Doshi (2) +

• Michael Dunphy (2)

• Andy Everall (1) +

• Thomas J. Fan (2)

• fancidev (60)

• Sergey Fedorov (2) +

• Sahil Garje (1) +

• Gabriel Gerlero (2)

• Yotam Gingold (1) +

• Ralf Gommers (111)

• Rohit Goswami (62)

• Anil Gurses (1) +

• Oscar Gustafsson (1) +

• Matt Haberland (392)

• Matt Hall (1) +

• Joren Hammudoglu (6) +

• CY Han (1) +

• Daniel Isaac (4) +

• Maxim Ivanov (1)

• Jakob Jakobson (2)

• Janez Demšar (4) +

• Chris Jerdonek (2) +

• Adam Jones (4) +

• Aditi Juneja (1) +

• Nuri Jung (1) +

• Guus Kamphuis (1) +

• Aditya Karumanchi (2) +

• Robert Kern (5)

• Agriya Khetarpal (11)

• Andrew Knyazev (7)

• Gideon Genadi Kogan (1) +

• Damien LaRocque (1) +

• Eric Larson (10)

• Gregory R. Lee (4)

• Linfye (1) +

• Boyu Liu (1) +

• Drew Allan Loney (1) +

• Christian Lorentzen (1)

• Loïc Estève (2)

• Smit Lunagariya (1)

• Henry Lunn (1) +

• Marco Maggi (4)

• Lauren Main (1) +

• Martin Spišák (1) +

• Mateusz Sokół (4)

• Jan-Kristian Mathisen (1) +

• Nikolay Mayorov (2)

• Nicholas McKibben (1)

• Melissa Weber Mendonça (62)

• João Mendes (10)

• Gian Marco Messa (1) +

• Samuel Le Meur-Diebolt (1) +

• Michał Górny (2)

• Naoto Mizuno (2)

• Nicolas Mokus (2)

• musvaage (18) +

• Andrew Nelson (88)

• Jens Hedegaard Nielsen (1) +

• Roman Nigmatullin (8) +

• Nick ODell (37)

• Yagiz Olmez (4)

• Matti Picus (9)

• Diogo Pires (5) +

• Ilhan Polat (96)

• Zachary Potthoff (1) +

• Tom M. Ragonneau (2)

• Peter Ralph (1) +

• Stephan Rave (1) +

• Tyler Reddy (192)

• redha2404 (2) +

• Ritvik1sharma (1) +

• Érico Nogueira Rolim (1) +

• Heshy Roskes (1)

• Pamphile Roy (34)

• Mikhail Ryazanov (1) +

• Sina Saber (1) +

• Atsushi Sakai (1)

• Clemens Schmid (1) +

• Daniel Schmitz (17)

• Moritz Schreiber (1) +

• Dan Schult (91)

• Searchingdays (1) +

• Matias Senger (1) +

• Scott Shambaugh (1)

• Zhida Shang (1) +

• Sheila-nk (4)

• Romain Simon (2) +

• Gagandeep Singh (31)

• Albert Steppi (40)

• Kai Striega (1)

• Anushka Suyal (143) +

• Alex Szatmary (1)

• Svetlin Tassev (1) +

• Ewout ter Hoeven (1)

• Tibor Völcker (4) +

• Kanishk Tiwari (1) +

• Yusuke Toyama (1) +

• Edgar Andrés Margffoy Tuay (124)

• Adam Turner (2) +

• Nicole Vadot (1) +

• Andrew Valentine (1)

• Christian Veenhuis (2)

• vfdev (2) +

• Pauli Virtanen (2)

• Simon Waldherr (1) +

• Stefan van der Walt (2)

• Warren Weckesser (23)

• Anreas Weh (1)

• Benoît Wygas (2) +

• Pavadol Yamsiri (3) +

• ysard (1) +

• Xiao Yuan (2)

• Irwin Zaid (12)

• Gang Zhao (1)

• ਗਗਨਦੀਪ ਸਿੰਘ (Gagandeep Singh) (10)

共有 149 人为此次发布做出了贡献。名字旁带有“+”的人是首次贡献补丁。此姓名列表是自动生成的，可能不完全。

已关闭的 1.15.0 版本问题

• #2011: DEP: spatial: ‘sokalmichener’ and ‘rogerstanimoto’ distances…

• #2035: ENH: Add methods to compute derivatives (Trac #1510)

• #2116: Cholesky decomposition - are elements in the ‘other’ triangle…

• #2162: MAINT: signal: lombscargle is unclear about normalisation…

• #2509: distributions cdf, sf evaluation in the wrong tail

• #4096: DOC: special: Clarify Mathieu function documentation

• #4275: BUG: linalg.interpolative: svd integer overflow

• #4517: MAINT: special.hankel2: (0, 0) delivers (nan+nan*j) instead…

• #4538: alternative parametrizations of univariate distributions

• #4708: ENH: linalg.funm: a possible speed-up

• #4952: DOC: stats: improvement suggestions for tutorial

• #5818: Clarify definition of preconditioner for sparse linear system…

• #6528: ENH: stats: Add Normal-inverse-gamma distribution

• #7099: ENH: stats.fisher_exact: support tables larger than 2x2

• #7242: ENH: implement at least one good robust scalar root-finding algorithm…

• #8053: Random variate distribution random_state set after init does…

• #8307: scipy.sparse.linalg.eigs gives incorrect largest magnitude eigenvalue

• #8344: BUG: special.nctdtr: incorrect results

• #8362: a function to convert a matrix into diagonal ordered form (ab)…

• #8787: BUG: signal.lombscargle: raises ZeroDivisionError

• #8888: MAINT: special.gegenbauer: case of alpha = 0 not handled

• #9249: BUG: linalg: test_interpolative::TestInterpolativeDecomposition::test_id…

• #9321: Easy access to scipy.stat.<distributionName> frozen distribution…

• #9509: BUG: special: test_kolmogorov fails on 32-bit platforms

• #10106: Slow random variate generation in scipy.stats

• #10328: DOC: stats: documentation is not distribution-specific

• #10364: Add Double Pareto-Lognormal Distribution

• #10374: ENH: make ARPACK-NG deterministic

• #11341: MAINT: constants: disparate electric permittivity constants

• #11465: Numerical stability of distributions: Add log_p as argument to…

• #11649: trust-constr error when attempting to keep bound constrained…

• #12019: scipy.linalg.orthogonal_procrustes

• #12282: API: stats.chisquare: must observed and expected frequencies…

• #12367: Allow whitespace alignment of tables of numbers?

• #12593: BUG: linalg: estimate_rank is quite unreliable

• #12651: BUG: cluster: fcluster assigning all points to a single cluster…

• #12662: Complex numpy exponentiation, nan, and/or inf causes segfault…

• #12895: BUG: special.riccati_yn: Definition missing minus sign

• #13019: TST, MAINT: test_maxiter_worsening on Python 3.9 + Linux ARM64

• #13137: BUG: linalg.schur: bug sorting complex eigenvalues in real form

• #13504: TST: stats: test distribution support method behavior for…

• #13643: BUG: signal.sepfir2d: fails with complex input on Windows

• #13986: Distributions cannot be freed by garbage collector due to self-references

• #14304: Jaccard distance greater than 1 if elements are strings

• #14467: DOC: constants: explain 0.0 uncertainty

• #14582: BUG: special.spherical_jn: Negative arguments lead to nans

• #14788: DOC: linalg.hankel: clarify that first entry of r is…

• #14895: ENH: All statistical distributions in scipy.stats should have…

• #14945: BUG: Overflows/NaNs cause segfault in integrate.quad on certain…

• #15012: ENH: linalg.cholesky: document that user is responsible…

• #15016: BUG: linalg.schur: sorting by imaginary part does not work for…

• #15021: scipy.linalg.qr with pivoting=True should return P as (N,N) permutation…

• #15533: BUG: test failure in test_x0_equals_Mb with bicgstab

• #15582: BUG: special.ncfdtr(dfn, dfd, nc, f) and stats.ncf.cdf(x,…

• #15620: BUG: signal.resample_poly returns an array consisting only of…

• #15888: BUG: linprog, with highs is killed by the OOM killer when called…

• #15896: Compiled code coverage with --gcov doesn’t work with Meson

• #15915: ENH: optimize: object-oriented interface to HiGHS

• #16494: MAINT: Speed up some slow tests

• #16531: DOC: Warnings/error in docstring examples.

