Switch more fully to measures terminology

NEWS.md

@@ -79,7 +79,7 @@ Breaking changes
   likelihood = MyLikeLihood(mydata)
   ```
 
-  This allows for defining likelihoods without depending on BAT. Avoid creating custom subtypes of `BAT.AbstractMeasureOrDensity`.
+  This allows for defining likelihoods without depending on BAT.
 
 * New behavior of `ValueShapes.NamedTupleShape` and  `ValueShapes.NamedTupleDist`: Due to changes in [ValueShapes](https://github.com/oschulz/ValueShapes.jl) v0.10, `NamedTupleShape` and `NamedTupleDist` now either (by default) use `NamedTuple` or (optionally) `ValueShapes.ShapedAsNT`, but no longer a mix of them. As a result, the behavior of BAT has changed as well when using a `NamedTupleDist` as a prior. For example, `mode(samples).result` returns a `NamedTuple` now directly.
 

Project.toml

@@ -4,6 +4,7 @@ version = "3.1.2"
 
 [deps]
 Accessors = "7d9f7c33-5ae7-4f3b-8dc6-eff91059b697"
+Adapt = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"
 AffineMaps = "2c83c9a8-abf5-4329-a0d7-deffaf474661"
 ArgCheck = "dce04be8-c92d-5529-be00-80e4d2c0e197"
 ArraysOfArrays = "65a8f2f4-9b39-5baf-92e2-a9cc46fdf018"
@@ -19,6 +20,7 @@ Distributed = "8ba89e20-285c-5b6f-9357-94700520ee1b"
 Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"
 DistributionsAD = "ced4e74d-a319-5a8a-b0ac-84af2272839c"
 DocStringExtensions = "ffbed154-4ef7-542d-bbb7-c09d3a79fcae"
+DomainSets = "5b8099bc-c8ec-5219-889f-1d9e522a28bf"
 DoubleFloats = "497a8b3b-efae-58df-a0af-a86822472b78"
 ElasticArrays = "fdbdab4c-e67f-52f5-8c3f-e7b388dad3d4"
 EmpiricalDistributions = "0bbb1fad-0f24-45fe-94a4-415852c5cc3b"
@@ -85,6 +87,7 @@ BATUltraNestExt = "UltraNest"
 
 [compat]
 Accessors = "0.1"
+Adapt = "3, 4"
 AdvancedHMC = "0.5"
 AffineMaps = "0.2.3, 0.3"
 ArgCheck = "1, 2.0"
@@ -102,6 +105,7 @@ Distributed = "1"
 Distributions = "0.25"
 DistributionsAD = "0.5, 0.6"
 DocStringExtensions = "0.8, 0.9"
+DomainSets = "0.5, 0.6, 0.7"
 DoubleFloats = "0.9, 1"
 ElasticArrays = "1.2.3"
 EmpiricalDistributions = "0.2, 0.3.1"
@@ -149,7 +153,7 @@ Tables = "0.2, 1.0"
 Transducers = "0.4"
 TypedTables = "1.2"
 UltraNest = "0.1"
-ValueShapes = "0.10.1"
+ValueShapes = "0.11"
 ZygoteRules = "0.2"
 julia = "1.6"
 

docs/src/experimental_api.md

@@ -14,25 +14,25 @@ bat_compare
 bat_integrated_autocorr_len
 bat_marginalmode
 BAT.auto_renormalize
+BAT.batsampleable
 BAT.BinnedModeEstimator
 BAT.DistributionTransform
 BAT.enable_error_log
 BAT.error_log
 BAT.EvalException
 BAT.ext_default
 BAT.get_adselector
-BAT.LogUniform
 BAT.PackageExtension
 BAT.pkgext
 BAT.set_rng
+batmeasure
 BridgeSampling
 EllipsoidalNestedSampling
 GridSampler
 HierarchicalDistribution
 PriorImportanceSampler
 ReactiveNestedSampling
-renormalize_density
 SobolSampler
-truncate_density
+truncate_batmeasure
 ValueAndThreshold
 ```

