Add MvNormalMeanScalePrecision distribution #206
Conversation
|
ping @Nimrais |
test/distributions/normal_family/mv_normal_mean_scale_precision_tests.jl
Outdated
Show resolved
Hide resolved
…{MvNormalMeanScalePrecision})
…{MvNormalMeanScalePrecision})
…malMeanScalePrecision})
…recision efficency test fix: remove unneeded code fix: remove not needed stuff fix: remove unused code test: add efficency test fix: return distributions_setuptests to HEAD test(fix): typo test(fix): remove unneeded testset test(fix): update efficency test
Nimrais
force-pushed
the
dev_mvscalenormal
branch
from
3812a9b to
77f4a0d
Compare
|
@bvdmitri I think PR is ready for review, but I need some help with efficient implementation of the fisher. The only tests that are failing are once that checking that fisher in this parametrisation is really faster. |
|
Thanks for refactoring this, @Nimrais! |
test/distributions/normal_family/mv_normal_mean_scale_precision_tests.jl
Outdated
Show resolved
Hide resolved
…n_tests.jl Co-authored-by: Bagaev Dmitry <bvdmitri@gmail.com>
|
@Nimrais ping, what needs to be done to finish this PR? |
|
@bvdmitri The basic functionality is now implemented. The only remaining task is to implement the multiplication of the Fisher matrix with another matrix for the Arrowhead matrix structure in BayesBase. Currently, this multiplication is only done in our tests. There's already a generic method for all matrix structures using the getindex method, so implementing this method is the last thing we need to do. |
|
@bvdmitri basically we need this PR ReactiveBayes/BayesBase.jl#25 be merged to finish this one |
| @@ -28,7 +28,7 @@ TinyHugeNumbers = "783c9a47-75a3-44ac-a16b-f1ab7b3acf04" | |||
|
|
|||
| [compat] | |||
| Aqua = "0.8.7" | |||
| BayesBase = "1.2" | |||
| BayesBase = "1.5.0" | |||
There was a problem hiding this comment.
Is there stuff in BayesBase 1.5.0 that you need specifically?
There was a problem hiding this comment.
yes, the arrowheadmatrix and the all methods for it to make the CI pass
There was a problem hiding this comment.
This PR was initially aimed at addressing ReactiveBayes/ReactiveMP.jl#387, which it still does. The distribution in question is parametrized by the mean and scale parameter of the precision matrix.
Initially, I implemented it as part of the
MultivariateNormalDistributionsFamily. However, the conversions betweenMvNormalMeanScalePrecisionand other distributions in this "class" don't always hold.During the process, @Nimrais suggested that this distribution could be particularly interesting for
ExponentialFamilyProjections.jl. To make it more useful, we need to optimize methods related to the computation of the Fisher information matrix. I made a first attempt to improve performance by modifying the computation ofkron(invη2, invη2). I believe further improvements are possible, but this serves as a starting point.Any suggestions for additional optimizations to enhance the distribution's effectiveness are much welcome.
UPD: I added the piece of code that actually fixes the ReactiveBayes/ReactiveMP.jl#387