• #16700: BUG: segfault on i386 in special test_kolmogorov.py TestSmirnovp

• #17059: ENH: Robust and fast numerical derivative for error propagation

• #17075: Test failures with Intel compilers

• #17136: BUG: logm sometimes raises ZeroDivisionError for matrices with…

• #17307: BUG: stats: frozen distribution domain error produced too late

• #17344: ENH: scipy.linalg.circulant should allow batching

• #17385: BUG: TestOnenormest.test_onenormest_table_6_t_1 is failing…

• #17404: DOC: comparison of optimizers in optimization guide

• #17494: MAINT: HiGHS build flag cleanup

• #17571: ENH: Add units to scipy.constants description

• #17905: ENH: Add softplus implementation

• #18014: Potential use-after-free bug in fcn_callback (Static Analyzer…

• #18250: BUG: complex256 type created by entering complex128 into…

• #18295: BUG: special: Loss of precision in logsumexp

• #18367: RFC: Get rid of linalg.interpolative Fortran code

• #18409: ENH: extensions of factorial{,2,k} for complex domains, recurrences,…

• #18445: ENH: add a callback feature to minimize_scalar and root_scalar

• #18879: BUG: integrate.quad_vec: Fatal error when using workers…

• #18880: TST: test_expm_multiply_dtype failure on aarch64 (flaky test)

• #18882: BUG: minimize does not satisfy nonlinear constraint even if keep_feasible=True

• #18907: BUG: presolve option in milp causing feasible problem…

• #18909: BUG: Cython3 linalg import order bug

• #18945: BUG: circe-ci SVD-LOBPCG benchmarks do not check accuracy so…

• #19042: DOC: sparse: BSR does not support slicing, __getitem__ not…

• #19071: BUG: scipy.special.gammasgn implementation & docs inconsistent…

• #19207: BUG: sparse.linalg: LinearOperator dtype determination broken

• #19223: BUG: io: scipy.io.loadmat error message recommend use of function…

• #19355: MAINT: lobpcg: add unit tests for accuracy matching benchmark…

• #19405: ENH: sparse.csr_array: keep axis functionality in getnnz()

• #19446: BUG: In test_b_orthonormalize, cannot parameterize Vdtype, Bdtype,…

• #19524: BUG: deepcopy on stats.rvs breaks seed / random generation.

• #19634: ENH: Dedicated Function for Envelope Extraction

• #19759: BUG: Bad result for stats.randint.pmf (edge case)

• #19791: DOC: linalg.schur: unclear signature for sort callable when…

• #19907: DEP: extend deprecation of private namespaces also to fortran-generated…

• #20026: ENH: ndimage: 1D rank filter speed up

• #20048: BUG: stats.rv_discrete.ppf: infinite loop in default implementation

• #20077: DEP: linalg: deprecate kron

• #20155: DOC: optimize.curve_fit: Inconsistent naming convention…

• #20196: MAINT: Audit usage of cython memoryviews, add const to allow…

• #20207: ENH: sparse: Validate dtype on sparse array/matrix creation

• #20239: DOC: Sparse arrays: todense() does not return numpy.matrix

• #20240: ENH: multiple small improvements to scipy.stats.circmean

• #20288: ENH: Poisson disk sampling for arbitrary bounds

• #20370: DOC: ndimage.convolve: clarify origin parameter description

• #20389: BUG: sparse.hstack does not respect dtype of {indptr,indices}…

• #20452: BUG:linalg:interpolative: Crashing if k is requested too high…

• #20552: DOC/DEV/MAINT: review distributing section of core-dev guide

• #20574: MAINT, BENCH: would be good to be able to run benchmarks with…

• #20602: MAINT/STY: fix UP031, UP032 linter errors

• #20609: BUG:linalg:interpolative: Inputs are not mapped to compatible…

• #20635: DOC: Titles of long function names in API Reference truncated…

• #20638: DOC: Specify cut-off frequency in firwin as half-ampltude vs…

• #20693: BUG: stats.noncentral_t: incorrect pdf values

• #20710: ENH: special.rel_entr: avoid premature overflow

• #20728: BUG: sparse.linalg: Segfault in arpack with ifx

• #20733: DOC: stats.mannwhitneyu: reversed options for alternative…

• #20739: DOC/DEV: update commit message guidance

• #20740: BLD/DEV: special: build warnings

• #20761: ENH: stats.cauchy: improve cdf and quantile accuracy in tail

• #20763: BUG: sparse.csgraph, array types: some functions should expect…

• #20813: BUG: optimize.nnls sometimes fails when input A is a…

• #20821: BUG: stats.levy_stable.rvs: “S0”-parameterization ignored

• #20844: DOC: update testing docs for alternative backends

• #20879: MAINT: clean up sparse._sputils.getdtype

• #20893: DOC/DEV: Developer docs should mention Accelerate support

• #20904: BUG: sparse.csgraph.dijkstra errors on inputs with int64 or no…

• #20910: BUG: positional argument DeprecationWarning message is overly…

• #20931: MAINT: Premature setting of attributes in HBInfo in scipy.io._harwell_boei…

• #20957: TST, MAINT: array API GPU test failures

• #20963: TST: special.tests.test_support_alternative_backends: failure…

• #20984: BUG: scipy.optimize.nnls 1.11.4 version has better performance…

• #20991: BUG: special.pro_rad1 returns ‘nan’ for any combination of…

• #20994: BUG: spatial.distance.cosine with complex arguments raises…

• #21009: BUG: Floating point exception when passing the invalid argument…

• #21010: BUG: Segmentation fault when passing invalid arguments to some…

• #21011: BUG: Double free or corruption when passing invalid arguments…

• #21016: BUG: Indexing broken for sparse arrays

• #21030: DOC, REL: release notes gh_lists touchups

• #21044: RFC: quo vadis, xp_assert_* infrastructure?

• #21045: BUG: scipy/optimize/tests/test_hessian_update_strategy.py::TestHessianUpdateSt…

• #21057: BUG: loong64 architecture fails TestQuad.test_complex

• #21059: TST, MAINT: TestHyp2f1.test_region5 tol issue with gcc 14.1.0

• #21078: BUG: linalg.expm: slower by a factor of ~4 in scipy>1.13

• #21088: DOC: optimize.InverseJacobian: document or deprecate?

• #21099: DOC: incorrect section ordering for classes

• #21106: BUG: stats.combine_pvalues gives result with wrong dimensionality…

• #21140: BUG: optimize.nnls: re-implementation not robust

• #21144: BUG: sparse: Two new XSLOW test failures

• #21145: BUG: dev.py build –debug doesn’t produce a debug build

• #21148: BUG: optimize.root_scalar pass in an array instead of float

• #21152: BUG: stats.bartlett: returned statistic can be negative…

• #21158: RFC: spatial: review calculation formula for Jaccard distance…

• #21166: ENH: linalg.null_space: expose lapack_driver and check_finite

• #21174: BUG: special tests failing on main when CuPy is installed with…

• #21178: DOC: optimize.root_scalar: arguments incorrectly marked…

• #21181: MAINT: spatial: weighted chebyshev distance revisited

• #21187: BUG: spatial.distance: kulczynski1 returns similarity rather…

• #21192: BUG: stats.differential_entropy: incorrect results with…

• #21193: BUG: optimize: Hessian update strategy fails on nested minimize…

• #21208: RFC: Should mutable default arguments be forbidden by the linter?