docs/src/internal_api.md

@@ -32,16 +32,16 @@ Order = [:macro, :function]
 # Documentation
 
 ```@docs
-BAT.AbstractMeasureOrDensity
-BAT.AbstractProposalDist
 BAT.AbstractSampleGenerator
-BAT.AnyIIDSampleable
-BAT.AnyMeasureOrDensity
 BAT.AnySampleable
 BAT.BasicMvStatistics
+BAT.BATMeasure
+BAT.BATPushFwdMeasure
+BAT.BATPwrMeasure
+BAT.BATWeightedMeasure
 BAT.CholeskyPartialWhitening
 BAT.CholeskyWhitening
-BAT.DistLikeMeasure
+BAT.DensitySampleMeasure
 BAT.ENSAutoProposal
 BAT.ENSBound
 BAT.ENSEllipsoidBound
@@ -52,63 +52,47 @@ BAT.ENSRandomWalk
 BAT.ENSSlice
 BAT.ENSUniformly
 BAT.FullMeasureTransform
-BAT.GenericDensity
 BAT.LFDensity
 BAT.LFDensityWithGrad
 BAT.LogDVal
 BAT.MCMCIterator
 BAT.MCMCSampleGenerator
+BAT.MeasureLike
 BAT.NoWhitening
 BAT.OnlineMvCov
 BAT.OnlineMvMean
 BAT.OnlineUvMean
 BAT.OnlineUvVar
-BAT.Renormalized
 BAT.SampleTransformation
 BAT.StandardMvNormal
 BAT.StandardMvUniform
 BAT.StandardUvNormal
 BAT.StandardUvUniform
 BAT.StatisticalWhitening
-BAT.Transformed
 BAT.UnshapeTransformation
 BAT.WhiteningAlgorithm
 
-
-BAT.WrappedNonBATDensity
-
-BAT.trafoof
 BAT.logvalof
 BAT.bat_report!
 BAT.fft_autocor
 BAT.fft_autocov
 BAT.argchoice_msg
-BAT.bat_sampler
 BAT.bg_R_2sqr
 BAT.checked_logdensityof
-BAT.default_val_numtype
-BAT.default_var_numtype
-BAT.density_valtype
 BAT.drop_low_weight_samples
 BAT.find_marginalmodes
-BAT.fromuhc
-BAT.fromuhc!
-BAT.fromui
 BAT.get_bin_centers
 BAT.getlikelihood
 BAT.getprior
 BAT.gr_Rsqr
 BAT.is_log_zero
 BAT.issymmetric_around_origin
-BAT.log_volume
 BAT.log_zero_density
-BAT.proposal_rand!
-BAT.proposaldist_logpdf
 BAT.repetition_to_weights
 BAT.smallest_credible_intervals
-BAT.spatialvolume
 BAT.sum_first_dim
+BAT.supports_rand
 BAT.trunc_logpdf_ratio
 BAT.truncate_dist_hard
-BAT.var_bounds
+BAT.measure_support
 ```

docs/src/tutorial_lit.jl

@@ -127,25 +127,20 @@ using BAT, DensityInterface, IntervalSets
 
 # ### Likelihood Definition
 #
-# First, we need to define the likelihood (function) for our problem.
+# First, we need to define the likelihood for our problem.
 #
-# BAT represents densities like likelihoods and priors as subtypes of
-# `BAT.AbstractMeasureOrDensity`. Custom likelihood can be defined by
-# creating a new subtype of `AbstractMeasureOrDensity` and by implementing (at minimum)
-# `DensityInterface.logdensityof` for that type - in complex uses cases, this may
-# become necessary. Typically, however, it is sufficient to define a custom
-# likelihood as a simple function that returns the log-likelihood value for
-# a given set of parameters. BAT will automatically convert such a
-# likelihood function into a subtype of `AbstractMeasureOrDensity`.
+# BAT expects likelihoods to implements the `DensityInterface` API. We
+# can simply wrap a log-likelihood function with
+# `DensityInterface.logfuncdensity` to make it compatible.
 #
 # For performance reasons, functions should [not access global variables
 # directly] (https://docs.julialang.org/en/v1/manual/performance-tips/index.html#Avoid-global-variables-1).
 # So we'll use an [anonymous function](https://docs.julialang.org/en/v1/manual/functions/#man-anonymous-functions-1)
 # inside of a [let-statement](https://docs.julialang.org/en/v1/base/base/#let)
 # to capture the value of the global variable `hist` in a local variable `h`
 # (and to shorten function name `fit_function` to `f`, purely for
-# convenience). `DensityInterface.logfuncdensity` turns a log-likelihood
-# function into a density object.
+# convenience). `DensityInterface.logfuncdensity` then turns the
+# log-likelihood function into a `DensityInterface` density object.
 
 likelihood = let h = hist, f = fit_function
     ## Histogram counts for each bin as an array:
@@ -208,12 +203,8 @@ prior = distprod(
 
 #md nothing # hide
 
-# In general, BAT allows instances of any subtype of `AbstractMeasureOrDensity` to
-# be uses as a prior, as long as a sampler is defined for it. This way, users
-# may implement complex application-specific priors. You can also
-# use `convert(AbstractMeasureOrDensity, distribution)` to convert any
-# continuous multivariate `Distributions.Distribution` to a
-# `BAT.AbstractMeasureOrDensity` that can be used as a prior (or likelihood).
+# BAT supports most `Distributions.Distribution` types, and combinations
+# of them, as priors.
 
