test/julia/test_pyjulia.py:62: AssertionError

[dweindl]

maybe tolerances in test_pyjulia_pipeline need to be adapted.

https://github.com/ICB-DCM/pyPESTO/actions/runs/7472405078/job/20334519310

____________________________ test_pyjulia_pipeline _____________________________

    def test_pyjulia_pipeline():
        """Test that a pipeline with julia objective works."""
        # just make sure display function works
        rng = np.random.default_rng(42)

        assert display_source_ipython(  # noqa: S101
            f"{os.path.dirname(__file__)}/../../doc/example/model_julia/LR.jl"
        )

        # define objective
        obj = JuliaObjective(
            module="LR",
            source_file=f"{os.path.dirname(__file__)}/../../"
            f"doc/example/model_julia/LR.jl",
            fun="fun",
            grad="grad",
        )

        n_p = obj.get("n_p")

        # call consistency
        x = rng.normal(size=n_p)
        assert obj.get_fval(x) == obj.get_fval(x)  # noqa: S101
        assert (obj.get_grad(x) == obj.get_grad(x)).all()  # noqa: S101

        # define problem
        lb, ub = [-5.0] * n_p, [5.0] * n_p
        # create 10 random starting points within the bounds
        x_guesses = rng.uniform(lb, ub, size=(10, n_p))
        problem = Problem(obj, lb=lb, ub=ub, x_guesses=x_guesses)

        # optimize
        result = optimize.minimize(problem, engine=SingleCoreEngine(), n_starts=10)

        # use parallelization
        result2 = optimize.minimize(
            problem, engine=MultiProcessEngine(), n_starts=10
        )

        # check results match
        assert np.allclose(  # noqa: S101
            result.optimize_result[0].x, result2.optimize_result[0].x
        )
        # optimal point won't be true parameter
        x_true = obj.get("p_true")
        assert not np.allclose(x_true, result.optimize_result[0].x)  # noqa: S101

        # check with analytical value
        p_opt = obj.get("p_opt")
>       assert np.allclose(result.optimize_result[0].x, p_opt)  # noqa: S101
E       assert False
E        +  where False = <function allclose at 0x7f94f5366d30>(array([ 0.67469106,  0.85365126, -0.27880515, -0.09968592,  0.60157911,\n        0.2020777 ,  0.00129849, -0.11284351]), array([ 0.67469117,  0.85365118, -0.27880519, -0.09968595,  0.60157918,\n        0.20207782,  0.00129851, -0.11284341]))
E        +    where <function allclose at 0x7f94f5366d30> = np.allclose
E        +    and   array([ 0.67469106,  0.85365126, -0.27880515, -0.09968592,  0.60157911,\n        0.2020777 ,  0.00129849, -0.11284351]) = {'id': '7', 'x': array([ 0.67469106,  0.85365126, -0.27880515, -0.09968592,  0.60157911,\n        0.2020777 ,  0.001298...B options={'disp': False, 'maxfun': 1000}>", 'free_indices': array([0, 1, 2, 3, 4, 5, 6, 7]), 'inner_parameters': None}.x

test/julia/test_pyjulia.py:62: AssertionError