Table of Contents

Getting Started

Install

dotnet add package OptunaSharp

OptunaSharp multi-targets .NET 8 and .NET 10 and depends on NumSharp.

Your first study

using OptunaSharp;
using OptunaSharp.Samplers;

var study = Optuna.CreateStudy(sampler: new TPESampler(seed: 42));

study.Optimize(trial =>
{
    double x = trial.SuggestFloat("x", -10, 10);   // define-by-run: the search space is declared here
    return (x - 2) * (x - 2);                      // the value to minimize
}, nTrials: 20);

Console.WriteLine(study.BestValue);        // 0.0004973004749973951
Console.WriteLine(study.BestParams["x"]);  // 2.0223002348641756

Python Optuna 5.0.0 prints the same two numbers for optuna.create_study(sampler=optuna.samplers.TPESampler(seed=42)) optimized for 20 trials: seeded sampling reproduces Optuna's draws exactly. Names follow .NET casing (suggest_float → SuggestFloat, n_trials → nTrials); parameter order and defaults are Optuna's. Next: Study and Trials covers ask-and-tell, multi-objective studies and attributes, and Samplers and Pruners cover the algorithms.

Build from source

git clone https://github.com/SciSharp/OptunaSharp
cd OptunaSharp
dotnet build SciSharp.OptunaSharp.sln -c Release

The NumSharp reference

OptunaSharp.Core depends on NumSharp and its native LAPACK/CBLAS backend NumSharp.Interop.OpenBLAS, both consumed as NuGet packages from nuget.org (version pinned by NumSharpVersion in Directory.Build.props). The OpenBLAS package ships the native LAPACK binaries, so np.linalg is bit-identical to NumPy with no extra setup — build and CI just restore from nuget.org.

The matching NumSharp source is vendored reference-only at refs/NumSharp (a pinned submodule on the same release tag) for browsing; it is not part of the build.

Test

# Unit tests (both target frameworks), as per-push CI runs them; drop `&TestCategory!=Slow` for the full run
dotnet test test/OptunaSharp.Tests/OptunaSharp.Tests.csproj -c Release \
  --filter "TestCategory!=OpenBugs&TestCategory!=HighMemory&TestCategory!=RequiresPython&TestCategory!=Slow"

# Live-Optuna interop parity: needs a discoverable Python with the vendored Optuna, the version parity is
# measured against (`git submodule update --init refs/optuna`, then `pip install -e refs/optuna`)
dotnet test test/OptunaSharp.Tests.Interop/OptunaSharp.Tests.Interop.csproj -c Release

The interop suite self-skips (Inconclusive) when no Python/Optuna is found, so it never turns a machine without Python red.

Pack

dotnet pack src/OptunaSharp.Core/OptunaSharp.Core.csproj -c Release -o packages

Build the documentation site locally

dotnet tool install -g docfx
docfx docs/website-src/docfx.json --serve

DocFX generates the API reference from OptunaSharp.Core and renders the conceptual docs into docs/website/.

Edit this page