Hi! I’m trying to sample from a MatrixNormal in Turing.jl running the model below (I’m just playing around with matrix distributions there is no deeper reason for this). However I keep running into the following error (which might be related to this Github issue):
ERROR: MethodError: no method matching bijector(::MatrixNormal{Float64, Matrix{Float64}, PDMats.PDMat{Float64, Matrix{Float64}}, PDMats.PDMat{Float64, Matrix{Float64}}})
Closest candidates are:
bijector(::Union{Kolmogorov, BetaPrime, Chi, Chisq, Erlang, Exponential, FDist, Frechet, Gamma, InverseGamma, InverseGaussian, LogNormal, NoncentralChisq, NoncentralF, Rayleigh, Weibull}) at ~/.julia/packages/Bijectors/vUc4m/src/transformed_distribution.jl:58
bijector(::Union{Arcsine, Beta, Biweight, Cosine, Epanechnikov, NoncentralBeta}) at ~/.julia/packages/Bijectors/vUc4m/src/transformed_distribution.jl:69
bijector(::Union{Levy, Pareto}) at ~/.julia/packages/Bijectors/vUc4m/src/transformed_distribution.jl:72
...
Stacktrace:
[1] macro expansion
@ ~/.julia/packages/DynamicPPL/xKo8W/src/varinfo.jl:784 [inlined]
[2] _link!(metadata::NamedTuple{(:mu,), Tuple{DynamicPPL.Metadata{Dict{AbstractPPL.VarName{:mu, Setfield.IdentityLens}, Int64}, Vector{MatrixNormal{Float64, Matrix{Float64}, PDMats.PDMat{Float64, Matrix{Float64}}, PDMats.PDMat{Float64, Matrix{Float64}}}}, Vector{AbstractPPL.VarName{:mu, Setfield.IdentityLens}}, Vector{Float64}, Vector{Set{DynamicPPL.Selector}}}}}, vi::DynamicPPL.TypedVarInfo{NamedTuple{(:mu,), Tuple{DynamicPPL.Metadata{Dict{AbstractPPL.VarName{:mu, Setfield.IdentityLens}, Int64}, Vector{MatrixNormal{Float64, Matrix{Float64}, PDMats.PDMat{Float64, Matrix{Float64}}, PDMats.PDMat{Float64, Matrix{Float64}}}}, Vector{AbstractPPL.VarName{:mu, Setfield.IdentityLens}}, Vector{Float64}, Vector{Set{DynamicPPL.Selector}}}}}, Float64}, vns::NamedTuple{(:mu,), Tuple{Vector{AbstractPPL.VarName{:mu, Setfield.IdentityLens}}}}, #unused#::Val{()})
@ DynamicPPL ~/.julia/packages/DynamicPPL/xKo8W/src/varinfo.jl:768
[3] _link!
@ ~/.julia/packages/DynamicPPL/xKo8W/src/varinfo.jl:766 [inlined]
[4] _link!
@ ~/.julia/packages/DynamicPPL/xKo8W/src/varinfo.jl:762 [inlined]
[5] link!!
@ ~/.julia/packages/DynamicPPL/xKo8W/src/varinfo.jl:717 [inlined]
[6] link!!