• #21212: DOC: doc build failing in CI

• #21217: BUG: SciPy won’t pick the correct BLAS when running the test…

• #21227: BUG: stats/fft/differentiate/optimize: test suite failures with…

• #21239: ENH: Add a Python level lapack wrapper for DSYTRS

• #21241: DOC: add sphinx-copybutton to the documentation

• #21248: BUG: ValueError: x0 violates bound constraints in minimize

• #21256: BUG: io.loadmat: failure with large file due to address position…

• #21266: DOC: CosineMixture function in go_benchmark_functions is wrong

• #21275: CI: wheel builds for Python 3.13 are failing

• #21286: BUG: Delaunay/qhull segfault on self-pass

• #21292: TST: ndimage: GPU test failures

• #21296: DOC: optimize.root: fix docs for inner_\* parameters

• #21300: BUG: integrate.simpson: x param is keyword only

• #21311: BUG: special.chdtr: torch failures

• #21351: ENH: spatial: Rotation: add split to break apart object with…

• #21357: DOC: signal.hilbert: Terminology issue

• #21366: DOC: stats.f_oneway: use attributes instead of tuple unpacking…

• #21369: BUG: stats.tukeylambda: support(lam) is incorrect when lam…

• #21379: BUG: spatial: Voronoi diagram missing ridges

• #21383: BUG: stats.sobol_indices: in-place modification of func…

• #21394: DEV: jupytext notebooks break caching of doc builds

• #21405: MAINT: a few potential minor cleanups

• #21421: DOC/DEV: direct to building guide from quickstart guide

• #21439: DOC/DEV: replace mambaforge with miniforge

• #21447: DEV/MAINT: Allow unicode characters ± and ∞ in source…

• #21452: DOC: signal.firls: reference unavailable

• #21453: ENH: Add support for Xi Correlation in scipy

• #21460: MAINT, DOC: make dist pipefail issue

• #21461: BUG: special.pro_rad2: incorrect results since translation…

• #21486: MAINT: ndimage test failures with CuPy

• #21504: DOC: add note about args/kwargs to description of callable…

• #21507: BUG: fft.fft: real-valued array-api-strict inputs fail

• #21510: DOC: signal.freqz: problem with fs

• #21513: TST, MAINT: test_differentiate torch GPU failures

• #21519: BUG: optimize.minimize: method='Powell' gives array not scalar

• #21534: TST, MAINT: test_matrix_input failing

• #21556: DOC: Lack of table of contents in cluster

• #21566: DOC: stats.pearsonr: error in notes of API reference

• #21571: DOC: interpolate.interp1d: clarify status and alternatives

• #21576: DOC: building: specify .ps1 for windows example

• #21582: BUG: Squeezed output from batched scipy.linalg.det

• #21583: BUG: optimize: test failures in scikit-learn following LBFGS…

• #21584: BUG: linalg.expm: nightly inaccurate for complex64

• #21596: MAINT: Update constants to CODATA 2022 recommendation

• #21610: BUG: special.logsumexp: imaginary component exceeds (-pi, pi]

• #21615: BUG: Invalid treatment of ellipsis in indexing of sparse matrices

• #21627: DOC: optimize.root: outdated naming of fprime in method=’hybr’

• #21630: BUG: optimize.nnls: precision problems

• #21641: BUG: io.mmwrite: auto-appending of .mtx extension

• #21660: BLD, MAINT: linker warnings with newer AppleClang/ld

• #21661: BUG: fft.fht: should set u.imag[-1] = 0 only when n is…

• #21670: BUG: ndimage: _normalize_sequence fails on 0d array

• #21671: BUG: signal.ShortTimeFFT: inverse tranform error with multichannel…

• #21675: BUG: Errors at compiling through pip for python 3.13 with option…

• #21677: BLD: build warnings from quadpack

• #21696: MAINT: lombscargle numerical backward-compat

• #21704: DOC: stats.bootstrap: clarify meaning of paired argument

• #21709: BUG: logsumexp returning incorrect results in Scipy 1.15.0.dev0

• #21724: MAINT: signal deprecation cleanups

• #21733: BUG: cluster: incorrect type of default value of dist in…

• #21738: BUG: “ERROR: Dependency “OpenBLAS” not found” in macOS tests…

• #21745: TST, MAINT: array API GPU test fails in test_cubature.py

• #21747: BUG: boolean indexing of sparse arrays broken on main

• #21758: BENCH/DEV: Add conda-build to environment.yml

• #21759: DEP: remove special.btdtr and special.btdtri

• #21760: BUG: failure in scipy/optimize/tests/test_minimize_constrained.py::test_gh1164…

• #21769: BENCH: Warnings from linprog

• #21772: BUG: optimize.curve_fit with nan_policy="omit" fails…

• #21775: BUG: sparse matrix-vector multiplication fails with flattened…

• #21788: BUG: stats: qmc.Sobol raises ValueError in multi-threading

• #21791: BUG: sparse: setdiag broken when not entire diagonal belongs…

• #21807: BUG: signal: Confusing error when giving an invalid mode to correlation_lags

• #21810: RFC: special: Behavior of gamma function and related functions…

• #21814: TST: interpolate: tests on griddata are not parametrized

• #21817: QUERY: optimize.isotonic_regression: cannot replicate results…

• #21820: BUG: stats: New XSLOW failures in test_fit.py::TestFit

• #21829: BLD: accelerate detection with GNU toolchain on ARM mac

• #21830: BLD: threads.h with MacOS 14.x ARM + gcc 14.2.0

• #21833: SPEC 7 Transition Tracker

• #21837: BUG: linalg.svd: Segmentation Fault, Integer overflow in LAPACK…

• #21838: ENH: sparse: revisit default index dtype selection in sparray…

• #21855: TST, MAINT: torch + GPU failures for test_create_diagonal

• #21862: BUG: large number of fails with macOS 15.1 using Accelerate

• #21885: BUG: interpolate/tests/test_interpnd.py::TestLinearNDInterpolation::test_threa…

• #21900: BUG: stats: New XSLOW test failure in test_sampling.py

• #21908: BUG: integrate.trapezoid: broadcasting failure after #21524

• #21927: TST: failures in test_riemann_zeta_complex on windows in…

• #21934: BUG: Intel oneAPI tests / py3.12, dev.py: meson.build:1:0:…

• #21940: DOC, REL: 1.15.0 author mappings

• #21946: BUG: several failing tests in interpolate on macOS15

• #21949: BUG: stats: XSLOW test failure in scipy.stats.tests.test_fit::TestFit

• #21952: DOC: stats.goodness_of_fit: improve examples

• #21957: CI: failure for “Oldest GCC …” Linux CI job (related to pre-release…

• #21963: DOC: Deprecation warning in sphinx when used with Python…

• #21988: refguide_check currently failing

• #22005: TST: TestJacobian::test_attrs tol bump?

• #22022: TST: tolerance violation in test_x0_working[tfqmr] on windows

• #22029: Test_SVDS_LOBPCG.test_svd_rng_3 test failure in wheel build…

• #22031: BUG: mypy failure in main

• #22077: DOC, REL: a few release notes/process issues

• #22094: API: unannounced breaking change: scipy.integrate.AccuracyWarning…

• #22095: DOC: sparse: sparse.eye_array does not accept tuple[int,…

• #22097: DEP: interpolate.interpnd.GradientEstimationWarning still…

• #22112: BUG/DOC: sparse: ND COO unexpected behaviour 1.15.0rc1

• #22123: DOC: stats: random variable transition guide launches wrong notebook

• #22128: BUG/DOC: it’s not clear how to use differentiate.jacobian…

• #22137: BUG: stats._distribution_infrastructure._Domain.symbols class…

• #22143: BUG: Fail to call BSpline after unpickling with mmap_mode="r"

• #22146: BUG:stats.ContinuousDistribution.llf: should not be public

• #22204: BUG: signal.ShortTimeFFT: istft with mfft > len(win)…

1.15.0 版本拉取请求

• #11345: MAINT: constants: revise way ‘exact’ values are recomputed

• #12071: ENH: linalg: update _procrustes.py to handle complex matrices…

• #12824: ENH: linalg.solve: detect and exploit matrix structure

• #15993: TST: sparse.linalg: Add iterative step test of solvers with LU…

• #16088: DOC: signal: Make _filter_design.py plot labels consistent

• #16090: ENH: vectorize companion matrix function

• #16467: ENH: sparse: add nonzero functionality to min, max, argmin,…

• #16877: MAINT: remove un-necessary all-true array.

