https://github.com/pymc-devs/pymc3

sort by:
Revision Author Date Message Commit Date
864ecb3 Fix bug in switch mixture logp The True and False branches were being mixed up 09 June 2023, 14:55:00 UTC
8b5f437 Fix `check_all_tests_are_covered` hook 08 June 2023, 11:40:56 UTC
2ac88af Rename basic "joint_logprob" functions to "conditional_logp" 08 June 2023, 08:25:32 UTC
a32c5e7 More informative error message for unused step sampler arguments (#6738) Co-authored-by: Joseph Hall <joseph.hall@bp.com> 07 June 2023, 15:17:32 UTC
261862d Fix auto-naming of multi-output logprobs 05 June 2023, 19:03:51 UTC
5b68edc Fix nested and single output IfElse logp 05 June 2023, 19:03:51 UTC
a30e0d4 Allow composition of interdependent container variables Join/MakeVector/IfElse can output multiple interdependent variables. These are potentially measurable because in the logp each output is given a distinct value variable. However, this isn't known during the IR rewrites. To circumvent this issue, we run an inner IR rewrite after giving dummy value variables to each output 05 June 2023, 19:03:51 UTC
9bba026 Fix logprob of `check_and_raise` 05 June 2023, 19:03:51 UTC
7ed5b71 Cleanup RV in graph checks 05 June 2023, 19:03:51 UTC
13f9894 New logprob inference logic This commit changes the logic used for logprob inference. Instead of eager bottom-up conversion to measurable variables in the IR rewrites, we only convert nodes whose outputs were marked as "needs_measuring". This is achieved with the new `PreserveRVMappings.request_measurable` method. This strategy obviates the need to undo unnecessary conversions. It also obviates a subtle need for graph cloning via the `ignore_logprob` helper, which prevented intermediate measurable rewrites from being reversed when they were needed to derive the logprob of valued variables, but were not directly valued. This indirect role of `ignore_logprob` is now done more explicitly and efficiently via the `request_measurable` method. All other uses of `ignore_logprob` (and `reconsider_logprob`) were removed from the codebase The `get_measurable_outputs` dispatching was also abandoned in favor of only considering outputs associated with value variables. A new MergeOptimizerRewrite was written to further target local rewrites to only those nodes whose variables have been marked as `needs_measuring`. 05 June 2023, 19:03:51 UTC
0fa051d Remove unused test util 05 June 2023, 19:03:51 UTC
eb88380 uncommented bambi lines 04 June 2023, 14:13:20 UTC
4d2b2ef update installation instructions to request version >5 (#6752) 02 June 2023, 02:19:04 UTC
f3df36b fix tiny error in overload definition 30 May 2023, 09:42:52 UTC
b17a60d Ignore named variables that are not traceable in `get_vars_in_point_list` 30 May 2023, 09:40:01 UTC
fbc62d5 Remove IR `Op`s from final logprob graph 27 May 2023, 08:33:37 UTC
4d0360c Fix nested replacement of useless IR conversions 27 May 2023, 08:33:37 UTC
8c93bb5 Fix Minibatch for multiple variables 24 May 2023, 08:44:44 UTC
db15ae4 Avoid cloning of Minibatch values 24 May 2023, 08:44:44 UTC
cafb60b Update PyTensor dependency 23 May 2023, 08:45:12 UTC
bfbc8cc Handle Scan in collect_default_updates This allows proper seeding in CustomDists with Scans 23 May 2023, 08:32:28 UTC
0073639 Remove `joint_logprob` function from `tests.logprob.utils` (#6650) * Remove joint_logprob function from tests.logprob.utils * Move the joint logprob test for subtensors to test_mixture.py 19 May 2023, 11:48:40 UTC