 
 # ### Bayesian Model Definition

ext/BATAdvancedHMCExt.jl

@@ -15,24 +15,26 @@ using Random
 using DensityInterface
 using HeterogeneousComputing, AutoDiffOperators
 
-using BAT: AnyMeasureOrDensity, AbstractMeasureOrDensity
+using BAT: MeasureLike, BATMeasure
 
 using BAT: get_context, get_adselector, _NoADSelected
-using BAT: getalgorithm, getmeasure
+using BAT: getalgorithm, mcmc_target
 using BAT: MCMCIterator, MCMCIteratorInfo, MCMCChainPoolInit, MCMCMultiCycleBurnin, AbstractMCMCTunerInstance
 using BAT: AbstractTransformTarget
 using BAT: RNGPartition, set_rng!
 using BAT: mcmc_step!, nsamples, nsteps, samples_available, eff_acceptance_ratio
 using BAT: get_samples!, get_mcmc_tuning, reset_rng_counters!
 using BAT: tuning_init!, tuning_postinit!, tuning_reinit!, tuning_update!, tuning_finalize!, tuning_callback
-using BAT: totalndof, var_bounds, checked_logdensityof
+using BAT: totalndof, measure_support, checked_logdensityof
 using BAT: CURRENT_SAMPLE, PROPOSED_SAMPLE, INVALID_SAMPLE, ACCEPTED_SAMPLE, REJECTED_SAMPLE
 
 using BAT: HamiltonianMC
 using BAT: AHMCSampleID, AHMCSampleIDVector
 using BAT: HMCMetric, DiagEuclideanMetric, UnitEuclideanMetric, DenseEuclideanMetric
 using BAT: HMCTuningAlgorithm, MassMatrixAdaptor, StepSizeAdaptor, NaiveHMCTuning, StanHMCTuning
 
+using ValueShapes: varshape
+
 using Accessors: @set
 
 

ext/BATCubaExt.jl

@@ -11,9 +11,9 @@ end
 using BAT
 BAT.pkgext(::Val{:Cuba}) = BAT.PackageExtension{:Cuba}()
 
-using BAT: AnyMeasureOrDensity, AbstractMeasureOrDensity
+using BAT: MeasureLike, BATMeasure
 using BAT: CubaIntegration
-using BAT: var_bounds, bat_integrate_impl
+using BAT: measure_support, bat_integrate_impl
 using BAT: transform_and_unshape, auto_renormalize
 
 using Base.Threads: @threads
@@ -125,17 +125,16 @@ function _integrate_impl_cuba(integrand::CubaIntegrand, algorithm::CuhreIntegrat
 end
 
 
-function BAT.bat_integrate_impl(target::AnyMeasureOrDensity, algorithm::CubaIntegration, context::BATContext)
-    measure = convert(AbstractMeasureOrDensity, target)
-    transformed_measure, _ = transform_and_unshape(algorithm.trafo, target, context)
+function BAT.bat_integrate_impl(target::MeasureLike, algorithm::CubaIntegration, context::BATContext)
+    measure = batmeasure(target)
+    transformed_measure, _ = transform_and_unshape(algorithm.trafo, measure, context)
 
-    vb = var_bounds(transformed_measure)
-    if !(all(isapprox(0), vb.vol.lo) && all(isapprox(1), vb.vol.hi))
-        throw(ArgumentError("CUBA integration requires measures that (can be converted to) have unit volume support"))
+    if !BAT.has_uhc_support(transformed_measure)
+        throw(ArgumentError("CUBA integration requires measures are supported only on the unit hypercube"))
     end
 