• #17318: ENH: signal: Add functionality of Complex Chirp waveform

• #18605: ENH: special: add softplus

• #18979: DOC: Add units to scipy.constants description

• #19058: DOC:special/signal: Add examples for berp_zeros and band_stop_obj

• #19145: ENH: stats: add Landau distribution

• #19209: BUG: sparse.linalg: fix LinearOperator dtype determination

• #19255: ENH: Use highspy in linprog

• #19361: BENCH: sparse.linalg: check accuracy in SVD-LOBPCG benchmarks

• #19475: ENH: stats.lmoment: add function to calculate sample L-moments

• #19764: MAINT: stats: fix axis_nan_policy decorator non-broadcastable…

• #19970: ENH: interpolate: replicate splrep and splprep in Python

• #19988: MAINT: unify factorial implementations

• #19989: MAINT: special: factorial clean-ups

• #20040: DOC: optimize: add comparison of optimizers to guide

• #20058: DOC:signal: Suggest remedies for slow speed in resample when…

• #20097: ENH: signal: Compute envelope of a real- or complex-valued signal

• #20194: BUG: linalg.cossin: fix for nonsymmetric cases

• #20242: DOC: sparse: Correct todense documentation

• #20303: DOC: stats: Convert sampling tutorial to MyST-md

• #20408: DOC: ndimage.convolve: modify origin param description

• #20496: DOC: stats: added Raises section to a few functions

• #20514: TST/BUG: linalg.expm: empty array support

• #20517: DOC: sparse.linalg: updated preconditioner doc for iterative…

• #20519: MAINT: signal: further refactor spline filters

• #20520: BUG: linalg.polar: empty array support

• #20539: ENH: special: Overhaul of Legendre functions

• #20543: ENH: ndimage: log(n) implementation for 1D rank filter

• #20558: ENH: linalg: Cythonize id_dist FORTRAN code

• #20589: TST: optimize.linprog/milp: add tests for various bug reports

• #20671: ENH: interpolate: fix concurrency issues throughout

• #20695: MAINT: special.ndtr: adjust implementation to more closely match…

• #20701: TST/MAINT: special: test with CuPy, make some CUDA fixes

• #20708: ENH: differentiate.hessian: use jacobian to compute…

• #20713: ENH: optimize.HessianUpdateStrategy: add __matmul__

• #20719: MAINT: sparse: fix __init__ func sig to allow maxprint…

• #20743: ENH: stats._xp_mean, an array API compatible mean with…

• #20754: ENH: sparse: add dtype validation in __init__ and astype

• #20759: MAINT: sparse.linalg: adjust norm, eigs, and lsqr…

• #20766: MAINT: stats: minor numerical improvements to circular statistics

• #20767: ENH: stats.qmc: add bounds parameters to PoissonDisk

• #20771: ENH: stats.ttest_ind: add array API support

• #20773: BUG: sparse.csgraph, array types: support non-zero fill_value...

• #20785: ENH: stats.nct.pdf: increase range in left tail using boost

• #20793: ENH: stats: end-to-end array-API support for NHSTs with beta…

• #20794: ENH: stats: add array API support for directional_stats

• #20800: ENH: optimize.elementwise: vectorized scalar optimization and…

• #20809: ENH: ndimage: extend ndimage filter axes support to correlate…

• #20811: ENH: ndimage: extend filter axes support to remaining filters…

• #20816: ENH: special.rel_entr: Avoid overflow before computing…

• #20822: CI: Add Linux workflow to test on free-threaded Python builds

• #20827: REL: set version to 1.15.0.dev0

• #20829: MAINT: special: fix typo in four_gammas used by hyp2f1

• #20830: DOC: optimize.differential_evolution: change convergence…

• #20833: BUG: interpolate: make BSpline.integrate always return an array

• #20834: ENH: integrate.nsum: elementwise evaluation of finite or infinite…

• #20837: MAINT: linalg: add const to Cython function signatures

• #20843: DOC/DEV: add docs for enabling interactive examples

• #20846: DOC: Wrap long titles in docs pages

• #20849: DOC/DEV: mention -b option in contributor guide on testing

• #20855: TST: add additional margin to fail_slows

• #20856: TYP: _lib.doccer: add type annotations

• #20857: ENH: sparse: add axis parameter to count_nonzero method

• #20859: DEP: signal: remove cmplx_sort

• #20862: MAINT: special: Add kokkos mdspan

• #20864: DEP: integrate: remove quadrature and romberg

• #20865: DEP: signal: remove wavelet functions

• #20866: DEP: stats: remove rvs_ratio_uniforms

• #20867: DEP: integrate.cumulative_trapezoid: raise ValueError…

• #20868: DEP: interpolate: deprecate complex dtypes in {Akima1D, Pchip}Interpolator

• #20869: DEP: special.factorial: raise error for non-integer scalars and…

• #20872: MAINT: interpolate: add const to Cython function signatures

• #20873: MAINT: sparse: add const to Cython function signatures

• #20874: MAINT: spatial: add const to Cython function signatures

• #20875: BLD/DEV: special: Fix warning due to mixed initializers

• #20876: DOC: use intersphinx_registry for easier intersphinx mapping…

• #20882: CI: Add workflow to build and upload free-threaded wheels

• #20883: ENH: stats: rewrite ttest_rel in terms of ttest_1samp

• #20884: ENH: stats: end-to-end array-API support for NHSTs with Student’s…

• #20885: BUG: fix incorrect intersphinx-registry entry in environment.yml

• #20886: CI/DEV: fix Node.js 16 warnings by bumping actions

• #20887: MAINT: signal: add const to Cython function signatures

• #20889: MAINT: sparse: Align matmul tests in test_base.py for spmatrix…

• #20891: MAINT: stats: add const to Cython function signatures

• #20895: TST: sparse: Finish allowing test_base.py to easily switch…

• #20897: DOC: Fix bug with parallel doc build

• #20898: MAINT: sparse: clean up _sputils.getdtype docstring

• #20900: ENH: stats: add array API support to combine_pvalues

• #20906: DOC: linalg.schur: update doc for the argument sort

• #20907: CI: Make sure nightly free-threaded wheels are tested with GIL…

• #20908: DOC: signal.dbode: improve docstring

• #20912: DOC: Add more information about how to use Accelerate

• #20913: BUG: sparse.csgraph.dijkstra: fix dtype and shape bugs

• #20915: DOC: integrate.quad_vec: Add example when using workers

• #20916: DOC: Mention that sparse.bsr_array does not support slicing.

• #20922: BUG: stats.mstats: fix mstats.{ttest_rel, ttest_1samp} when…

• #20924: BUG: _lib: ensure reasonable length _deprecate_positional_args…

• #20926: DOC: sparse: Add migration guide for converting code from spmatrix…

• #20928: ENH: optimize._differentiate: add array API support

• #20932: MAINT: io: fix premature setting of attributes in HBInfo

• #20934: TST: stats.combine_pvalues: parameterise tests and update…

• #20941: DOC/MAINT: single to double backticks to remove improper linking

• #20942: CI: Use Cython nightly wheel on free-threaded CI

• #20944: DOC: update distributing section

• #20946: ENH: stats.gmean: add array API support

• #20951: CI: Add MacOS to free-threaded wheel release CI

• #20954: MAINT: stats.hmean/pmean: simplify prior to array API conversion

• #20955: DOC: Single to double backticks for non-targets

• #20962: DOC/MAINT: stats.gmean/gstd/hmean/pmean: document/treat invalid…

• #20965: ENH: stats.tmean: add array API support

• #20968: MAINT: fix some misspellings

• #20969: DOC: linalg: add # may vary to a linalg.schur example

• #20971: TST: special: use standard_normal to generate arguments in…

• #20974: ENH: stats.combine_pvalues: add native axis support

• #20975: DOC: single to double backticks

• #20976: BUG: Update scipy-optimise directive in view of new default role

• #20977: DOC: More single to double backtick

• #20980: CI, MAINT: test_plot_iv NumPy 2 shim

• #20985: BLD: Add build only CI workflow for Windows using MSVC + ifx…

• #20986: DEV: gh_lists: single -> double backticks

• #20987: DOC/DEV: update commit message guidance

• #20989: ENH: stats.chi2_contingency: add method parameter

• #20995: CI: test cp313-dev

• #20998: MAINT: signal: fix code comment typo

• #21003: MAINT: odr: fix a refcounting issue in __odrpack.c

• #21004: DOC: stats: Convert chisquare example to notebook

• #21005: CI: one invocation for all tests in array API job

• #21017: MAINT: smoke-docs: add special/_precompute to –ignore list,…

• #21018: MAINT: sparse: better error message on matmul mismatch

• #21021: ENH:MAINT:optimize: Re-rewrite nnls in Cython

• #21022: BUG: sparse: Fix advanced indexing using both slice and array

• #21023: DEV: lint: enforce newlines at end of files

• #21025: API: signal: rename sosfreqz to freqz_sos

• #21028: ENH: stats.tmin/tmax: add array API support

• #21029: MAINT/STY: Fix UP031 AND UP032 linter errors

• #21032: DOC/DEV: update vendored-code page

• #21033: DOC: interpolate: discuss linear interpolation with extrapolation

• #21034: ENH: stats.xp_var: array-API compatible variance with scipy.stats…

• #21035: ENH: stats.hmean/pmean: add array API support

• #21036: ENH: stats.tvar/tstd/tsem: add array API support

• #21037: MAINT: forward port 1.14.0 relnotes

• #21041: ENH: differentiate: add sub-package for array-API compatible…

• #21042: DOC: stats: Move biomedical examples to notebooks

• #21047: TST: optimize: fix exception test on PyPy3.10

• #21050: ENH: stats.Normal: add new continuous distribution infrastructure…

• #21051: BUG: interpolate.LinearNDInterpolator: fix for precomputed triangulation