@ ~/.julia/packages/DynamicPPL/xKo8W/src/abstract_varinfo.jl:383 [inlined]
[7] initialstep(rng::Random._GLOBAL_RNG, model::DynamicPPL.Model{typeof(estimateA), (:A, :U, :V, :Uₐ, :Vₐ), (), (), NTuple{5, Matrix{Float64}}, Tuple{}, DynamicPPL.DefaultContext}, spl::DynamicPPL.Sampler{NUTS{Turing.Essential.ForwardDiffAD{0}, (), AdvancedHMC.DiagEuclideanMetric}}, vi::DynamicPPL.TypedVarInfo{NamedTuple{(:mu,), Tuple{DynamicPPL.Metadata{Dict{AbstractPPL.VarName{:mu, Setfield.IdentityLens}, Int64}, Vector{MatrixNormal{Float64, Matrix{Float64}, PDMats.PDMat{Float64, Matrix{Float64}}, PDMats.PDMat{Float64, Matrix{Float64}}}}, Vector{AbstractPPL.VarName{:mu, Setfield.IdentityLens}}, Vector{Float64}, Vector{Set{DynamicPPL.Selector}}}}}, Float64}; init_params::Nothing, nadapts::Int64, kwargs::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})
@ Turing.Inference ~/.julia/packages/Turing/KOb5J/src/inference/hmc.jl:153
[8] step(rng::Random._GLOBAL_RNG, model::DynamicPPL.Model{typeof(estimateA), (:A, :U, :V, :Uₐ, :Vₐ), (), (), NTuple{5, Matrix{Float64}}, Tuple{}, DynamicPPL.DefaultContext}, spl::DynamicPPL.Sampler{NUTS{Turing.Essential.ForwardDiffAD{0}, (), AdvancedHMC.DiagEuclideanMetric}}; resume_from::Nothing, init_params::Nothing, kwargs::Base.Pairs{Symbol, Int64, Tuple{Symbol}, NamedTuple{(:nadapts,), Tuple{Int64}}})
@ DynamicPPL ~/.julia/packages/DynamicPPL/xKo8W/src/sampler.jl:111
[9] macro expansion
@ ~/.julia/packages/AbstractMCMC/fnRmh/src/sample.jl:120 [inlined]
[10] macro expansion
@ ~/.julia/packages/ProgressLogging/6KXlp/src/ProgressLogging.jl:328 [inlined]
[11] macro expansion
@ ~/.julia/packages/AbstractMCMC/fnRmh/src/logging.jl:9 [inlined]
[12] mcmcsample(rng::Random._GLOBAL_RNG, model::DynamicPPL.Model{typeof(estimateA), (:A, :U, :V, :Uₐ, :Vₐ), (), (), NTuple{5, Matrix{Float64}}, Tuple{}, DynamicPPL.DefaultContext}, sampler::DynamicPPL.Sampler{NUTS{Turing.Essential.ForwardDiffAD{0}, (), AdvancedHMC.DiagEuclideanMetric}}, N::Int64; progress::Bool, progressname::String, callback::Nothing, discard_initial::Int64, thinning::Int64, chain_type::Type, kwargs::Base.Pairs{Symbol, Int64, Tuple{Symbol}, NamedTuple{(:nadapts,), Tuple{Int64}}})
@ AbstractMCMC ~/.julia/packages/AbstractMCMC/fnRmh/src/sample.jl:111
[13] sample(rng::Random._GLOBAL_RNG, model::DynamicPPL.Model{typeof(estimateA), (:A, :U, :V, :Uₐ, :Vₐ), (), (), NTuple{5, Matrix{Float64}}, Tuple{}, DynamicPPL.DefaultContext}, sampler::DynamicPPL.Sampler{NUTS{Turing.Essential.ForwardDiffAD{0}, (), AdvancedHMC.DiagEuclideanMetric}}, N::Int64; chain_type::Type, resume_from::Nothing, progress::Bool, nadapts::Int64, discard_adapt::Bool, discard_initial::Int64, kwargs::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})
@ Turing.Inference ~/.julia/packages/Turing/KOb5J/src/inference/hmc.jl:133
[14] sample
@ ~/.julia/packages/Turing/KOb5J/src/inference/hmc.jl:103 [inlined]
[15] #sample#2
@ ~/.julia/packages/Turing/KOb5J/src/inference/Inference.jl:145 [inlined]
[16] sample
@ ~/.julia/packages/Turing/KOb5J/src/inference/Inference.jl:138 [inlined]
[17] #sample#1
@ ~/.julia/packages/Turing/KOb5J/src/inference/Inference.jl:135 [inlined]
[18] sample(model::DynamicPPL.Model{typeof(estimateA), (:A, :U, :V, :Uₐ, :Vₐ), (), (), NTuple{5, Matrix{Float64}}, Tuple{}, DynamicPPL.DefaultContext}, alg::NUTS{Turing.Essential.ForwardDiffAD{0}, (), AdvancedHMC.DiagEuclideanMetric}, N::Int64)
@ Turing.Inference ~/.julia/packages/Turing/KOb5J/src/inference/Inference.jl:129
[19] top-level scope
@ ~/RandomStuff/scratch.jl:133
this is the model I’m using:
using Random, Turing, Bijectors
Random.seed!(123)
#Estimate a MatrixNormal as simulated here
U = rand(LKJ(2, 0.5))
V = rand(LKJ(2, 0.5))
Uₐ = rand(LKJ(2, 0.5))
Vₐ = rand(LKJ(2, 0.5))
Asample = rand(MatrixNormal(zeros(Float64, 2, 2), U, V))
#Create the model
@model function estimateA(A, U, V, Uₐ, Vₐ)
mu ~ MatrixNormal(zeros(Float64,size(A,1), size(A,2)), Uₐ, Vₐ)
A ~ MatrixNormal(mu, U, V)
end
#Estimate!
