rel-2.32.0

What's Changed

Major Changes 🛠

• Improve dot lift rewrites by @ricardoV94 in #1471
• Change Dot Op to only accept matrix inputs by @ricardoV94 in #1538

New Features 🎉

• Refactor and update QR Op by @jessegrabowski in #1518
• Optimize matmuls involving block diagonal matrices by @jessegrabowski in #1493

Bugfixes 🐛

• Fix CheckAndRaise Op C implementation by @ricardoV94 in #1521
• Fix mean, var and std for XTensorVariables by @ricardoV94 in #1533
• Fix AdvancedIncSubtensor1 C-compilation with empty indices by @ricardoV94 in #1551

Documentation 📖

• Fix missing xtensor.type cross-reference targets by @OriolAbril in #1530

Maintenance 🔧

• Converts negative constant axis to positive if present in Join(COp) by @PabloRoque in #1527
• Preserve static shape information in block_diag by @jessegrabowski in #1529
• Get rid of pydot-ng because it's deprecated since 7 years by @maresb in #1534
• Canonicalize subtensor negative integer indices by @ricardoV94 in #1541
• Use stricter numerical tolerance in rewrites and allow casting in PatternNodeRewriter by @lciti in #1526

New Contributors

• @PabloRoque made their first contribution in #1527

Full Changelog: rel-2.31.7...rel-2.32.0

rel-2.31.7

What's Changed

New Features 🎉

• Implement full_like, ones_like, and zeros_like for XTensorVariables by @ricardoV94 in #1519

Documentation 📖

• Add xtensor docs by @ricardoV94 in #1504

Maintenance 🔧

• Extend C implementation of psi to support negative inputs by @Abhinav-Khot in #1523
• Make convolve mode symbolic to avoid unnecessary large convolution in gradient graph by @ricardoV94 in #1522

Full Changelog: rel-2.31.6...rel-2.31.7

rel-2.31.6

What's Changed

New Features 🎉

• Implement broadcast for XtensorVariables by @ricardoV94 in #1513

Bugfixes 🐛

• Do not skip validation between consecutive Elemwise inplace replacements by @ricardoV94 in #1494
• Fix shape errors in scalar_solve_to_division rewrite by @ricardoV94 in #1516

Full Changelog: rel-2.31.5...rel-2.31.6

rel-2.31.5

What's Changed

New Features 🎉

• Implement vectorized jacobian and allow arbitrary expression shapes by @ricardoV94 in #1228
• Rewrite scalar solve to division by @jessegrabowski in #1453

Bugfixes 🐛

• Fix xtensor (...,) Transpose by @ricardoV94 in #1500

Maintenance 🔧

• Small fixups to xtensor type and XRV by @ricardoV94 in #1503

Full Changelog: rel-2.31.4...rel-2.31.5

rel-2.31.4

What's Changed

New Features 🎉

• Decompose Tridiagonal Solve into core steps by @ricardoV94 in #1382
• Implement symbolic minimize and root Ops by @jessegrabowski in #1182
• Add Feature that can go back and forward in rewrite history by @ricardoV94 in #874
• Reuse cholesky decomposition with cho_solve in graphs with multiple pt.solve when assume_a = "pos" by @jessegrabowski in #1467
• Cover more cases of log1mexp stabilization by @lciti in #1483
• Add JAX dispatch for CholeskySolve Op by @jessegrabowski in #1491
• Implement xarray-like labeled tensors and semantics by @ricardoV94 in #1411

Bugfixes 🐛

• Use grad to compute jacobian when input shape is known to be (1,) by @jessegrabowski in #1454
• Fix invalid numpy dtype by @ricardoV94 in #1456
• Fix bug in gradient of Blockwise'd Scan by @ricardoV94 in #1482

Documentation 📖

• Improve creating an Op documentation page by @ricardoV94 in #1086
• Fix optimize example notebook folder name by @jessegrabowski in #1474
• Correct indexing links in docs by @williambdean in #1463

Maintenance 🔧

• Simplify Numba implementation of Alloc by @ricardoV94 in #1329
• Use lapack func instead of scipy.linalg.cholesky by @Fyrebright in #1487
• Avoid miniKanren v1.0.4 by @maresb in #1496

