Update turing_inference to accommodate Turing's updated syntax.

[Vaibhavdixit02]

Hi @xukai92 I saw that there have been quite a few changes in the last few months in Turing.jl and thus turing_inference needs to be updated too, could you help out with the required changes?

[xukai92]

Sure, I will update Turing-related codes.

[Vaibhavdixit02]

That would be a great help 👍

[Vaibhavdixit02]

@xukai92 Could you provide a loose timeframe in which you will be able to take a look. I will be presenting a talk at JuliaCon'18 on DiffEqBayes.jl so it would be helpful for me to plan out my work for the talk accordingly. Thanks!

[xukai92]

I will do it within two days.

[xukai92]

I was trying to upgrade related packages for this repo. but failed to build SymEngine.

=======================[ ERROR: SymEngine ]=======================

LoadError: failed process: Process(setenv(`/Users/kai/.julia/v0.6/Conda/deps/usr/bin/conda install -y symengine=0.3.0`,String["XPC_FLAGS=0x0", "LSCOLORS=Gxfxcxdxbxegedabagacad", "PATH=/Applications/Julia-0.6.app/Contents/Resources/julia/bin:/Applications/Julia-0.6.app/Contents/Resources/julia/bin:/usr/bin:/bin:/usr/sbin:/sbin:/Users/kai/sbin:/Users/kai/bin:/Users/kai/.local/sbin:/Users/kai/.local/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/opt/local/sbin:/opt/local/bin:/sbin:/bin:/opt/X11/bin:/Library/TeX/texbin:/Users/kai/cling/bin:/Users/kai/Library/Android/sdk:/Users/kai/miniconda2/bin", "PWD=/Users/kai", "DISPLAY=/private/tmp/com.apple.launchd.LyGpKQDTb9/org.macosforge.xquartz:0", "XPC_SERVICE_NAME=0", "TERM_PROGRAM=iTerm.app", "ARCHFLAGS=-arch x86_64", "SHELL=/bin/zsh", "__CF_USER_TEXT_ENCODING=0x1F5:0x0:0x2", "COLORFGBG=15;0", "LC_CTYPE=en_GB.UTF-8", "TMPDIR=/var/folders/xv/35q9th6x10b926ntt6cwgfjh0000gn/T/", "TK_LIBRARY=/System/Library/Frameworks/Tk.framework/Versions/8.5/Resources/Scripts", "CONDARC=/Users/kai/.julia/v0.6/Conda/deps/usr/condarc-julia.yml", "CPPFLAGS=-I/opt/X11/include", "FONTCONFIG_PATH=/Applications/Julia-0.6.app/Contents/Resources/julia/etc/fonts", "LANG=en_GB.UTF-8", "ITERM_PROFILE=Default", "SHLVL=2", "LOGNAME=kai", "CONDA_PREFIX=/Users/kai/.julia/v0.6/Conda/deps/usr", "TERM_SESSION_ID=w0t0p1:1CC0BEF1-AFAA-4FA6-8785-13D6D5780D0D", "SSH_AUTH_SOCK=/private/tmp/com.apple.launchd.dMhY3eLd6k/Listeners", "OLDPWD=/Users/kai", "DEFAULT_USER=kai", "_=/Users/kai/bin/julia", "PAGER=less", "Apple_PubSub_Socket_Render=/private/tmp/com.apple.launchd.NiLzn0XvJa/Render", "USER=kai", "ITERM_SESSION_ID=w0t0p1:1CC0BEF1-AFAA-4FA6-8785-13D6D5780D0D", "TERM=xterm-256color", "HOME=/Users/kai", "TERM_PROGRAM_VERSION=3.2.0beta3", "OPENBLAS_MAIN_FREE=1", "COLORTERM=truecolor", "LESS=-R", "ZSH=/Users/kai/.oh-my-zsh"]), ProcessExited(1)) [1]
while loading /Users/kai/.julia/v0.6/SymEngine/deps/build.jl, in expression starting on line 56

==================================================================

=========================[ BUILD ERRORS ]=========================

WARNING: SymEngine had build errors.