model = estimateA(Asample, U, V, Uₐ, Vₐ);
#Bijectors.bijector(::MatrixNormal) = PDBijector()
chains = sample(model, NUTS(), 100);
I’ve also tried the suggestion of the related Github issue to add the line Bijectors.bijector(::MatrixNormal) = PDBijector() (uncomment above) however that also results in an error:
ERROR: UndefVarError: PDBijector not defined
Stacktrace:
[1] bijector(#unused#::MatrixNormal{Float64, Matrix{Float64}, PDMats.PDMat{Float64, Matrix{Float64}}, PDMats.PDMat{Float64, Matrix{Float64}}})
@ Main ~/RandomStuff/scratch.jl:133
[2] macro expansion
@ ~/.julia/packages/DynamicPPL/xKo8W/src/varinfo.jl:784 [inlined]
[3] _link!(metadata::NamedTuple{(:mu,), Tuple{DynamicPPL.Metadata{Dict{AbstractPPL.VarName{:mu, Setfield.IdentityLens}, Int64}, Vector{MatrixNormal{Float64, Matrix{Float64}, PDMats.PDMat{Float64, Matrix{Float64}}, PDMats.PDMat{Float64, Matrix{Float64}}}}, Vector{AbstractPPL.VarName{:mu, Setfield.IdentityLens}}, Vector{Float64}, Vector{Set{DynamicPPL.Selector}}}}}, vi::DynamicPPL.TypedVarInfo{NamedTuple{(:mu,), Tuple{DynamicPPL.Metadata{Dict{AbstractPPL.VarName{:mu, Setfield.IdentityLens}, Int64}, Vector{MatrixNormal{Float64, Matrix{Float64}, PDMats.PDMat{Float64, Matrix{Float64}}, PDMats.PDMat{Float64, Matrix{Float64}}}}, Vector{AbstractPPL.VarName{:mu, Setfield.IdentityLens}}, Vector{Float64}, Vector{Set{DynamicPPL.Selector}}}}}, Float64}, vns::NamedTuple{(:mu,), Tuple{Vector{AbstractPPL.VarName{:mu, Setfield.IdentityLens}}}}, #unused#::Val{()})
@ DynamicPPL ~/.julia/packages/DynamicPPL/xKo8W/src/varinfo.jl:768
[4] _link!
@ ~/.julia/packages/DynamicPPL/xKo8W/src/varinfo.jl:766 [inlined]
[5] _link!
@ ~/.julia/packages/DynamicPPL/xKo8W/src/varinfo.jl:762 [inlined]
[6] link!!
@ ~/.julia/packages/DynamicPPL/xKo8W/src/varinfo.jl:717 [inlined]
[7] link!!
@ ~/.julia/packages/DynamicPPL/xKo8W/src/abstract_varinfo.jl:383 [inlined]
[8] initialstep(rng::Random._GLOBAL_RNG, model::DynamicPPL.Model{typeof(estimateA), (:A, :U, :V, :Uₐ, :Vₐ), (), (), NTuple{5, Matrix{Float64}}, Tuple{}, DynamicPPL.DefaultContext}, spl::DynamicPPL.Sampler{NUTS{Turing.Essential.ForwardDiffAD{0}, (), AdvancedHMC.DiagEuclideanMetric}}, vi::DynamicPPL.TypedVarInfo{NamedTuple{(:mu,), Tuple{DynamicPPL.Metadata{Dict{AbstractPPL.VarName{:mu, Setfield.IdentityLens}, Int64}, Vector{MatrixNormal{Float64, Matrix{Float64}, PDMats.PDMat{Float64, Matrix{Float64}}, PDMats.PDMat{Float64, Matrix{Float64}}}}, Vector{AbstractPPL.VarName{:mu, Setfield.IdentityLens}}, Vector{Float64}, Vector{Set{DynamicPPL.Selector}}}}}, Float64}; init_params::Nothing, nadapts::Int64, kwargs::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})
@ Turing.Inference ~/.julia/packages/Turing/KOb5J/src/inference/hmc.jl:153
[9] step(rng::Random._GLOBAL_RNG, model::DynamicPPL.Model{typeof(estimateA), (:A, :U, :V, :Uₐ, :Vₐ), (), (), NTuple{5, Matrix{Float64}}, Tuple{}, DynamicPPL.DefaultContext}, spl::DynamicPPL.Sampler{NUTS{Turing.Essential.ForwardDiffAD{0}, (), AdvancedHMC.DiagEuclideanMetric}}; resume_from::Nothing, init_params::Nothing, kwargs::Base.Pairs{Symbol, Int64, Tuple{Symbol}, NamedTuple{(:nadapts,), Tuple{Int64}}})
