Changelog

0.20. Kernels, kernels, kernels. All you can think about is covariance functions. I am sick of your kernels. I am sick of cleaning the leftover eigenspaces you leave on the carpet when you come home, late in the night, after spending all day hewing positive semidefinite operators. I am sick of hearing your breath, the stale unique acrid smell that fills the algebraists’ workrooms, wafting through the linen to my nostrils. Go away. Go away from here and rest with your beloved kernels! (2023-08-29)

Release highlights

  • Improvements to the kernel system.

  • GP objects are immutable.

Kernels

  • Coherent logic to determine the class of the result of operations on kernels.

  • Each generic class has it own Cross* superclass representing crosskernels.

  • Decorators to make crosskernels.

  • Transformation system with clear semantics for linear operators and algebraic operations.

  • Exponentiation scalar ** kernel.

  • Replaced diff and where with linear operators.

  • Hardcoded non-fuzzy derivability check. May yield false positives when additional derivatives are taken by an outer context, but disabling it is easy.

  • Fix derivation on crosskernels when one input is scalar and the other is structured.

GP

  • GP objects are immutable: operations can only return a new, different object. This makes them more compatible with jax and reduces chances of errors.

  • Change name and signature of methods to define processes: the prefix is def instead of add, and the first argument is always the key of the new process.

  • GP methods that return dictionaries use ordinary Python dicts instead of gvar.BufferDict.

0.19. Miss the Forest for Two Forests and an Auxiliary Regression Term (2023-08-22)

Release highlights

  • GP version of BCF (Bayesian Causal Forests).

  • Extension of the copula submodule to a full PPL (Probabilistic Programming Language) to specify the hyperparameter priors.

bayestree submodule

  • The new class bcf implements the infinite trees limit of the BCF model for causal inference in deconfounded observational studies. Additionally, it allows specifying an arbitrary auxiliary regression term.

  • bart adds a parameter marginalize_mean to avoid having the mean as hyperparameter.

copula submodule

The new class Distr replaces CopulaFactory and can be instantiated. It represents a probability distribution over a tensor, much like PyMC. Distr objects can be used as parameters to other Distr objects to define a model. The distribution is always parametrized to be a multivariate standard Normal.

Improvements to StructuredArray

  • StructuredArrays created from numpy structured arrays with same names and types, but different padding, will now compare as equal instead of different.

  • StructuredArray.from_dict to create a StructuredArray from a dict of arrays.

  • Define len(StructuredArray).

  • Support numpy.concatenate.

  • Support numpy.lib.recfunctions.append_fields.

gvar-related

  • Solve bug that could make jax.jit fail if a gvar.BufferDict was passed as argument.

  • Decorator gvar_gufunc to make a function support gvars.

0.18. Our code, hallowed be thy dependencies, but deliver us from pins and torment (2023-08-02)

Recent versions of lsqfitgp would impose version 0.4.6 of jax and jaxlib due to incompatibilities with newer releases. This limitation is now gone. Thanks to the experimental Windows support added to jax, lsqfitgp can be pip installed on Windows. The unit tests currently have some failures and crash midway, so I guess it’s not fully usable.

New decomposition system

The covariance matrix decomposition system has been scrapped and rewritten. Consequences:

  • All solvers but 'chol' are gone, they were never useful in practice anyway. I may add new solvers in the future. Consequently, it is not necessary to specify GP(..., solver='chol') to speed up the fit.

  • It is not possible to pass a decomposition in place of givencov to use the Woodbury matrix formula.

  • Operations with decompositions can not be differentiated; instead, there is a method that computes the Normal density and its derivatives with custom code. This is faster because it avoids redundant calculations.

  • GP.marginal_likelihood loses the separate option.

  • lsqfitgp.raniter fails if the matrix is not p.s.d. instead of issuing a warning and then working around the problem.

  • The 'fisher' method in empbayes_fit now works decently enough that it can be used in practice, same for covariance='fisher'.

empbayes_fit

  • empbayes_fit(..., minkw=dict(method='trust-constr') did not work, fixed.

  • Now jit=True by default.

