Running Pathfinder on Turing.jl models

This tutorial demonstrates how Turing can be used with Pathfinder.

We'll demonstrate with a regression example.

using AbstractMCMC, AdvancedHMC, DynamicPPL, FlexiChains, Pathfinder, Random, Turing
Random.seed!(39)

@model function regress(x)
    α ~ Normal()
    β ~ Normal()
    σ ~ truncated(Normal(); lower=0)
    μ = α .+ β .* x
    y ~ product_distribution(Normal.(μ, σ))
end
x = 0:0.1:10
true_params = (; α=1.5, β=2, σ=2)
# simulate data
y = rand(regress(x) | true_params)[@varname(y)]

model = regress(x) | (; y)
n_chains = 8

For convenience, pathfinder and multipathfinder can take Turing models as inputs and produce MCMCChains.Chains or FlexiChains.VNChain objects as outputs. To access this, we run Pathfinder normally; the chains representation of the draws (defaulting to Chains) is stored in draws_transformed.

result_single = pathfinder(model; ndraws=1_000)
Single-path Pathfinder result
  tries: 1
  draws: 1000
  fit iteration: 14 (total: 16)
  fit ELBO: -213.66 ± 0.09
  fit distribution: MvNormal{Float64, Pathfinder.WoodburyPDMat{Float64, LinearAlgebra.Diagonal{Float64, Vector{Float64}}, Matrix{Float64}, Matrix{Float64}, Pathfinder.WoodburyPDFactorization{Float64, LinearAlgebra.Diagonal{Float64, Vector{Float64}}, LinearAlgebra.QRCompactWYQ{Float64, Matrix{Float64}, Matrix{Float64}}, LinearAlgebra.UpperTriangular{Float64, Matrix{Float64}}}}, Vector{Float64}}(
dim: 3
μ: [1.6508971085750253, 1.9311753921107158, 0.5801261338729266]
Σ: [0.11087798427694032 -0.016527256928018597 -0.0014208078651489035; -0.016527256928018624 0.0034071472308229607 0.00019433068465847872; -0.0014208078651489287 0.00019433068465846744 0.004718299969699233]
)
result_single.draws_transformed
Chains MCMC chain (1000×6×1 Array{Float64, 3}):

Iterations        = 1:1:1000
Number of chains  = 1
Samples per chain = 1000
parameters        = α, β, σ
internals         = logprior, loglikelihood, logjoint

Use `describe(chains)` for summary statistics and quantiles.

To request a different chain type (e.g. VNChain), we can specify the chain_type directly.

pathfinder(model; ndraws=1_000, chain_type=VNChain).draws_transformed
FlexiChain (1000 iterations, 1 chain)
↓ iter=1:1000 | → chain=1:1

Parameter type   VarName
Parameters       α, β, σ
Extra keys       :logprior, :loglikelihood, :logjoint

Note that while Turing's sample methods default to initializing parameters from the prior with InitFromPrior, Pathfinder defaults to uniformly sampling them in the range [-2, 2] in unconstrained space (equivalent to Turing's InitFromUniform(-2, 2)). To use Turing's default in Pathfinder, specify init_sampler=InitFromPrior().

result_multi = multipathfinder(model, 1_000; nruns=n_chains, init_sampler=InitFromPrior())
Multi-path Pathfinder result
  runs: 8
  draws: 1000
  Pareto shape diagnostic: 0.39 (good)

The Pareto shape diagnostic indicates that it is likely safe to use these draws to compute posterior estimates.

chns_pf = result_multi.draws_transformed
describe(chns_pf)
Chains MCMC chain (1000×6×1 Array{Float64, 3}):

Iterations        = 1:1:1000
Number of chains  = 1
Samples per chain = 1000
parameters        = α, β, σ
internals         = logprior, loglikelihood, logjoint

Summary Statistics

  parameters      mean       std      mcse   ess_bulk   ess_tail      rhat   ess_per_sec 
      Symbol   Float64   Float64   Float64    Float64    Float64   Float64       Missing 

           α    1.6542    0.3244    0.0104   976.4619   904.1276    1.0026       missing
           β    1.9309    0.0586    0.0019   948.4244   966.2907    1.0007       missing
           σ    1.8086    0.1246    0.0041   908.0277   944.1954    1.0031       missing


Quantiles

  parameters      2.5%     25.0%     50.0%     75.0%     97.5% 
      Symbol   Float64   Float64   Float64   Float64   Float64 

           α    1.0317    1.4321    1.6640    1.8631    2.2388
           β    1.8116    1.8937    1.9278    1.9706    2.0448
           σ    1.6006    1.7206    1.8007    1.8855    2.0972

We can also use these draws to initialize MCMC sampling with InitFromParams.