• #21063: MAINT: gcc-14 test_region5 tol bump

• #21068: ENH: stats.zmap/zscore/gzscore: add array API…

• #21076: ENH: stats.differential_entropy: add array API support

• #21081: DOC: Add default options for COBYQA

• #21083: MAINT: simplify _integrate_pdf

• #21085: DEP: spatial: deprecate complex input to cosine and correlation

• #21086: DOC: spatial: Fix typo in seuclidean docstring

• #21087: DOC: optimize: remove inadvertent block quote indentation

• #21089: ENH: stats.alexandergovern: vectorize calculation for n-D…

• #21094: DOC: sparse.linalg.gcrotmk: fix backticks and add maxiter…

• #21096: DOC: sparse.linalg.gcrotmk: add missing backticks

• #21097: ENH: stats.boxcox_llf: add array API support

• #21098: DEV: don’t add sparse label for submodules

• #21101: DOC: special.ellipj: fix order of parameters in docstring

• #21103: MAINT: itemsize pybind cleanup

• #21109: MAINT: stats.combine_pvalues: fix native axis support…

• #21110: ENH:sparse.linalg: Update vendored ARPACK version to 3.9.1

• #21111: BUG: ndimage.binary_erosion: avoid divide by zero by capping…

• #21112: DOC: optimize.differential_evolution: fix interval for…

• #21113: MAINT, DOC: simplify docs warn filter

• #21115: BENCH: Corrections to benchmarks README and sparse.Arithmetic…

• #21116: BUG: optimize.root_scalar: let bracket be passed as a NumPy…

• #21117: TST: interpolate: use xp_assert infrastructure

• #21118: DOC: optimize: Add docstring to InverseJacobian

• #21119: MAINT: remove another mpl test shim

• #21120: BUG: cluster: Avoid OOB write when distances are NaN in centroid

• #21121: BUG: cluster: Fix fcluster "maxclust" binary search logic

• #21123: DEV/CI: add gmpy2 back to test dependencies

• #21124: MAINT: use xp_vector_norm instead of xp.linalg.vector_norm

• #21125: MAINT: remove unused minpack2 fortran code

• #21127: TST: fix boxcox_llf test skips

• #21128: TST: special.logsumexp: modernise tests

• #21129: TST: sparse.linalg: test all dtypes for lobpcg B-orthonormalization

• #21131: ENH:optimize: Rewrite MINPACK in C

• #21132: TST: sparse.linalg: simplify dtypes def in test_expm_multiply.py

• #21133: TST: sparse.csgraph: simplify dtypes def in test_graph_laplacian.py

• #21141: DOC: Remove outdated comment about macos/musl in Cirrus CI config

• #21143: MAINT: signal: Don’t redefine PyArray_MIN macro

• #21146: MAINT: sparse.linalg: spsolve simplification

• #21149: ENH: special.logsumexp: add array API standard support

• #21150: ENH: ndimage: add array API standard support

• #21151: DOC: interpolate: fix py:obj reference target not found warning

• #21154: DOC: fix documentation about verbose in minimize_trustregion_constr.py

• #21157: DEP: stats.find_repeats: deprecate function

• #21162: MAINT: optimize.root_scalar: ensure user function gets…

• #21163: MAINT: special: Make fixes needed for ellipkinc and ellipeinc…

• #21167: ENH: interpolate: add AAA algorithm for rational approximation

• #21168: BLD: remove optional test dependencies from cibuildwheel config

• #21171: DEV: fix --debug and add --release to dev.py build

• #21173: BUG/CI: Compile and run tests with ifx + MKL on Linux

• #21176: MAINT: spatial.distance.jaccard: correct numerator calculation

• #21179: BUG: special: cupy import guard

• #21182: MAINT: spatial: improve weighted Chebyshev distance doc and zero…

• #21185: DOC: optimize: Mark x1 as optional for secant root finding

• #21186: DOC: array->``array_like`` in a few functions

• #21194: MAINT: stats.differential_entropy: fix results with integer…

• #21196: DEV: add label glob for scipy.differentiate

• #21197: ENH: sparse: extend COO arrays to n-dimensions

• #21199: MAINT: bump pybind11, mark C/C++/Cython extension modules as…

• #21201: ENH:integrate: Rewrite QUADPACK in C

• #21203: BUG: optimize: Avoid sharing BFGS HessianUpdateStrategy between…

• #21204: MAINT: ensure Python.h is included first in __minpack.c

• #21210: MAINT: special: Make sine and cosine integrals work in CUDA

• #21211: DEP: signal.{correlate,convolve,lfilter}: deprecate object arrays…

• #21213: MAINT, DOC: bump jupyterlite-sphinx lower bound

• #21215: MAINT: _lib: update _docscrape from upstream

• #21216: ENH: integrate._tanhsinh: add array API support

• #21220: TST: sparse.linalg.lobpcg: add unit tests for accuracy

• #21221: BENCH: sparse.linalg: check accuracy properly in LOBPCG benchmarks

• #21222: BUG: special: cython_special missing dep

• #21223: MAINT: interpolate: remove several fused type uses from _bspl.pyx

• #21225: MAINT: io: move _test_fortran under tests

• #21230: ENH: differentiate: support array initial_step

• #21231: ENH: optimize.elementwise.find_minimum: add array API support

• #21235: MAINT: interpolate: remove README

• #21240: ENH: sparse: refactor 2-D COO sparse-dense matrix-matrix multiplication

• #21242: BUG: sparse: fix two slow tests that fail in numpy 2 inside A.nnz

• #21243: BENCH: fix --compare in python dev.py bench

• #21245: DOC: add sphinx-copybutton

• #21247: ENH: linalg.nullspace: expose SVD options

• #21252: DOC: integrate.solve_bvp: make notation consistent

• #21254: CI: Test icx + icpx + ifx + MKL build of SciPy

• #21257: MAINT: BUG: fixed issue where optimization method trust-constr…

• #21259: BUG: interpolate: avoid OOB in the periodic spline constructor

• #21260: BUG: io.loadmat: fix read of large Matlab 4 arrays

• #21264: MAINT: utilise array_api_compat v1.8

• #21265: BUG: fix cstddef include

• #21267: DOC: add Stack Overflow link to readme file

• #21268: MAINT/BENCH: optimize: fix CosineMixture reference

• #21276: MAINT: Fix incorrectly named 1234 Hz fs .wav test file

• #21277: ENH: signal.lombscargle: update to the generalized Lomb-Scargle…

• #21278: BLD: bump cibuildwheel closes #21275

• #21281: MAINT: test_maxiter_worsening cleanup

• #21282: TST: ndimage: unskip test_boundary_spline_accuracy for torch

• #21284: MAINT: Remove mutable defaults in function definitions, forbid…

• #21290: MAINT: linalg: run dos2unix on cython_lapack_signatures.txt

• #21294: TST: fix GPU failures

• #21295: DOC: find_simplex no self

• #21297: MAINT: special: rename C++ library to xsf

• #21299: DEV: gh_lists: improve sanitization of backticks

• #21301: ENH: stats: Implement _logcdf and _logsf for rv_continuous

• #21304: TST: Fix bugs in various tests found via linter

• #21309: MAINT:integrate: Fix an off-by-one error in QUADPACK

• #21310: MAINT/DOC: clean up _lib._array_api, update docs

• #21312: MAINT: special: fix Pylance typing nit

• #21313: BUG: find_simplex shape () segfault

• #21314: ENH: optimize: Rewrite LBFGSB in C

• #21316: MAINT: special.chdtr: fix generic chdtr

• #21318: BUG: special: Fix bug in beta ppf by setting correct Boost policy

• #21321: BUG:linalg: fix solve_banded raising IndexError when…

• #21322: BUG: stats: Allow betaprime._ppf to accept scalars.

• #21326: MAINT: stats: Improve precision of argus.sf.

• #21328: ENH: linalg: add Python wrapper of ?gtcon

• #21329: MAINT: interpolate.AAA: improve input validation of max_terms

• #21330: ENH: integrate: multidimensional integration of array-valued…

• #21331: ENH: linalg: add Python wrapper of ?trcon

• #21332: BUG: stats.levy_stable: fix ignored parameterization

• #21333: DOC:integrate.simpson: Remove reference to removed even param

• #21334: TST: add xfail_xp_backends

• #21335: BUG/ENH: stats: updates for cauchy.

• #21336: ENH: linalg: Add sy/hetrs LAPACK wrappers

• #21339: ENH: stats.poisson_binom: add Poisson Binomial distribution

• #21342: MAINT: stats: fix test that discrete distribution methods accept…

• #21344: BLD: require Meson 1.5.0, fix a missing build dependency for…

• #21346: TST: special: Fix two XSLOW tests.