c57769c Rename _replace_rvs_in_graphs and fix bug when replacing input 17 May 2023, 20:05:22 UTC
e1060de Add redirects from old v3 notebooks (#6719) * add redirects from old v3 notebooks * fix redirects 16 May 2023, 08:14:07 UTC
bee65b4 update gaussian_processes.rst (#6693) * moved and cleaned up GP guide * added GP guide to core_notebooks index * implement oriol's suggestions * implement oriol's suggestions * Update docs/source/learn/core_notebooks/Gaussian_Processes.rst Co-authored-by: Oriol Abril-Pla <oriol.abril.pla@gmail.com> * Update docs/source/learn/core_notebooks/Gaussian_Processes.rst Co-authored-by: Oriol Abril-Pla <oriol.abril.pla@gmail.com> --------- Co-authored-by: Oriol Abril-Pla <oriol.abril.pla@gmail.com> 15 May 2023, 20:14:32 UTC
8f43799 ⬆️ UPGRADE: Autoupdate pre-commit config 15 May 2023, 08:59:56 UTC
9886f8b Change logger names from all "pymc" to module name (__name__) 13 May 2023, 13:53:51 UTC
f632a34 Type overloading for return_inferencedata in pm.sample() (#6709) 12 May 2023, 22:42:48 UTC
9b72c2e Add ICDF function to pareto distribution Adds ICDF (quantile) function for the Pareto distribution. Source https://en.wikipedia.org/wiki/Pareto_distribution Issue #6612 11 May 2023, 08:45:02 UTC
10eab32 Add ICDF function to laplace distribution Adds ICDF (quantile) function for the laplace distribution. Source https://en.wikipedia.org/wiki/Laplace_distribution Issue #6612 11 May 2023, 08:45:02 UTC
c377bf6 target GLM_linear >> glm_linear (#6714) 10 May 2023, 17:23:43 UTC
a53e865 Implement Hurdle distributions 10 May 2023, 12:02:34 UTC
970db18 Move distributions to more consistent places ZeroInflated to mixture.py DiracDelta to distribution.py 10 May 2023, 12:02:34 UTC
83e3545 Remove deprecated Constant alias 10 May 2023, 12:02:34 UTC
4242faf Speedup Slice sampler 08 May 2023, 16:04:00 UTC
fee9a02 Return tune information from Slice sampler This avoids a crash when mixing slice sampling with another sampler that has `tune` stats 08 May 2023, 16:04:00 UTC
9e91935 ⬆️ UPGRADE: Autoupdate pre-commit config 07 May 2023, 19:28:31 UTC
a617bf2 Added scale parameterization to `Exponential` (#6677) 28 April 2023, 14:48:31 UTC
371472d Add logprob inference for not operations (#6689) 28 April 2023, 10:50:45 UTC
d4bb701 Derive logprob for hyperbolic and error transformations (#6664) * cleaned up if block in find_measurable_transform * Adapt default `RVTransform.log_jac_det` to univariate and vector transformations. * Use np.testing in check_jacobian_det Co-authored-by: Luke LB <ll17354@bristol.ac.uk> Co-authored-by: Ricardo Vieira <28983449+ricardoV94@users.noreply.github.com> 28 April 2023, 09:44:16 UTC
890c2cc Updated Pytest video series link 28 April 2023, 08:29:15 UTC
5cf7efe Fix numerical precision issues in discrete ICDFs. 27 April 2023, 15:29:51 UTC
910d9ef Add logcdf implementation for Truncated distributions (#6690) 27 April 2023, 12:07:46 UTC
9f01be2 Rectify type-hints for `set_data` 26 April 2023, 17:32:49 UTC
a86fe7c Fix WhiteNoise subclassing from Covariance (#6674) * fix WhiteNoise subclassing from Covariance (#6673) Since #6458, Covariance is now the base class for kernels/covariance functions with input_dim and active_dims, which does not include WhiteNoise and Constant kernels. * add regression test for #6673 * fix WhiteNoise input to marginal GP 26 April 2023, 17:24:39 UTC
55d915c Make logprob inference for binary ops independent of order of inputs (#6682) 26 April 2023, 13:21:11 UTC
a59c9cd Adding book cover page image into the documentation page. 21 April 2023, 10:58:16 UTC