params = AbstractMCMC.to_samples(DynamicPPL.ParamsWithStats, chns_pf[1:n_chains, :, :], model)
initial_params = [InitFromParams(p.params) for p in vec(params)]
chns = sample(model, Turing.NUTS(), MCMCThreads(), 1_000, n_chains; initial_params, progress=false)
describe(chns)
┌ Warning: Only a single thread available: MCMC chains are not sampled in parallel
@ AbstractMCMC ~/.julia/packages/AbstractMCMC/oqm6Y/src/sample.jl:544
┌ Info: Found initial step size
  ϵ = 0.025
┌ Info: Found initial step size
  ϵ = 0.05
┌ Info: Found initial step size
  ϵ = 0.025
┌ Info: Found initial step size
  ϵ = 0.05
┌ Info: Found initial step size
  ϵ = 0.2
┌ Info: Found initial step size
  ϵ = 0.05
┌ Info: Found initial step size
  ϵ = 0.05
┌ Info: Found initial step size
  ϵ = 0.025
Chains MCMC chain (1000×17×8 Array{Float64, 3}):

Iterations        = 501:1:1500
Number of chains  = 8
Samples per chain = 1000
Wall duration     = 4.43 seconds
Compute duration  = 2.98 seconds
parameters        = α, β, σ
internals         = n_steps, is_accept, acceptance_rate, log_density, hamiltonian_energy, hamiltonian_energy_error, max_hamiltonian_energy_error, tree_depth, numerical_error, step_size, nom_step_size, logprior, loglikelihood, logjoint

Summary Statistics

  parameters      mean       std      mcse    ess_bulk    ess_tail      rhat   ess_per_sec 
      Symbol   Float64   Float64   Float64     Float64     Float64   Float64       Float64 

           α    1.6387    0.3304    0.0052   4009.3111   4441.1806    1.0035     1344.5040
           β    1.9331    0.0580    0.0009   4049.6772   4048.3682    1.0032     1358.0406
           σ    1.8170    0.1289    0.0019   4717.5282   4372.3596    1.0005     1582.0014


Quantiles

  parameters      2.5%     25.0%     50.0%     75.0%     97.5% 
      Symbol   Float64   Float64   Float64   Float64   Float64 

           α    0.9950    1.4127    1.6399    1.8645    2.2844
           β    1.8200    1.8927    1.9334    1.9723    2.0453
           σ    1.5885    1.7267    1.8095    1.9006    2.0847

We can use Pathfinder's estimate of the metric and only perform enough warm-up to tune the step size.

inv_metric = result_multi.pathfinder_results[1].fit_distribution.Σ
metric = Pathfinder.RankUpdateEuclideanMetric(inv_metric)
kernel = HMCKernel(Trajectory{MultinomialTS}(Leapfrog(0.0), GeneralisedNoUTurn()))
adaptor = StepSizeAdaptor(0.8, 1.0)  # adapt only the step size
nuts = AdvancedHMC.HMCSampler(kernel, metric, adaptor)

n_adapts = 50
n_draws = 1_000
chns = sample(
    model,
    externalsampler(nuts),
    MCMCThreads(),
    n_draws + n_adapts,
    n_chains;
    n_adapts,
    initial_params,
    progress=false,
)[n_adapts + 1:end, :, :]  # drop warm-up draws
describe(chns)
┌ Warning: Only a single thread available: MCMC chains are not sampled in parallel
@ AbstractMCMC ~/.julia/packages/AbstractMCMC/oqm6Y/src/sample.jl:544
[ Info: Found initial step size 1.6
[ Info: Found initial step size 1.6
[ Info: Found initial step size 3.2
[ Info: Found initial step size 3.2
[ Info: Found initial step size 3.2
[ Info: Found initial step size 1.6
[ Info: Found initial step size 3.2
[ Info: Found initial step size 1.7000000000000002
Chains MCMC chain (1000×17×8 Array{Float64, 3}):

Iterations        = 51:1:1050
Number of chains  = 8
Samples per chain = 1000
Wall duration     = 1.41 seconds
Compute duration  = 1.25 seconds
parameters        = α, β, σ
internals         = n_steps, is_accept, acceptance_rate, log_density, hamiltonian_energy, hamiltonian_energy_error, max_hamiltonian_energy_error, tree_depth, numerical_error, step_size, nom_step_size, logprior, loglikelihood, logjoint

Summary Statistics

  parameters      mean       std      mcse     ess_bulk    ess_tail      rhat   ess_per_sec 
      Symbol   Float64   Float64   Float64      Float64     Float64   Float64       Float64 

           α    1.6446    0.3326    0.0032   10751.4061   6051.0525    1.0002     8580.5316
           β    1.9315    0.0585    0.0006   10493.5935   6011.7122    1.0007     8374.7754
           σ    1.8144    0.1273    0.0013    9918.0779   5971.6501    1.0011     7915.4652


Quantiles

  parameters      2.5%     25.0%     50.0%     75.0%     97.5% 
      Symbol   Float64   Float64   Float64   Float64   Float64 

           α    0.9826    1.4201    1.6510    1.8688    2.3024
           β    1.8160    1.8924    1.9317    1.9703    2.0472
           σ    1.5886    1.7245    1.8056    1.8967    2.0824

See Initializing HMC with Pathfinder for further examples.