• #21347: BUG/ENH: stats: Cauchy distribution fixes

• #21349: TST: linalg: skip svd_gesdd test for large matrices for/in…

• #21352: ENH: linalg.lantr: add Python wrapper

• #21353: MAINT: stats.randint.pmf: fix zero PMF values within the…

• #21354: MAINT: stats.rv_discrete.ppf: fix infinite loop bug

• #21355: MAINT: stats.geom.rvs: ensure that output is not negative…

• #21360: BLD: use OpenBLAS v0.3.28 with fewer kernels, fix OpenBLAS licences

• #21361: BLD: switch to more robust git hash determination

• #21363: ENH: linalg.solve: use structure to speed up finite check, apply…

• #21365: BUG: interpolate: FITPACK: remove fpchec.f in-line if-then-endif…

• #21367: MAINT: stats.mannwhitneyu: simplify alternative hypotheses

• #21368: DOC: fixes ISSUE#21366

• #21371: BUG: stats.tukeylambda: Fix the support() method.

• #21373: ENH: interpolate.AAA: add cleanup function for removing spurious…

• #21375: MAINT: Add local resources for smoke-tutorials

• #21377: BENCH: Update BlockDiagSparseConstruction to use a coo_matrix…

• #21378: DOC: signal.hilbert: update to reflect implementation

• #21387: DOC: signal: fix typos in _short_time_fft.py and signal.rst

• #21388: ENH: special: Improve precision of special.logit.

• #21389: ENH: scipy.stats: add normal-inverse-gamma distribution

• #21395: DOC: stats.ecdf: fix example plot

• #21397: DOC/DEV: add pooch to virtualenv dev quickstart guide

• #21399: MAINT: optimize: remove unnecessary isnan check

• #21407: BUG: optimize.minimize: set trust-constr success=False…

• #21412: ENH: sparse: add CSR/CSC _broadcast_to method

• #21413: MAINT: cleaner 0-D/scalar checks for xp assertions, round…

• #21419: ENH: linalg.circulant: allow batching

• #21423: MAINT: linalg.logm: avoid ZeroDivisionError, emit…

• #21425: DOC: linalg.qr: add example about explicit permutation matrix

• #21427: DOC: linalg.cholesky: document when/whether only selected half…

• #21428: DOC: linalg.hankel: emphasize that first element of r is…

• #21429: MAINT: linalg.schur: fix sort='iuc'/'ouc', correct…

• #21430: MAINT: forward port 1.14.1 relnotes

• #21433: DOC: Cache jupyter notebooks build

• #21435: ENH: sparse: Add nD COO support for matmul, dot and tensordot

• #21437: TST/DOC: smoke docs: strict checks

• #21440: ENH: linalg.funm: Pythranize double for loop

• #21441: DOC: special: prolate spheroidal docs wrong

• #21443: DOC/DEV/CI: mambaforge -> miniforge

• #21444: DOC: optimize: make lsq_linear example smaller

• #21446: ENH: linalg: enable N-D batch support in special matrix functions

• #21449: DEV/MAINT: Add ± and ∞ to the extra set of allowed Unicode…

• #21450: Use miniforge for uploading wheel builds

• #21451: DOC: interpolate: add missing call to example code

• #21454: ENH: special: Add root finding tools to xsf and implement gdtrib…

• #21455: TYP: ignore missing sphinx import in mypy.ini

• #21457: MAINT: few scipy cleanups

• #21458: DOC: signal.firls: fix broken link

• #21464: CI: remove fetch-depth: 0 from wheel build jobs

• #21467: STY: fix new lint rules that are popping up in CI

• #21468: MAINT: special: remove more functions from functions.json

• #21469: DOC/MAINT: add references to Boost

• #21471: BUG: sparse: undesired behaviour of 1D/2D matmul

• #21472: MAINT: special: refactor ufunc.h and move to xsf

• #21473: ENH: integrate.cubature: array API standard support

• #21474: DOC: use https URLs in pull request template

• #21477: DOC: update Contributor Quickstart guide to refer to the Building…

• #21478: DOC: Fix version matching by extending the length of GITVER

• #21481: MAINT/DEV: pin sphinx in environment.yml, bump jupyterlite-sphinx

• #21483: ENH: special: computing derivatives by simple autodifferentiation

• #21485: MAINT: interpolate: abstract out barycentric representation of…

• #21492: MAINT: Specify SHELL=/bin/bash in doc/Makefile

• #21493: MAINT: Check for all submodule paths in check_installation

• #21496: TST: Use pytest-run-parallel against free-threaded CI

• #21497: ENH: interpolate: add FloaterHormannInterpolator

• #21499: MAINT, TST: CuPy skip ndimage test

• #21502: MAINT: Update COBYQA to the latest release

• #21505: BUG: special: Use Boost for ncfdtr, fixing accuracy issues

• #21508: BUG: fft: fix real input to standard funcs

• #21512: TST: signal: convert to xp_assert_* infrastructure (pt….

• #21514: DOC: signal.freqz: fix example

• #21515: MAINT: differentiate: manually promote dtype before element assignment

• #21516: TST: remove redundant torch skips

• #21517: DEV: pytorch -> torch

• #21518: ENH: stats.boxcox_llf: add GPU support

• #21520: TST: special._smirnovp: remove test xfail after translation…

• #21524: ENH: integrate.trapezoid: add array API standard support

• #21525: MAINT: stats: Fix typos principle -> principal

• #21526: BUG: optimize.minimize.powell: stop squeezing everything

• #21528: BUG: special.logsumexp: fix type promotion

• #21530: TST: signal: convert to xp_assert_* infrastructure, pt 2

• #21536: TST: interpolate: fix a spurious failure with -b all

• #21539: MAINT: interpolate: move an internal utility from cython to python

• #21542: DOC: integrate.cubature: fix doc formatting and references issues

• #21544: DOC: optimize: document recipe for \*args and \*\*kwargs

• #21545: STY: ignore rule UP038

• #21549: ENH: ndimage: add axes support to most morphology functions

• #21551: MAINT: interpolate: trim down _bspl extension

• #21553: ENH:linalg: Rewrite expm in C

• #21564: TST, MAINT: skip some torch GPU tests

• #21565: ENH: HiGHs re-integration

• #21567: MAINT: stats.wilcoxon: small improvements/fixes

• #21568: MAINT: special: performance optimization for simple autodifferentiation

• #21570: MAINT, BLD: 3.13 to classifiers

• #21572: DEP: spatial.distance: deprecate kulczynski1 and sokalmichener…

• #21573: DOC/MAINT: stats: fix some typos

• #21575: DOC/MAINT: doc: fix some typos

• #21577: MAINT: special.hankel2: fix edge case

• #21578: DOC: special.mathieu_c/sem: add information to docstring

• #21579: TST/DEV: allow stacking of skip_xp_backends

• #21585: DOC/MAINT: special: fix some typos

• #21586: DOC/MAINT: spatial: fix some typos

• #21587: BUG:linalg: Fix single precision expm UV computation

• #21588: BUG: linalg.det: Fix edge case handling of stacked 1x1 arrays

• #21590: DOC: signal.firwin: emphasise cutoff freq is half-amplitude

• #21593: DOC: special.roots_legendre: fix typo in example

• #21595: DOC: building: specify .ps1 for windows example

• #21597: ENH: special.logsumexp: improve precision when one element…

• #21598: DOC: spatial.voronoi_plot_2d: notes on degeneracy

• #21599: CI: use gha-update [wheel build]

• #21600: MAINT: _lib: vendor and use array-api-extra

• #21602: BUG:optimize: Correct guarding, previosuly Fortran indexed, mod…

• #21603: DOC: special.riccati_yn: highlight sign convention

• #21604: MAINT: special.gegenbauer: fix behavior for n=0;…

• #21605: MAINT: constants: update codata values to 2022

• #21607: DOC/MAINT: sparse: fix some typos

• #21608: TST: ndimage: skip fourier tests for JAX

• #21609: ENH: integrate.lebedev_rule: points and weights of Lebedev…

• #21612: MAINT: special: fix delegation to generic implementations

• #21616: BUG: sparse: fix indexing after ellipsis and 2D array indexing

• #21617: DOC/MAINT: signal: fix some typos

• #21621: DOC/MAINT: ndimage: fix some typos

• #21622: MAINT: special.logsumexp: enforce branch cut convention

• #21624: DOC/MAINT: linalg: fix some typos

• #21625: ENH: optimize.differential_evolution: respect workers when…

• #21626: BUG: sparse.linalg: Set t=2 in test_onenormest_table_6_t_1…

• #21629: MAINT: special: use reflection formulas for spherical Bessel…

• #21633: DEP: linalg: deprecate kron

• #21635: TST: special: fix XSLOW failure in test_mpmath.TestSystematic.test_spherical_j…

