search

STOR-i / GaussianProcesses.jl

A Julia package for Gaussian Processes
https://stor-i.github.io/GaussianProcesses.jl/latest/
Other
308 stars 53 forks source link

Error when trying to optimize GPR example #211

Closed Cyberface closed 2 years ago

Cyberface commented 2 years ago

Hi,

I'm following the GPR example here but when I try to run

optimize!(gp)

I get the following error

InexactError: Int64(103.91337059292768)

Stacktrace:
  [1] (::GaussianProcesses.var"#ltarget_and_dltarget!#102"{Base.Iterators.Pairs{Symbol, Bool, Tuple{Symbol, Symbol, Symbol}, NamedTuple{(:kern, :domean, :noise), Tuple{Bool, Bool, Bool}}}, GPE{LinearAlgebra.Adjoint{Float64, Vector{Float64}}, Vector{Float64}, MeanZero, SEIso{Float64}, GaussianProcesses.FullCovariance, GaussianProcesses.IsotropicData{Matrix{Float64}}, PDMats.PDMat{Float64, Matrix{Float64}}, GaussianProcesses.Scalar{Int64}}})(grad::Vector{Float64}, hyp::Vector{Float64})
    @ GaussianProcesses ~/.julia/packages/GaussianProcesses/kzIIW/src/optimize.jl:85
  [2] value_gradient!!(obj::OnceDifferentiable{Float64, Vector{Float64}, Vector{Float64}}, x::Vector{Float64})
    @ NLSolversBase ~/.julia/packages/NLSolversBase/GRQ1x/src/interface.jl:82
  [3] value_gradient!(obj::OnceDifferentiable{Float64, Vector{Float64}, Vector{Float64}}, x::Vector{Float64})
    @ NLSolversBase ~/.julia/packages/NLSolversBase/GRQ1x/src/interface.jl:69
  [4] value_gradient!(obj::Optim.ManifoldObjective{OnceDifferentiable{Float64, Vector{Float64}, Vector{Float64}}}, x::Vector{Float64})
    @ Optim ~/.julia/packages/Optim/3K7JI/src/Manifolds.jl:50
  [5] (::LineSearches.var"#ϕdϕ#6"{Optim.ManifoldObjective{OnceDifferentiable{Float64, Vector{Float64}, Vector{Float64}}}, Vector{Float64}, Vector{Float64}, Vector{Float64}})(α::Float64)
    @ LineSearches ~/.julia/packages/LineSearches/Ki4c5/src/LineSearches.jl:84
  [6] (::LineSearches.HagerZhang{Float64, Base.RefValue{Bool}})(ϕ::Function, ϕdϕ::LineSearches.var"#ϕdϕ#6"{Optim.ManifoldObjective{OnceDifferentiable{Float64, Vector{Float64}, Vector{Float64}}}, Vector{Float64}, Vector{Float64}, Vector{Float64}}, c::Float64, phi_0::Float64, dphi_0::Float64)
    @ LineSearches ~/.julia/packages/LineSearches/Ki4c5/src/hagerzhang.jl:139
  [7] HagerZhang
    @ ~/.julia/packages/LineSearches/Ki4c5/src/hagerzhang.jl:101 [inlined]
  [8] perform_linesearch!(state::Optim.LBFGSState{Vector{Float64}, Vector{Vector{Float64}}, Vector{Vector{Float64}}, Float64, Vector{Float64}}, method::LBFGS{Nothing, LineSearches.InitialStatic{Float64}, LineSearches.HagerZhang{Float64, Base.RefValue{Bool}}, Optim.var"#17#19"}, d::Optim.ManifoldObjective{OnceDifferentiable{Float64, Vector{Float64}, Vector{Float64}}})
    @ Optim ~/.julia/packages/Optim/3K7JI/src/utilities/perform_linesearch.jl:59
  [9] update_state!(d::OnceDifferentiable{Float64, Vector{Float64}, Vector{Float64}}, state::Optim.LBFGSState{Vector{Float64}, Vector{Vector{Float64}}, Vector{Vector{Float64}}, Float64, Vector{Float64}}, method::LBFGS{Nothing, LineSearches.InitialStatic{Float64}, LineSearches.HagerZhang{Float64, Base.RefValue{Bool}}, Optim.var"#17#19"})
    @ Optim ~/.julia/packages/Optim/3K7JI/src/multivariate/solvers/first_order/l_bfgs.jl:204
 [10] optimize(d::OnceDifferentiable{Float64, Vector{Float64}, Vector{Float64}}, initial_x::Vector{Float64}, method::LBFGS{Nothing, LineSearches.InitialStatic{Float64}, LineSearches.HagerZhang{Float64, Base.RefValue{Bool}}, Optim.var"#17#19"}, options::Optim.Options{Float64, Nothing}, state::Optim.LBFGSState{Vector{Float64}, Vector{Vector{Float64}}, Vector{Vector{Float64}}, Float64, Vector{Float64}})
    @ Optim ~/.julia/packages/Optim/3K7JI/src/multivariate/optimize/optimize.jl:57
 [11] optimize(d::OnceDifferentiable{Float64, Vector{Float64}, Vector{Float64}}, initial_x::Vector{Float64}, method::LBFGS{Nothing, LineSearches.InitialStatic{Float64}, LineSearches.HagerZhang{Float64, Base.RefValue{Bool}}, Optim.var"#17#19"}, options::Optim.Options{Float64, Nothing}) (repeats 2 times)
    @ Optim ~/.julia/packages/Optim/3K7JI/src/multivariate/optimize/optimize.jl:35
 [12] optimize!(::GPE{LinearAlgebra.Adjoint{Float64, Vector{Float64}}, Vector{Float64}, MeanZero, SEIso{Float64}, GaussianProcesses.FullCovariance, GaussianProcesses.IsotropicData{Matrix{Float64}}, PDMats.PDMat{Float64, Matrix{Float64}}, GaussianProcesses.Scalar{Int64}}, ::LBFGS{Nothing, LineSearches.InitialStatic{Float64}, LineSearches.HagerZhang{Float64, Base.RefValue{Bool}}, Optim.var"#17#19"}; domean::Bool, kern::Bool, noise::Bool, lik::Bool, meanbounds::Nothing, kernbounds::Nothing, noisebounds::Nothing, likbounds::Nothing, kwargs::Base.Iterators.Pairs{Union{}, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})
    @ GaussianProcesses ~/.julia/packages/GaussianProcesses/kzIIW/src/optimize.jl:28
 [13] optimize! (repeats 2 times)
    @ ~/.julia/packages/GaussianProcesses/kzIIW/src/optimize.jl:23 [inlined]
 [14] top-level scope
    @ In[10]:1
 [15] eval
    @ ./boot.jl:360 [inlined]
 [16] include_string(mapexpr::typeof(REPL.softscope), mod::Module, code::String, filename::String)
    @ Base ./loading.jl:1116

Any ideas?

I'm quite new to Julia and don't have much experience debugging sorry.

Thanks!

Cyberface commented 2 years ago

I found the error.

I was using an integer for logObsNoise = -1.

Simply changing this to a float -> logObsNoise = -1.0 fixes my problem.