3f7881e update link to pymc port of DBDA code 20 April 2023, 06:50:09 UTC
61be336 Add logprob derivation for >= and <= operations 19 April 2023, 17:42:01 UTC
9b712bf Derive logprob of less and greater than comparisons (#6662) 19 April 2023, 11:53:59 UTC
f2bb88b Make metropolis elemwise updates independent of each other Not updating q0 after each elemwise update rendered subsequent proposals dependent on the previous ones. 19 April 2023, 07:34:01 UTC
1ed4475 Allow logcdf inference in CustomDist 15 April 2023, 05:10:22 UTC
e2eb26d Do not require class_name for CustomDist and Simulator dists * Also remove the experimental warning when using CustomSymbolicDists 15 April 2023, 05:10:22 UTC
aae97a2 Arrange distributions and sub-contents alphabetically (#6653) 13 April 2023, 11:29:47 UTC
23572e1 Fix dtype casting bug in icdf function 13 April 2023, 11:21:36 UTC
fa82fef Fix bug in random function of HalfStudent (#6658) Co-authored-by: Ricardo Vieira <ricardo.vieira1994@gmail.com> 13 April 2023, 10:44:49 UTC
2a324bc Bump Pytensor dependency 12 April 2023, 08:54:36 UTC
5d68bf3 Fix changed files filters in tests workflow Introduced in 433703ab0156c88de4eb3b180ea85ec21dd3e3fb 08 April 2023, 11:44:14 UTC
0d7413e Add all_tests job to be used as single requirement 07 April 2023, 13:47:04 UTC
7973967 Add Type Hints to distribution parameters (#6635) 07 April 2023, 13:15:48 UTC
09dc9d0 Raise warning if RVs are present in derived probability graphs 06 April 2023, 19:16:46 UTC
30c9179 Improve docstrings in GP module (#6652) 06 April 2023, 15:56:59 UTC
e5a200f Add welcome bot (#6602) * config.yml file >>> add welcome bot links * Update links to contributing guide --------- Co-authored-by: Oriol Abril-Pla <oriol.abril.pla@gmail.com> 06 April 2023, 14:21:07 UTC
b1a75ae Add timestamps to list of possible contributions (#6623) 06 April 2023, 14:08:39 UTC
041e2f9 Add Gitpod video to contributing page (#6646) 06 April 2023, 13:10:27 UTC
48c8dc8 ⬆️ UPGRADE: Autoupdate pre-commit config 04 April 2023, 19:34:33 UTC
264afdc ⬆️ UPGRADE: Autoupdate pre-commit config (#6624) * ⬆️ UPGRADE: Autoupdate pre-commit config --------- Co-authored-by: pymc-bot <pymc-devs@users.noreply.github.com> 02 April 2023, 21:51:00 UTC
b7764dd Added ICDF for the continuous exponential distribution. (#6641) * Added ICDF for the continuous exponential distribution. * Removed redundant out of bounds parameter tests. 01 April 2023, 18:37:28 UTC
eb4e353 fix docstrings for censored and timeseries (#6638) 01 April 2023, 09:00:33 UTC
617f93c install theme from conda (#6639) 01 April 2023, 08:59:38 UTC
8d5b236 update links to sponsor images 31 March 2023, 22:53:19 UTC
6404805 Add model debug helper 31 March 2023, 14:36:03 UTC
4c64eb9 Don't use `check_parameters` in `get_tau_sigma`. The default use of `check_parameters` indicates that an expression can be replaced by -inf, if the constraints aren't met. Instead, if `can_be_replaced_by_ninf=False`, sampling would fail for negative tau/sigma. To avoid this, the conversion now returns the right value for positive tau or sigma, but negative images if the inputs were negative. The methods that then validate the paramters (such as logp, logcdf, random), can later catch this. 31 March 2023, 14:36:03 UTC
3f2a1da Improve collect_default_updates * It works with nested RNGs * It raises error if RNG used in SymbolicRandomVariable is not given an update * It raises warning if same RNG is used in multiple nodes 30 March 2023, 10:21:00 UTC