• #21636: MAINT: differentiate.differentiate: rename function to…

• #21637: MAINT: interpolate: deduplicate input validation for NdBSpline

• #21638: ENH: differentiate.jacobian: support array step_direction

• #21642: BUG: io.mmwrite: fix auto-appending of .mtx extension

• #21646: DOC: sparse: Migration Guide rewording and tweaks

• #21647: DOC: optimize.root: replace ‘fprime’ with ‘jac’ in method=’hybr’

• #21648: DOC/TST: integrate.cubature: Clean up docstring and add…

• #21656: API: integrate.cubature: change some arguments to keyword-only

• #21657: ENH: integrate.cubature: support for infinite limits

• #21658: ENH: stats.chatterjeexi: add xi correlation function

• #21663: CI: fix failing conda setup

• #21668: BUG: fft.fht: set u.imag[-1] = 0 only when n is even

• #21672: BUG: ndimage: fix 0d arrays in _normalize_sequence

• #21673: BUG: signal.ShortTimeFFT: fix multichannel roundtrip with mfft…

• #21678: BUG: fix nan output of special.betaincinv

• #21680: MAINT:integrate: Silence a few QUADPACK compiler warnings

• #21682: DOC: Reduce duplication in user guide

• #21686: BUG: signal: int handling for resample_poly

• #21689: BUG: optimize: fix bugs found in L-BFGS-B implementation

• #21695: ENH: stats.chisquare: add sum_check to disable check…

• #21697: MAINT: signal.lombscargle: fix numerical backward-compat

• #21699: BUG: sparse.linalg: remove faulty test of eigval order from ARPACK

• #21700: ENH: stats.Mixture: add mixture distributions for new infrastructure

• #21701: MAINT: interpolate: remove undocumented nu arg of BSpline.design_matrix

• #21702: MAINT: special.clean up factorial corner cases, including complex…

• #21703: DOC: sparse: update spdiags handling in doc_string and migration…

• #21705: MAINT: Allow greek Unicode symbols in linter.

• #21706: DEP: signal: deprecate object arrays in sosfilt

• #21707: ENH: stats.make_distribution: generate a ContinuousDistribution…

• #21710: ENH: sparse.linalg: convert sparse.linalg to use sparray internally

• #21711: BUG: special: Fix formula for division of dual numbers in xsf

• #21712: MAINT: special.logsumexp: preserve tiny imaginary components

• #21714: TST: stats.sampling: add fail slow exception

• #21718: DOC: stats.bootstrap: improve description of paired…

• #21719: BLD: macos-12 image deprecated [wheel build]

• #21721: MAINT: signal: lombscargle docstring tweaks and fix handling…

• #21723: DOC: Fixed a typo in the description of scipy.sparse.csgraph.maximum_bipartite…

• #21726: ENH: linalg.solve: add assume_a='banded'

• #21728: ENH: special: use boost in nctdtr

• #21729: MAINT: signal.lombscargle: vectorization changes and handling…

• #21730: MAINT: sparse.linalg: use @ for matmul in docs/tests for Linear…

• #21731: ENH: stats.dpareto_lognorm: add double Pareto lognormal…

• #21734: BUG: cluster: correct type of default value of dist in ClusterNode

• #21737: TST: differentiate: small tolerance bump on failing test

• #21739: BLD: fix issue with capitalisation when meson detects Accelerate

• #21744: ENH: special: Extend Riemann Zeta function to complex inputs

• #21746: ENH: stats: exponential, logarithm, and reciprocal of a random…

• #21748: MAINT: Updated pybind11 min version to 2.13.2

• #21749: MAINT/DEV: fix python dev.py ipython under Debian 12 / Python…

• #21751: BUG: array API conformance in cubature

• #21752: ENH: stats.truncate: truncated random variables

• #21753: DOC: sparse: migration to sparray guide updates

• #21754: MAINT: interpolate: rename interpnd -> _interpnd

• #21756: DOC: interpolate: detail interp1d recommended replacements

• #21763: BUG: handle pearsonr constant case with n=2 vectors

• #21765: MAINT: remove equality constraint that was impossible to satisfy…

• #21767: MAINT: stats.ContinuousDistribution: protect attributes,…

• #21768: BUG: sparse: fix ellipsis and bool array indexing combination

• #21770: DOC: stats: correct conclusion associated with high p-value

• #21773: DEP: special: Remove special.btdtr and special.btdtri

• #21774: API: optimize.differential_evolution: transition to Generator…

• #21776: BUG/ENH/DOC: spatial: Better handle & Document Rotation splitting

• #21777: ENH: stats: absolute value of a random variable

• #21778: ENH: optimize: add array api support to rosen and friends

• #21779: ENH: sparse.csgraph: migrate to use sparray (code changes only)

• #21780: BUG: signal: Fix ShortTimeFFT.extent() for fft_mode ==…

• #21784: BUG: sparse: fix Asp @ vsp for 1D sparse vector and add tests

• #21785: MAINT: optimize: migration to sparray pass 1 changes

• #21789: ENH: integrate.nsum: support unimodal functions and infinite…

• #21790: ENH: stats: distributions of order statistics

• #21792: BUG: sparse: fix setdiag for matrices with missing diagonal entries

• #21793: CI: Move free-threaded-wheels.yml into wheels.yml

• #21794: BUG: stats.qmc: fix get_poly_vinit for multi-threading

• #21796: MAINT: Update array_api_compat submodule for the 2023.12…

• #21797: BUG: special: fix array index out of bound for pro_rad2

• #21799: ENH: sparse.linalg: add is_sptriangular and spbandwidth…

• #21800: DOC: signal.find_peaks: Document that widths are returned

• #21801: ENH: extend factorial{,2,k} to allow complex inputs

• #21802: MAINT: special: remove dependence of xsf::numpy on sf_error

• #21808: BUG: signal.correlation_lags: fail with meaningful error…

• #21811: ENH: differentiate: add array API support to jacobian and…

• #21812: API: interpolate.BarycentricInterpolator: transition to Generator…

• #21815: TST: interpolate: parametrize tests on griddata

• #21818: CI: Replace deadsnakes 3.13t by Quansight-Labs/setup-python action

• #21819: ENH: stats.ContinuousDistribution: improve method resolution…

• #21821: TST: optimize: mark MIP6 tests xslow

• #21822: TST: stats.fit: adjust tests for seed->``rng`` transition

• #21823: API: optimize.dual_annealing: adopt SPEC007

• #21824: BUG: sparse: fix idx_dtype when building index arrays in…

• #21825: DOC: 使 _transition_to_rng 替换 rng 的文档

• #21827: BUG: special: 修复 gamma 和 gammasgn 在极点的行为…

• #21831: DOC: 修复 _transition_to_rng 最旧的 GCC "'rng' is not…

• #21832: CI: 限制 array-api-strict 版本以允许转换为…

• #21834: DOC: fft.fht: 修复示例中的拼写错误

• #21839: MAINT:sparse.linalg: 分离 ARPACK 和 PROPACK 代码共享

• #21840: DOC: interpolate: 为平滑样条添加教程页面

• #21841: DOC: 缓和“遗留”警告文本的语气

• #21843: MAINT: stats.lmoment: 修复 keepdims 行为

• #21845: API: optimize.check_grad: 转换为 Generator (SPEC…

• #21847: API: optimize.basinhopping: 转换为 Generator (SPEC 7)

• #21848: API: optimize.quadratic_assignment: 转换为 rng (SPEC…

• #21849: DOC: interpolate: 将一维 FITPACK 包装器声明为遗留

• #21850: BUG: linalg.svd: 避免分段错误

• #21851: STY: 添加有关表中允许的空格的注释

• #21853: DOC: 仅替换 rng 文档的第一段

• #21854: API: stats._resampling: 转换为 rng (SPEC 7)

• #21856: MAINT, TST: test_create_diagonal GPU 支持

• #21857: BLD: 加速 + 非原生时发出警告

• #21858: ENH: sparse: 为 sparse 索引到 sputils 添加安全转换函数

• #21859: DOC: signal.unit_impulse: 添加关于 delta 的解释

• #21860: MAINT: 改进对线程局部存储的处理

• #21861: MAINT: special: 在 a… 中使用一致的 std::isnan 和 std::isinf

• #21864: DEP: 移除 scipy.misc

• #21865: MAINT: fft: 移除过时的文档解决方法

• #21867: MAINT: stats.wilcoxon: 改进文档和测试

• #21868: BUG: io.matlab: 暴露 varmats_from_mat

• #21869: MAINT: stats.ContinuousDistribution: 澄清条件…

• #21870: MAINT: spatial/special: 跟进 std::isnan, std::isinf

• #21871: ENH: stats.ContinuousDistribution.sample: 生成 QMC 样本

• #21873: DOC: sparse.csgraph: 更新 csgraph 文档以使用 sparray

• #21875: MAINT: interpolate.barycentric_interpolate: 添加 rng 参数

• #21878: API: cluster: 转换为 rng (SPEC 7)