0.17. And God spoke through the interface, and the exception was raised KeyError(‘fagiolino’) (2023-07-17)

This release mostly features improvements to the BART simplified interface.

bayestree.bart

  • Example on the ACIC data challenge

  • Error weights

  • New methods bart.gp(), bart.data(), bart.pred()

  • All methods can sample hyperparameters with Laplace

  • The prior mean is a free hyperparameter

  • The prior on the error variance is scaled with the data, in particular bart now works well with non-standardized data

raniter

  • Accept user provided random number generator

  • Solved bug with scalar elements in dictionaries

empbayes_fit

  • The arguments in gpfactorykw are now passed through a function call instead of being closed over. This implies that, if jit=True, large arrays can be passed efficiently without being hardcoded in the compiled code, and that arguments not understood by JAX can not be passed through gpfactorykw.

0.16. If fitting a nonparametric regression became so easy, many professional statisticians would be left without a job! (2023-04-24)

This release adds a simplified interface that packages a Gaussian process regression with the BART kernel.

0.15. EUROCIM special edition, 10 exclusive bugs for the first 500 users, download now (2023-04-17)

This release mostly features further improvements to the BART kernel. According to the benchmarks I am going to show at EUROCIM 2023, GP regression with this kernel and hyperparameter tuning beats the original cross-validated BART MCMC both in statistical and computational performance. Following the ACIC challenge 2022 results, this implies that it’s probably SoTA on the standard causal inference in decounfounded observational study setting with $n<10000$, although that is still to be verified.

BART

  • Strongly optimized implementation, both on memory usage and running time.

  • Computation is batched by default to avoid creating large intermediate matrices when there are many covariates.

  • Good choices for technical parameters are documented.

  • BART.correlation also accepts total splits and indices w.r.t. the grid instead of split count separations.

  • Example script with full BART inference.

Hyperparameters fit (empbayes_fit)

  • New parameters:

    • covariance to set how to estimate the posterior covariance matrix

    • fix to fix some hyperparameters to the initial values

    • mlkw to pass additional keyword arguments to GP.marginal_likelihood

    • forward to use forward autodiff for the likelihood gradient; it uses a partially reverse autodiff calculation to avoid the $O(n^3d)$ complexity of full-forward mode

  • Removed the 'hessian' and 'hessmod' values of parameter method. Use 'fisher' instead. <– BREAKING CHANGE

  • New attributes:

    • prior: the prior on the hyperparameters, as gvars

    • initial: the starting values for the MAP search

    • fix: the mask indicating locked hyperparameters

    • gpfactory: the gpfactory argument

  • Additional logging: per task timing and function calls count.

  • Copula factories for the Beta and Inverse Gamma distributions, in the submodule copula.

StructuredArray

  • New function unstructured_to_structured to create a StructuredArray from an unstructured array.

  • Arrays are now read-only. Setting the fields requires a JAX-like syntax with .at[field].set(value). <– BREAKING CHANGE

  • Added methods squeeze and astype, although the latter refuses to change the type.

  • Pytree unflattening tries to infer shape and dtype if the leaves are not compatible with the original structure. The inference is incomplete and ambiguous. <– BREAKING CHANGE

  • Implemented numpy functions ix, empty_like, and empty.

Kernels

  • New CrossKernel parameter batchbytes to batch the calculation, and corresponding method batch. Requires jax-traceability of the kernel implementation.

  • The Decaying kernel gets an explicit real exponent to indicate infinite divisibility.

GP

  • By default, even if checksym=False, the full covariance matrix is computed for diagonal blocks, unless the additional new option halfmatrix=True is set. <– BREAKING CHANGE

Dependencies

  • Pinned jax and jaxlib to 0.4.6. Earlier versions severely break the optimization of the BART kernel, while later versions break the jit option in empbayes_fit. Sorry for this pin, I do know that dependency pins in libraries are evil. <– BREAKING CHANGE

  • Tested on Python 3.11

0.14.1. It wasn’t me! (2023-03-28)

Limits the maximum supported jax version to 0.4.6 due to the new jax.jit implementation triggering a bug I have not been able yet to fix.

0.14. Bayes Area Rapid Transit (2023-02-16)

Improvements to the BART kernel:

  • nonuniform splitting probability along covariates

  • interpolation between lower and upper bound

  • recursion reset to increase accuracy

A good but fast approximation to the exact BART kernel can now be obtained with BART(..., maxd=4, reset=2, gamma=0.95). Other features:

  • StructuredArray.from_dataframe converts pandas and polars dataframes to a jax-compatible columnar format

  • Added LOBPCG to the decompositions, it computes a low-rank approximation of the covariance matrix, less accurate but faster than ‘lowrank’, renamed Lanczos

  • Faster and coherent positivity check of GP objects. Only the highest and lowest eigenvalues, estimated with LOBPCG, are checked, which stays fast even with large matrices. The check is repeated as many times as necessary on the minimal set of covariance blocks touched by any operation.

