Add compile-time options to change default LogProbType

[penelopeysm]

With this PR:

julia> using DynamicPPL

julia> set_logprob_type!(Float32) # Then restart session
[ Info: DynamicPPL's log probability type has been set to Float32; please note you will need to restart your Julia session for this change to take effect.

julia> using DynamicPPL, LogDensityProblems, ForwardDiff, Distributions, ADTypes

julia> @model function f()
           x ~ Normal(0.0f0, 1.0f0)
       end
f (generic function with 2 methods)

julia> typeof(rand(f())[@varname(x)])
Float32

julia> typeof(logjoint(f(), (; x=0.0f0)))
Float32

julia> ldf = LogDensityFunction(f(), getlogjoint_internal, LinkAll(); adtype=AutoForwardDiff());

julia> typeof(rand(ldf))
Vector{Float32} (alias for Array{Float32, 1})

julia> typeof(LogDensityProblems.logdensity(ldf, [0.0f0]))
Float32

julia> typeof(LogDensityProblems.logdensity_and_gradient(ldf, [0.0f0]))
Tuple{Float32, Vector{Float32}}

Prior to this PR, logjoint and LogDensityProblems.logdensity would return Float64 instead. See TuringLang/Turing.jl#2212.

[github-actions]

Benchmark Report

• this PR's head: 9f2975d2c43d2f7281af8625b605c57e137cb689
• base branch: 34b8230ac30bb948798e58ec95adb65a9fbad4b4

Computer Information

Julia Version 1.11.9
Commit 53a02c0720c (2026-02-06 00:27 UTC)
Build Info:
  Official https://julialang.org/ release
Platform Info:
  OS: Linux (x86_64-linux-gnu)
  CPU: 4 × Intel(R) Xeon(R) Platinum 8370C CPU @ 2.80GHz
  WORD_SIZE: 64
  LLVM: libLLVM-16.0.6 (ORCJIT, icelake-server)
Threads: 1 default, 0 interactive, 1 GC (on 4 virtual cores)

Benchmark Results

┌───────────────────────┬───────┬─────────────┬────────┬───────────────────────────────┬────────────────────────────┬─────────────────────────────────┐
│                       │       │             │        │       t(eval) / t(ref)        │     t(grad) / t(eval)      │        t(grad) / t(ref)         │
│                       │       │             │        │ ─────────┬──────────┬──────── │ ───────┬─────────┬──────── │ ──────────┬───────────┬──────── │
│                 Model │   Dim │  AD Backend │ Linked │     base │  this PR │ speedup │   base │ this PR │ speedup │      base │   this PR │ speedup │
├───────────────────────┼───────┼─────────────┼────────┼──────────┼──────────┼─────────┼────────┼─────────┼─────────┼───────────┼───────────┼─────────┤
│               Dynamic │    10 │    mooncake │   true │   282.98 │   280.26 │    1.01 │   7.72 │    8.13 │    0.95 │   2185.29 │   2277.45 │    0.96 │
│                   LDA │    12 │ reversediff │   true │  3485.55 │  3306.00 │    1.05 │   3.29 │    1.95 │    1.69 │  11483.28 │   6439.73 │    1.78 │
│   Loop univariate 10k │ 10000 │    mooncake │   true │ 30908.25 │ 30837.72 │    1.00 │   6.36 │    6.31 │    1.01 │ 196476.77 │ 194681.42 │    1.01 │
├───────────────────────┼───────┼─────────────┼────────┼──────────┼──────────┼─────────┼────────┼─────────┼─────────┼───────────┼───────────┼─────────┤
│    Loop univariate 1k │  1000 │    mooncake │   true │  3144.22 │  3144.12 │    1.00 │   6.99 │    6.20 │    1.13 │  21964.97 │  19501.99 │    1.13 │
│      Multivariate 10k │ 10000 │    mooncake │   true │ 31896.32 │ 31989.73 │    1.00 │   9.86 │    9.73 │    1.01 │ 314375.20 │ 311353.11 │    1.01 │
│       Multivariate 1k │  1000 │    mooncake │   true │  3428.76 │  3417.58 │    1.00 │   9.31 │    9.30 │    1.00 │  31927.86 │  31792.76 │    1.00 │
├───────────────────────┼───────┼─────────────┼────────┼──────────┼──────────┼─────────┼────────┼─────────┼─────────┼───────────┼───────────┼─────────┤
│ Simple assume observe │     1 │ forwarddiff │  false │     0.88 │     1.01 │    0.87 │  10.50 │    9.08 │    1.16 │      9.22 │      9.14 │    1.01 │
│           Smorgasbord │   201 │ forwarddiff │  false │   960.16 │   933.92 │    1.03 │  71.50 │   71.84 │    1.00 │  68650.05 │  67091.60 │    1.02 │
│           Smorgasbord │   201 │      enzyme │   true │  1297.65 │  1292.76 │    1.00 │   6.47 │    4.84 │    1.34 │   8395.18 │   6258.34 │    1.34 │
├───────────────────────┼───────┼─────────────┼────────┼──────────┼──────────┼─────────┼────────┼─────────┼─────────┼───────────┼───────────┼─────────┤
│           Smorgasbord │   201 │ forwarddiff │   true │  1296.15 │  1291.82 │    1.00 │  67.36 │   67.71 │    0.99 │  87306.86 │  87475.17 │    1.00 │
│           Smorgasbord │   201 │    mooncake │   true │  1301.63 │  1298.80 │    1.00 │   4.68 │    4.75 │    0.99 │   6090.86 │   6167.60 │    0.99 │
│           Smorgasbord │   201 │ reversediff │   true │  1296.65 │  1282.38 │    1.01 │ 127.53 │  130.76 │    0.98 │ 165365.71 │ 167686.37 │    0.99 │
├───────────────────────┼───────┼─────────────┼────────┼──────────┼──────────┼─────────┼────────┼─────────┼─────────┼───────────┼───────────┼─────────┤
│              Submodel │     1 │    mooncake │   true │     0.88 │     0.87 │    1.01 │  27.04 │   26.88 │    1.01 │     23.71 │     23.40 │    1.01 │
└───────────────────────┴───────┴─────────────┴────────┴──────────┴──────────┴─────────┴────────┴─────────┴─────────┴───────────┴───────────┴─────────┘

[codecov]

Codecov Report

❌ Patch coverage is 34.48276% with 19 lines in your changes missing coverage. Please review.
✅ Project coverage is 77.93%. Comparing base (34b8230) to head (613dc1d).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/utils.jl	25.00%	18 Missing ⚠️
src/transformed_values.jl	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1328      +/-   ##
==========================================
- Coverage   78.26%   77.93%   -0.34%     
==========================================
  Files          50       50              
  Lines        3566     3585      +19     
==========================================
+ Hits         2791     2794       +3     
- Misses        775      791      +16     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[github-actions]

DynamicPPL.jl documentation for PR #1328 is available at:
https://TuringLang.github.io/DynamicPPL.jl/previews/PR1328/