• #21880: API: linalg.clarkson_woodruff_transform: 转换为…

• #21881: API: stats: 转换 goodness_of_fit, sobol_indices…

• #21883: DOC/MAINT: 修复关于 GitHub 的一些拼写错误

• #21886: API: stats.{PermutationMethod,BootstrapMethod}: 转换…

• #21888: API: sparse: 将 random 类函数转换为…

• #21891: BUG: spatial: 保护 distance_wrap 指令

• #21892: MAINT: _lib: 使用 compat 中的 is_numpy 等辅助函数…

• #21893: MAINT: linalg.interpolative: 规范化 rng 参数 (SPEC7)

• #21895: MAINT: bump array-api-extra

• #21897: MAINT: Hausdorff Generator 处理

• #21901: MAINT: stats.dpareto_lognorm._cdf: 处理 x=0 处的特殊情况

• #21903: MAINT: linprog 基准测试期间静默警告

• #21904: BUG: 修复 nogil LinearNDInterpolator

• #21905: MAINT: io: 迁移到在 IO 中使用 sparray

• #21906: CI: gha 更新

• #21909: DOC: io.loadmat: 为 uint16_codec 参数添加文档

• #21912: BUG: integrate.trapezoid: 修复广播问题

• #21913: API: spatial.directed_hausdorff: 转换为 'rng' 关键字…

• #21914: MAINT: spatial.transform.Rotation.random: 转换为…

• #21916: MAINT, TST: differentiate: test_examples 容差增加

• #21917: BLD: fortran: 仅对 gfortran 使用 -std=legacy

• #21918: BUG: optimize.curve_fit: 修复 sigma 维度问题…

• #21919: 调用 lcov 为 C、C++ 和 Fortran 生成覆盖率报告…

• #21921: DEP: stats.ttest_ind: 弃用 random_state 和 permutation…

• #21926: REV: Revert “MAINT: io: 将 _test_fortran 移到 tests…”

• #21928: DEV: gh_lists: 修复警告

• #21929: MAINT: bump array-api-extra to 0.2.0

• #21930: TST: sparse.linalg: 调整 expm_multiply 测试的 rtol 以避免不稳定的情况

• #21931: MAINT: stats.qmc: 转换为 rng (SPEC 7)

• #21932: ENH: integrate.nsum: 添加 array API 标准支持

• #21937: TST: linalg.blas: 修复测试并发，将 \*ger 标记为不安全

• #21938: TST: linalg: 不要调用 np.asarray_chkfinite(np.empty((3,…

• #21941: DOC: 添加关于并行执行和线程安全性的文档

• #21943: MAINT: 将 f2py 生成的扩展模块标记为可以安全运行，无需…

• #21944: MAINT: sparse: 将 broadcast_shapes 函数添加到 _sputils.py

• #21947: TST: stats.chatterjeexi: 修复错误消息

• #21948: TST: 添加更全面的 SPEC 7 测试

• #21950: CI: 修复 Intel oneAPI tests 作业

• #21953: TST: stats.rdist: 跳过失败的 xslow 测试

• #21954: ENH: sparse: 确保线程安全

• #21955: ENH: spatial: 确保线程安全

• #21961: MAINT: free-threading: 声明 HiGHS, ndimage._rank_filter_1d…

• #21964: TST: optimize: 为 dual-annealing 测试稍微增加容差

• #21965: ENH: special: 确保测试是线程安全的

• #21967: DOC: stats.goodness_of_fit: 阐明已知/拟合之间的区别…

• #21973: DEV: lint: 禁用 UP031

• #21974: DEV: 将 conda-build 添加到 environment.yml

• #21975: CI: 使用稳定的 NumPy 进行“旧版本构建”

• #21976: TST: optimize.elementwise.bracket_root: 修复 torch 测试…

• #21977: ENH: integrate.tanhsinh: 使 _tanhsinh 公开

• #21979: API: integrate.simpson: 允许 x 按位置传递

• #21981: MAINT: 清除 from __future__ import annotations

• #21982: DOC: SciPy 1.15.0 版本说明

• #21983: BUG: linalg: 修复 cython 导入顺序

• #21984: BUG: signal: 实际拒绝 correlate/convolve 中的对象

• #21985: DOC: optimize.root: 修复 inner_\* 参数的文档

• #21989: DOC: integrate.tanhsinh: 修复引用格式；示例标题

• #21990: CI: 移除 macOS 10.9 的 SDK 和部署目标设置

• #21991: BUG: stats.sobol_indices: 修复输入参数的变异

• #21992: DOC: 在 scipy.cluster 中取消隐藏 toctree 并设置 maxdepth…

• #21994: BLD: 在 macOS 上移除 ld_classic 的使用

• #21996: ENH: stats.fisher_exact: 扩展到 R x C 表

• #21998: MAINT, DOC: 使用 Jupytext 的 API，并修复跨平台使用…

• #22002: TST: 仅在支持的平台上运行复杂 zeta 避免下溢测试…

• #22003: DEV: 工具的统一 git 子模块排除

• #22009: TST: differentiate.jacobian: float32 的容差增加

• #22024: MAINT: 1.15.0rc1 的版本固定/准备

• #22025: DOC: stats: 概率教程/转换指南

• #22026: MAINT: stats.Mixture: 修复默认 weights

• #22027: MAINT: stats.ContinuousDistribution: 改进文档生成；…

• #22030: MAINT: stats.FoldedDistribution: 适配私有…

• #22032: MAINT: 修复 mypy 抱怨

• #22033: TST: 修复 tfmqr 和 svds 的 sparse.linalg 失败

• #22036: DOC: 适应 NumPy 2.2 关于大数组缩写的更改

• #22037: MAINT: stats: 为转换后的分布添加自定义 repr

• #22040: MAINT: stats.make_distribution: 支持更多现有分布

• #22043: ENH: sparse: 使两个 sputils 公开，以便更容易地转换索引数组

• #22048: TST: integrate.tanhsinh: 修复绝对值/权重“压缩”错误

• #22050: MAINT: stats.order_statistic: 覆盖 support

• #22058: DOC: stats.order_statistic: 添加“Returns”部分

• #22059: TST: 暂时跳过扩展测试

• #22067: MAINT: 1.15.0rc1 backports

• #22078: REL: 将 1.15.0rc2 设置为未发布

• #22081: MAINT: 在新基础结构中为分布添加 __str__ 覆盖…

• #22082: BUG, DOC: 修复 md5 哈希报告

• #22085: DOC: sparse: nonzero() 的显式 dtypes

• #22091: DOC: 更新 special 版本说明

• #22098: DOC: 提及已移除 AccuracyWarning

• #22099: DEP: 更新 interpnd 弃用警告中的版本

• #22104: DOC: 1.15.0 版本说明更新

• #22106: DOC: sparse: 修正 eye_array 文档中第一个 shape 输入…

• #22107: MAINT/DOC: Doctests 修复 scpdt 1.6

• #22113: ENH: sparse: 增强构造函数中的 dtype 检查

• #22124: DOC: 修复对“Random Variable Transition Guide”的错误引用…

• #22129: ENH: sparse: nD 清理和文档

• #22135: MAINT: _lib: 为 _deprecate_positional_args 添加缺失的 f 字符串

• #22139: MAINT: stats._SimpleDomain: 确保实例不共享…

• #22149: MAINT: stats.ContinuousDistribution.llf: 移除方法

• #22150: MAINT: SciPy 1.15.0rc2 backports

• #22156: DEP: special.lpn 和 [c]lpmn 的弃用警告

• #22158: MAINT: 接受 interpolate._dierckx 中的 ndarray 子类

• #22162: TYP: 暂时忽略 numpy==2.2.1 的 mypy 错误

• #22167: DEP: special: sph_harm 的弃用警告 + 注释

• #22168: BUG: 修复 uint dtypes 下 0 的阶乘值不正确

• #22175: MAINT: stats: 修复 free-threaded CPython 下的线程安全问题

• #22177: MAINT: 修复未声明 free-threading 支持的扩展模块…

• #22181: REL: 将 1.15.0rc3 设置为未发布

• #22193: DEP: linalg.solve_toeplitz/matmul_toeplitz: 警告 n-D c…

• #22225: DOC: differentiate.jacobian: 关于…的正确/改进的文档