  • The Woodbury formula, triggered by passing a pre-decomposed error covariance matrix, now works well for ill-conditioned priors (but not still for the error covariance).

  • GP.add(proc)lintransf define the default process if not otherwise specified.

  • The GP solvers now require two separate parameters epsrel and epsabs instead of eps to specify the regularization. epsabs may be useful to fix the regularization independently of the hyperparameters when using empbayes_fit.

  • Decompositions produced by GP.decompose have additional operations and properties (de)correlate(..., transpose=True), matrix(), eps, m

Fixed many bugs, in particular all JAX JIT incompatibilities are gone.

New requirements:

  • scipy >= 1.5

  • jax, jaxlib >= 0.4.1

0.13. The mark of Riemann (2022-11-01)

Major changes:

  • The Fourier kernel has been replaced with Zeta, which is the generalization to a continuous order parameter.

New features:

  • New method GP.decomp to decompose any positive semidefinite matrix.

  • GP.pred and GP.marginal_likelihood now accept a Decomposition produced by GP.decomp as value to the ycov parameter, indicating that the decomposition of the prior covariance matrix shall be done with Woodbury’s formula. This is computationally convenient if the subset of process values is a low rank linear transformation of some other values, and ycov is a fixed matrix that can be computed and decomposed once and for all. The low rank is automatically detected and exploited, although only to one level of depth in the transformations defined with GP.addtransf or GP.addlintransf. The current implementation of Woodbury does not work properly with degenerate or ill-conditioned matrices.

  • GP.addcov accepts a new parameter decomps to provide precomputed decompositions of the diagonal blocks of the user-provided covariance matrix. These decompositions are pre-loaded in the cache used by GP.pred and GP.marginal_likelihood.

  • New function sample, equivalent of gvar.sample.

  • New parameter initial to empbayes_fit to set the starting point of the minimization.

  • New parameter verbosity to empbayes_fit to set console logging level.

  • New attributes pmean, pcov and minargs to empbayes_fit. minargs provides the complete argument list passed to the minimizer.

  • New parameter MA(norm=...) to standardize the moving average covariance kernel to unit variance.

  • New method BART.correlation to compute the BART correlation with more options than the standard kernel interface.

  • Optimization in the implementation of BART to make the computational cost linear with two levels of depth.

Bugs fixed

  • Some compatibility issues with new JAX versions.

0.12. The wealth of a man is measured not in gold, neither in cattle; but in his most treasured covariance functions (2022-07-16)

This release improves the built-in collection of covariance functions.

New kernels:

  • Cauchy: generalized Cauchy kernel, replaces and extends RatQuad

  • HoleEffect

  • Bessel

  • Sinc, Pink, Color: kernels with power-law spectra

  • StationaryFracBrownian: a stationary analogue of FracBrownian

  • Decaying: a kernel for exponentially decaying functions

  • CausalExpQuad

  • Log

  • Circular: a periodic kernel, less smooth than Periodic

  • AR, MA: discrete autoregressive and moving average processes

  • BART: Bayes Additive Regression Trees

Improved kernels:

  • Constant, White: support non-numerical input

  • Matern: all derivatives work for any value of nu; extended to nu=0 as white noise

  • GammaExp: fixed problems with second derivatives

  • Gibbs: support multidimensional input

  • FracBrownian: the fractional Brownian motion is now bifractional with an additional index K; extended to negative values

  • Fourier: support arbitrarily large n

  • Wendland: renamed PPKernel, now accepts real exponent parameter

Deleted kernels:

  • RatQuad: replaced by Cauchy

  • PPKernel: replaced by Wendland

  • Matern12, Matern32, Matern52: replaced by Maternp

New general feature: kernels have a maxdim option to set a limit of the dimensionality of input.

Improvements to StructuredArray:

  • now supported numpy functions can be applied from numpy instead of using their duplicates in lsqfitgp

  • a StructuredArray can be converted back to a numpy array

  • a StructuredArray can be converted to an unstructured array with numpy.lib.recfunctions.structured_to_unstructured

0.11. With great linearity comes great responsibility (2022-06-11)

New features:

  • Two more generic GP methods to define linear transformations, addproclintransf and addlintransf, which take arbitrary callables

  • Support for jax’s just in time compiler

  • Second order methods in empbayes_fit, included Fisher scoring

And many bugfixes as usual.