@ DynamicPPL ~/.julia/packages/DynamicPPL/xKo8W/src/sampler.jl:111
[10] macro expansion
@ ~/.julia/packages/AbstractMCMC/fnRmh/src/sample.jl:120 [inlined]
[11] macro expansion
@ ~/.julia/packages/ProgressLogging/6KXlp/src/ProgressLogging.jl:328 [inlined]
[12] macro expansion
@ ~/.julia/packages/AbstractMCMC/fnRmh/src/logging.jl:9 [inlined]
[13] mcmcsample(rng::Random._GLOBAL_RNG, model::DynamicPPL.Model{typeof(estimateA), (:A, :U, :V, :Uₐ, :Vₐ), (), (), NTuple{5, Matrix{Float64}}, Tuple{}, DynamicPPL.DefaultContext}, sampler::DynamicPPL.Sampler{NUTS{Turing.Essential.ForwardDiffAD{0}, (), AdvancedHMC.DiagEuclideanMetric}}, N::Int64; progress::Bool, progressname::String, callback::Nothing, discard_initial::Int64, thinning::Int64, chain_type::Type, kwargs::Base.Pairs{Symbol, Int64, Tuple{Symbol}, NamedTuple{(:nadapts,), Tuple{Int64}}})
@ AbstractMCMC ~/.julia/packages/AbstractMCMC/fnRmh/src/sample.jl:111
[14] sample(rng::Random._GLOBAL_RNG, model::DynamicPPL.Model{typeof(estimateA), (:A, :U, :V, :Uₐ, :Vₐ), (), (), NTuple{5, Matrix{Float64}}, Tuple{}, DynamicPPL.DefaultContext}, sampler::DynamicPPL.Sampler{NUTS{Turing.Essential.ForwardDiffAD{0}, (), AdvancedHMC.DiagEuclideanMetric}}, N::Int64; chain_type::Type, resume_from::Nothing, progress::Bool, nadapts::Int64, discard_adapt::Bool, discard_initial::Int64, kwargs::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})
@ Turing.Inference ~/.julia/packages/Turing/KOb5J/src/inference/hmc.jl:133
[15] sample
@ ~/.julia/packages/Turing/KOb5J/src/inference/hmc.jl:103 [inlined]
[16] #sample#2
@ ~/.julia/packages/Turing/KOb5J/src/inference/Inference.jl:145 [inlined]
[17] sample
@ ~/.julia/packages/Turing/KOb5J/src/inference/Inference.jl:138 [inlined]
[18] #sample#1
@ ~/.julia/packages/Turing/KOb5J/src/inference/Inference.jl:135 [inlined]
[19] sample(model::DynamicPPL.Model{typeof(estimateA), (:A, :U, :V, :Uₐ, :Vₐ), (), (), NTuple{5, Matrix{Float64}}, Tuple{}, DynamicPPL.DefaultContext}, alg::NUTS{Turing.Essential.ForwardDiffAD{0}, (), AdvancedHMC.DiagEuclideanMetric}, N::Int64)
@ Turing.Inference ~/.julia/packages/Turing/KOb5J/src/inference/Inference.jl:129
[20] top-level scope
@ ~/RandomStuff/scratch.jl:134
The solution might be obvious but I’m relatively new to Turing. Can anyone help me? Thanks!
Update
The solution might be that Turing.org has not defined the bijector class for this case as the following also fails (same error):
using Bijectors, Distributions
dist = MatrixNormal(zeros(2,2), rand(LKJ(2, 0.5)), rand(LKJ(2, 0.5)))
x = rand(dist)
b = bijector(dist)
however I’m unfamiliar with how to create a ad-hoc new bijector.
PS
Yes I know I can vectorice the matrix and built it from an MvNormal using the propererty in the Wikipedia definition. However I’m trying to directly sample from the MatrixNormal.