 - packages with build errors remain installed in /Users/kai/.julia/v0.6
 - build the package(s) and all dependencies with `Pkg.build("SymEngine")`
 - build a single package by running its `deps/build.jl` script

==================================================================

which leads to

ERROR: LoadError: error compiling #ode_def_opts#1: error compiling symbols: error compiling _symbol: error compiling Type: could not load library "libsymengine"
dlopen(libsymengine.dylib, 1): image not found
Stacktrace:
 [1] include_from_node1(::String) at ./loading.jl:576
 [2] include(::String) at ./sysimg.jl:14
 [3] process_options(::Base.JLOptions) at ./client.jl:305
 [4] _start() at ./client.jl:371
while loading /Users/kai/projects/DiffEqBayes.jl/test/turing.jl, in expression starting on line 4

Tried to Pkg.update, Pkg.rm and Pkg.add for that package but didn't help. Any idea how to resolve this?

[Vaibhavdixit02]

I can't recreate the error. What version of Julia and SymEngine are you using?
@ChrisRackauckas do you know something about this?

[xukai92]

I'm using Julia 0.6.2 and SymEngine 0.4.0.

[ChrisRackauckas]

@isuruf

[isuruf]

Can you run /Users/kai/.julia/v0.6/Conda/deps/usr/bin/conda install -y symengine=0.3.0 in bash and post the error here?

[xukai92]

@isuruf Here is it

➜  DiffEqBayes.jl (update-turing) ✗ /Users/kai/.julia/v0.6/Conda/deps/usr/bin/conda install -y symengine=0.3.0
Solving environment: failed

PackagesNotFoundError: The following packages are not available from current channels:

  - symengine=0.3.0

Current channels:

  - https://repo.anaconda.com/pkgs/main/osx-64
  - https://repo.anaconda.com/pkgs/main/noarch
  - https://repo.anaconda.com/pkgs/free/osx-64
  - https://repo.anaconda.com/pkgs/free/noarch
  - https://repo.anaconda.com/pkgs/r/osx-64
  - https://repo.anaconda.com/pkgs/r/noarch
  - https://repo.anaconda.com/pkgs/pro/osx-64
  - https://repo.anaconda.com/pkgs/pro/noarch

To search for alternate channels that may provide the conda package you're
looking for, navigate to

    https://anaconda.org

and use the search bar at the top of the page.

[isuruf]

Run /Users/kai/.julia/v0.6/Conda/deps/usr/bin/conda config --add channels conda-forge and then the command from my previous comment.

[xukai92]

That works! Thanks!

[ChrisRackauckas]

@mohamed82008 posted

function turing_inference(prob::DiffEqBase.DEProblem,alg,t,data,priors = nothing;
                          num_samples=1000, delta=0.65, kwargs...)

  N = length(priors)
  _theta = Vector{Real}(undef, N)
  @model bif(x) = begin
    for i in 1:length(priors)
      _theta[i] ~ priors[i]
    end
    σ ~ InverseGamma(2, 3)

    theta = convert(Array{typeof(first(_theta))}, _theta)
    p_tmp = remake(prob, u0 = convert.(eltype(theta), (prob.u0)), p = theta) 
    sol_tmp = solve(p_tmp, alg; saveat = t, kwargs...)
    fill_length = length(t) - length(sol_tmp.u)

    for i in 1:fill_length
      if eltype(sol_tmp.u) <: Number
        push!(sol_tmp.u, Inf)
      else
        push!(sol_tmp.u, fill(Inf, size(sol_tmp[1])))
      end
    end
    for i = 1:length(t)
      res = sol_tmp.u[i]
      x[:,i] ~ MvNormal(res, σ*ones(2))
    end
  end

  model = bif(data)
  chn = sample(model, Turing.NUTS(num_samples, delta))
end