0.10. A rose by any other name would be differentiable (2022-06-01)

Ported from autograd to jax. Consequences:

  • Does not naively pip-install anymore on Windows

  • Internal bug: backward derivatives of bilinear forms are broken

  • Scalar-Kernel multiplication in autodiff context now works

  • Problems with second derivatives of some kernels are now solved

0.9. Free as in “you can peek at the prior” (2022-05-15)

  • Lifted the restriction that points can’t be added any more to a GP object after getting the prior in gvar form (thanks to G. P. Lepage for implementing the necessary gvar feature)

  • New GP parameter posepsfac to change the tolerance of the prior covariance matrix positive definiteness check

  • The positiveness check is done when avoiding gvars too

0.8. Fourier (2022-05-05)

Fixes a lot of bugs. New features and improvements:

  • Fourier series, experimental and half-broken (loc and scale won’t work)

  • New process-transformation methods GP.addkernelop, GP.addprocderiv, GP.addprocxtransf, GP.addprocrescale

  • User-defined prior covariance matrices

  • Cache of prior covariance matrices decompositions, to get the posterior for different variables separately

  • New kernel class StationaryKernel

  • Improved the numerical accuracy of the Fourier kernel

  • Option saveargs in Kernel to remember the initialization arguments

  • New regularizations svdcut- and svdcut+ that preserve the sign of the eigenvalues

  • Now BlockDecomp implements correlate and decorrelate with the block Cholesky decomposition

0.7. Little washing bear (2022-04-22)

  • With the new family of methods GP.addproc* one can define independent processes and combinations of them.

  • Fixed error which arose sometimes when taking the gradient of GP.marginal_likelihood with multiple derivatives taken on the process.

  • The derivability checking system is now correct but fuzzy: it emits warnings if the derivatives may be illegal, and raises an exception only when they are surely illegal.

  • Added axes: int parameter to GP.addtransf to contract more indices.

0.6.4. gvar fix^2 (2022-03-13)

A fix to adapt to a gvar fix.

0.6.3. gvar fix (2022-03-10)

A fix to adapt to a change in gvar internals.

0.6.2. ar kernel (2020-05-19)

  • Renamed the AR2 kernel “Celerite”, because it did not really cover all the AR(2) cases.

  • The Harmonic kernel now supports taking a derivative w.r.t Q even at Q=1.

0.6.1. Night fix (2020-05-19)

  • Simplified the parametrization of AR2 and Harmonic kernels.

  • Fixed numerical accuracy problems in the Harmonic kernel.

0.6. ARRH (2020-05-18)

  • New kernels AR2, BrownianBridge, Harmonic.

0.5.1. Flat prior (2020-05-17)

  • Fixed bug in GP.prior(key, raw=True) which would return a flat array whatever the original shape

0.5. Fastiter (2020-05-16)

  • New generator lsqfitgp.raniter; like gvar.raniter but takes the mean and covariance matrix separately.

  • Solved bug in GP.prior when raw=True

0.4. Classy (2020-05-15)

  • Converted empbayes_fit to a class and optimized it a bit.

0.3. Quad (2020-05-12)

  • Improved numerical accuracy.

  • New kernel OrnsteinUhlenbeck.

0.2.1. No SVD (2020-05-09)

  • Use a gvar 11.5 feature to replace SVD with diagonalization in gvar.raniter.

0.2. Series (2020-05-08)

  • New kernels Taylor and Fourier.

0.1.4. Forcekron (2020-05-05)

  • Fixed kernels Cos, Periodic, Categorical and FracBrownian on multidimensional input. Now Cos and Periodic are not incorrectly considered isotropic.

0.1.3. Readonly (2020-05-04)

  • Non-field indexing of StructuredArray now returns a read-only StructuredArray that can not be assigned even on whole fields. Wrapping it again with StructuredArray removes the limitation.

  • New parameters raises of empbayes_fit to disable raising an exception when the minimization fails.

0.1.2. Asarray (2020-05-03)

  • Exposed StructuredArray-compatible numpy-like functions

  • Improved docstrings

  • empbayes_fit raises an exception when the minimization fails

0.1.1. Vector (2020-05-01)

  • Support decorating a np.vectorize function as kernel

0.1. Wiener (2020-04-30)

  • Added kernel WienerIntegral

  • Support scalars in GP.addx

0.0.2. Some fixes (2020-04-27)

  • Improved docstrings

  • Improved error messages

  • Optimized gvar priors for transformed processes

0.0.1. Initial release (2020-04-23)

No comment.