-    renormalized_measure, logrenormf = auto_renormalize(transformed_measure)
-    dof = totalndof(renormalized_measure)
+    renormalized_measure, logweight = auto_renormalize(transformed_measure)
+    dof = totalndof(varshape(renormalized_measure))
     integrand = CubaIntegrand(logdensityof(renormalized_measure), dof)
 
     r_cuba = _integrate_impl_cuba(integrand, algorithm, context)
@@ -152,9 +151,9 @@ function BAT.bat_integrate_impl(target::AnyMeasureOrDensity, algorithm::CubaInte
     end
 
     (value, error) = first(r_cuba.integral), first(r_cuba.error)
-    rescaled_value, rescaled_error = exp(BigFloat(log(value) - logrenormf)), exp(BigFloat(log(error) - logrenormf))
+    rescaled_value, rescaled_error = exp(BigFloat(log(value) - logweight)), exp(BigFloat(log(error) - logweight))
     result = Measurements.measurement(rescaled_value, rescaled_error)
-    return (result = result, logrenormf = logrenormf, cuba_result = r_cuba)
+    return (result = result, logweight = logweight, cuba_result = r_cuba)
 end
 
 

ext/BATNestedSamplersExt.jl

@@ -13,7 +13,7 @@ using HeterogeneousComputing
 
 BAT.pkgext(::Val{:NestedSamplers}) = BAT.PackageExtension{:NestedSamplers}()
 
-using BAT: AnyMeasureOrDensity, AbstractMeasureOrDensity
+using BAT: MeasureLike, BATMeasure
 using BAT: ENSBound, ENSNoBounds, ENSEllipsoidBound, ENSMultiEllipsoidBound
 using BAT: ENSProposal, ENSUniformly, ENSAutoProposal, ENSRandomWalk, ENSSlice 
 
@@ -63,16 +63,18 @@ end
 
 
 
-function BAT.bat_sample_impl(target::AnyMeasureOrDensity, algorithm::EllipsoidalNestedSampling, context::BATContext)
+function BAT.bat_sample_impl(m::BATMeasure, algorithm::EllipsoidalNestedSampling, context::BATContext)
     # ToDo: Forward RNG from context!
     rng = get_rng(context)
 
-    density_notrafo = convert(AbstractMeasureOrDensity, target)
-    density, trafo = BAT.transform_and_unshape(algorithm.trafo, density_notrafo, context)                 # BAT prior transformation
-    vs = varshape(density)
-    dims = totalndof(vs)
+    transformed_m, trafo = BAT.transform_and_unshape(algorithm.trafo, m, context)                 # BAT prior transformation
+    dims = totalndof(varshape(transformed_m))
 
-    model = NestedModel(logdensityof(density), identity);                                   # identity, because ahead the BAT prior transformation is used instead
+    if !BAT.has_uhc_support(transformed_m)
+        throw(ArgumentError("$algorithm doesn't measures that are not limited to the unit hypercube"))
+    end
+
+    model = NestedModel(logdensityof(transformed_m), identity);                                   # identity, because ahead the BAT prior transformation is used instead
     bounding = ENSBounding(algorithm.bound)
     prop = ENSprop(algorithm.proposal)
     sampler = Nested(
@@ -89,15 +91,15 @@ function BAT.bat_sample_impl(target::AnyMeasureOrDensity, algorithm::Ellipsoidal
     weights = samples_w[:, end]                                                             # the last elements of the vectors are the weights
     nsamples = size(samples_w,1)
     samples = [samples_w[i, 1:end-1] for i in 1:nsamples]                                   # the other ones (between 1 and end-1) are the samples
-    logvals = map(logdensityof(density), samples)                                           # posterior values of the samples
-    samples_trafo = vs.(BAT.DensitySampleVector(samples, logvals, weight = weights))
-    samples_notrafo = inverse(trafo).(samples_trafo)                                            # Here the samples are retransformed
+    logvals = map(logdensityof(transformed_m), samples)                                           # posterior values of the samples
+    transformed_smpls = BAT.DensitySampleVector(samples, logvals, weight = weights)
+    smpls = inverse(trafo).(transformed_smpls)                                            # Here the samples are retransformed
 
     logintegral = Measurements.measurement(state.logz, state.logzerr)
-    ess = bat_eff_sample_size(samples_notrafo, KishESS(), context).result
+    ess = bat_eff_sample_size(smpls, KishESS(), context).result
 
     return (
-        result = samples_notrafo, result_trafo = samples_trafo, trafo = trafo, 
+        result = smpls, result_trafo = transformed_smpls, trafo = trafo, 
         logintegral = logintegral, ess = ess,
         info = state
     )