New Contributors

• @lciti made their first contribution in #1483
• @Fyrebright made their first contribution in #1487

Full Changelog: rel-2.31.3...rel-2.31.4

rel-2.31.3

What's Changed

New Features 🎉

• Support RandomVariable graphs with scalar shape parameters in JAX backend by @ricardoV94 in #1029

Bugfixes 🐛

• Handle compute_map=None in Scan by @zaxtax in #1435
• Coerce dtype __props__ to string due to invalid hash of np.dtype() objects by @ricardoV94 in #1436

Maintenance 🔧

• Fix slow dot in numba by @ricardoV94 in #1426
• Constant fold branches of variadic add/mul by @ricardoV94 in #1422
• Reuse output of Join in C backend by @ricardoV94 in #1340
• Avoid default allocation for taps of length 1 in ScanSaveMem by @ricardoV94 in #1395

Full Changelog: rel-2.31.2...rel-2.31.3

rel-2.31.2

What's Changed

New Features 🎉

• C implementation of convolve1d by @ricardoV94 in #1404

Bugfixes 🐛

• Fix numba dispatch not returning arrays or wrong dtypes by @ricardoV94 in #1406
• Respect check_finite in LU decomposition rewrites by @ricardoV94 in #1417
• Fix GEMV dot case with empty output and beta 0 by @ricardoV94 in #1410
• Fix shape issues in jax tridiagonal solve by @benmaier in #1414

New Contributors

• @benmaier made their first contribution in #1414

Full Changelog: rel-2.31.1...rel-2.31.2

rel-2.31.1

What's Changed

New Features 🎉

• Reuse LU decomposition in Solve by @ricardoV94 in #1396

Bugfixes 🐛

• Fix bug in gemv c code by @ricardoV94 in #1408
• Fix bug in local_blockwise_advanced_inc_subtensor by @ricardoV94 in #1405

Full Changelog: rel-2.31.0...rel-2.31.1

rel-2.31.0

What's Changed

Major Changes 🛠

• Speedup AdvancedSubtensor1 and AdvancedIncSubtensor1 in C backend by @ricardoV94 in #1346
  • AdvancedIncSubtensor1 now explicitly forbids runtime broadcast

New Features 🎉

• Add Ops for LU Factorization by @jessegrabowski in #1218
• Implement several subtensor lift rewrites by @ricardoV94 in #1158
• Implement numba dispatch of StudentT by @ricardoV94 in #1402
• Implement C code for ExtractDiagonal and ARange by @ricardoV94 in #1392

Bugfixes 🐛

• Use shell=False for Popen on Windows by @michaelosthege in #1324
• Fix lu_solve with batch inputs by @ricardoV94 in #1394

Maintenance 🔧

• Update pyproject license schema according to deprecation warnings by @maresb in #1362
• Fix typo in warning message by @zaxtax in #1364
• Restore d3viz/html/template.html by @williambdean in #1380
• Faster convolve1d in numba backend by @ricardoV94 in #1378
• Use PyObject_CallNoArgs to call thunks from VM by @ricardoV94 in #1383
• Avoid copy of flipped A matrices in GEMV by @ricardoV94 in #1389
• Resolve deprecation warnings of regex library by @emmanuel-ferdman in #1393
• Speedup Python implementation of Blockwise by @ricardoV94 in #1391
• Remove dead C code importing time.h by @maresb in #1403
• Remove iostream include for C backend by @maresb in #1400

New Contributors

• @emmanuel-ferdman made their first contribution in #1393

Full Changelog: rel-2.30.3...rel-2.31.0

rel-2.30.3

What's Changed

New Features 🎉

• Support numba compiled sort and argsort functions by @victorgarcia98 in #1309
• Inplace Composite and ScalarLoop Ops with multiple outputs by @ricardoV94 in #1322
• Allow specializing shape of predefined tensor types by @ricardoV94 in #1355

Bugfixes 🐛

• Fix bug in infer_shape of Blockwise(Subtensor) by @ricardoV94 in #1353

Maintenance 🔧

• Propagate static output shapes in Split and avoid copy in C-impl by @ricardoV94 in #1343

New Contributors

• @victorgarcia98 made their first contribution in #1309

Full Changelog: rel-2.30.2...rel-2.30.3