Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix indexing for chains in different threads #154

Merged
merged 3 commits into from
Jan 10, 2025
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion Project.toml
Original file line number Diff line number Diff line change
Expand Up @@ -3,7 +3,7 @@ uuid = "80f14c24-f653-4e6a-9b94-39d6b0f70001"
keywords = ["markov chain monte carlo", "probabilistic programming"]
license = "MIT"
desc = "A lightweight interface for common MCMC methods."
version = "5.6.0"
version = "5.6.1"

[deps]
BangBang = "198e06fe-97b7-11e9-32a5-e1d131e6ad66"
Expand Down
18 changes: 13 additions & 5 deletions src/sample.jl
Original file line number Diff line number Diff line change
Expand Up @@ -391,13 +391,16 @@ function mcmcsample(

# Copy the random number generator, model, and sample for each thread
nchunks = min(nchains, Threads.nthreads())
chunksize = cld(nchains, nchunks)
interval = 1:nchunks
# `copy` instead of `deepcopy` for RNGs: https://github.com/JuliaLang/julia/issues/42899
rngs = [copy(rng) for _ in interval]
models = [deepcopy(model) for _ in interval]
samplers = [deepcopy(sampler) for _ in interval]

# If nchains/nchunks = m with remainder n, then the first n chunks will
# have m + 1 chains, and the rest will have m chains.
m, n = divrem(nchains, nchunks)

# Create a seed for each chain using the provided random number generator.
seeds = rand(rng, UInt, nchains)

Expand Down Expand Up @@ -437,12 +440,17 @@ function mcmcsample(
Distributed.@async begin
try
Distributed.@sync for (i, _rng, _model, _sampler) in
zip(1:nchunks, rngs, models, samplers)
chainidxs = if i == nchunks
((i - 1) * chunksize + 1):nchains
zip(interval, rngs, models, samplers)
if i <= n
chainidx_hi = i * (m + 1)
nchains_chunk = m + 1
else
((i - 1) * chunksize + 1):(i * chunksize)
chainidx_hi = i * m + n # n * (m + 1) + (i - n) * m
nchains_chunk = m
end
chainidx_lo = chainidx_hi - nchains_chunk + 1
chainidxs = chainidx_lo:chainidx_hi

Threads.@spawn for chainidx in chainidxs
# Seed the chunk-specific random number generator with the pre-made seed.
Random.seed!(_rng, seeds[chainidx])
Expand Down
Loading