a75af50 Add rtd preview action 30 March 2023, 09:02:47 UTC
1bfd128 Install theme from pypi Also updates the sphinx cap to match the one from the theme, which is more realistic. I haven't tried but doubt our doc build would still work with sphinx 1.5 30 March 2023, 09:02:47 UTC
46f8e2f Allow logcdf and icdf inference 30 March 2023, 08:43:50 UTC
ae9fcac Implemented logprob for SpecifyShape and CheckandRaise (#6538) 29 March 2023, 21:02:37 UTC
f7861b5 Updating Error message when checking the starting values for MCMC (#6632) 29 March 2023, 20:50:23 UTC
f3ce16f Infer logp for elemwise transformations of multivariate variables 29 March 2023, 10:49:11 UTC
d860f63 Fix broadcasting issue in ScaleTransform jacobian 29 March 2023, 10:49:11 UTC
6f80846 Fix jacobian dimensionality alignment in ChainedTransform The existing code only considered mixing of scalar and vector transforms, but not potentially higher dimensionality transforms (e.g., matrix) 29 March 2023, 10:49:11 UTC
83cd926 Do mamba and pre-commit updates in background 25 March 2023, 19:02:35 UTC
be560e4 Define PRE_COMMIT_HOME in dev.Dockerfile 25 March 2023, 19:02:35 UTC
3f4dc72 Bump dev container base image 25 March 2023, 19:02:35 UTC
9a3f504 Separate logp-related dispatch helpers from user facing functions 25 March 2023, 17:28:55 UTC
d1625da Simplify single variable logp inference 25 March 2023, 17:28:55 UTC
a584d98 Rename logprob/joint_logprob to logprob/basic.py 25 March 2023, 17:28:55 UTC
2fcce43 Added ICDF for the discrete uniform distribution. 24 March 2023, 06:32:52 UTC
067d89b Improved docstring for predictions argument in sample_posterior_predictive (#6616) * Improved docstring for predictions argument in sample_posterior_predictive * Fix typos Co-authored-by: Ravin Kumar <7213793+canyon289@users.noreply.github.com> --------- Co-authored-by: Ravin Kumar <7213793+canyon289@users.noreply.github.com> 22 March 2023, 22:51:18 UTC
473c952 Implement icdf for Univariate distribution (#6528) Also extended testing.check_icdf with skip_paradomain_outside_edge_test param --------- Co-authored-by: Michal Raczycki <michalraczycki@macbook-pro-michal.home> 19 March 2023, 15:28:37 UTC
67925df Fix a typo in the docstring of OderedLogistic (#6611) 18 March 2023, 08:20:45 UTC
c7279b5 Remove shape functions (#6556) 17 March 2023, 15:38:56 UTC
0334994 Derive logprob of IfElse graphs 17 March 2023, 12:13:39 UTC
239da11 Add definition of B in documentation for Beta distribution (#6604) * Add definition of B in documentation for Beta dist * Define B in Beta distribution and link Wikipedia. 17 March 2023, 11:24:26 UTC
5dcd101 Fix logprob inference for scans with carried deterministic states 16 March 2023, 13:14:02 UTC
a542581 Add test for scans over sequences 16 March 2023, 13:14:02 UTC
fafc020 Use better `assert_no_rvs` from logprob submodule This utility can find RVs in inner graphs 16 March 2023, 13:14:02 UTC
bae121a Add HSGP Latent GP approximation (#6458) * add HSGP implementation * Extend Covariances to support power spectral densities --------- Co-authored-by: Michael Osthege <michael.osthege@outlook.com> 14 March 2023, 23:05:15 UTC
c709abf Remove `auto` argument from `pm.Deterministic` docstring (#6592) * Remove auto argument from Deterministic * Add missing arguments in the deterministic docstring as per numpydoc style guide * Add missing arguments in the Potential docstring * Add spaces around colons --------- Co-authored-by: Michael Osthege <michael.osthege@outlook.com> 14 March 2023, 09:42:17 UTC
back to top