Test Suite

QUBODrivers includes a reusable test suite for sampler authors. It checks the MOI interface, model conversion, result queries, fixed variables, and example QUBO/Ising problems.

QUBODrivers.test is the public entry point. Its methods are provided by the QUBODrivers_Test_Ext package extension, which Julia loads automatically when both QUBODrivers and Test are present in the active environment. This keeps Test out of QUBODrivers' hard dependencies while still exposing a stable testing API.

The extension module names themselves, such as QUBODrivers_Test_Ext and MOI_PythonCall_Ext, are internal implementation details. Users should call the public APIs they enable, like QUBODrivers.test, instead of importing those extension modules directly.

Basic Usage

Add Test to the active test environment, then call QUBODrivers.test inside a regular @testset.

using Test
using QUBODrivers

@testset "My sampler" begin
    QUBODrivers.test(MySampler.Optimizer)
end

If your sampler needs deterministic configuration, pass a setup function. The function receives each optimizer instance before it is tested.

QUBODrivers.test(MySampler.Optimizer) do model
    MOI.set(model, MOI.RawOptimizerAttribute("seed"), 1)
    MOI.set(model, MOI.RawOptimizerAttribute("num_reads"), 100)
end

For slow backends, disable the example problem set and run only the interface checks:

QUBODrivers.test(MySampler.Optimizer; examples = false)

What the Suite Expects

A sampler should:

• be instantiable as an MOI.AbstractOptimizer;
• support binary and spin variable domains through MOI;
• accept linear, quadratic, and single-variable objectives;
• handle fixed variables consistently;
• return at least one result for solvable test models;
• expose result count, objective values, variable primals, solve time, and status through standard MOI attributes.

test(optimizer::Type{S}; examples::Bool=true) where {S<:AbstractSampler}
test(config!::Function, optimizer::Type{S}; examples::Bool=true) where {S<:AbstractSampler}