desdeo 1.2__py3-none-any.whl → 2.1.0__py3-none-any.whl

desdeo/problem/testproblems/mixed_variable_dimenrions_problem.py

@@ -0,0 +1,83 @@
+from desdeo.problem.schema import (
+    Objective,
+    Problem,
+    TensorVariable,
+    Variable,
+    VariableTypeEnum,
+)
+
+def mixed_variable_dimensions_problem():
+    """Defines a problem with variables with mixed dimensions.
+
+    Has both Variable and TensorVariable types of variables, where the TensorVariables have
+    varying number of dimensions. Mostly for testing purposes.
+    """
+    x = Variable(
+        name="Regular variable",
+        symbol="x",
+        variable_type=VariableTypeEnum.real,
+        lowerbound=-1.0,
+        upperbound=1.0,
+        initial_value=0.5,
+    )
+
+    y = TensorVariable(
+        name="1D vector",
+        symbol="Y",
+        shape=[5],
+        variable_type=VariableTypeEnum.real,
+        lowerbounds=5,
+        upperbounds=5,
+        initial_values=5,
+    )
+
+    z = TensorVariable(
+        name="2D vector",
+        symbol="Z",
+        shape=[5, 2],
+        variable_type=VariableTypeEnum.real,
+        lowerbounds=-100,
+        upperbounds=100,
+        initial_values=1,
+    )
+
+    a = TensorVariable(
+        name="2D vector",
+        symbol="A",
+        shape=[2, 3, 2],
+        variable_type=VariableTypeEnum.real,
+        lowerbounds=-100,
+        upperbounds=100,
+        initial_values=1,
+    )
+
+    dummy = Objective(
+        name="dummy objective, not relevant",
+        symbol="f_1",
+        func="x - Y[1]",
+        maximize=False,
+        ideal=-1000,
+        nadir=1000,
+        is_linear=True,
+        is_convex=True,
+        is_twice_differentiable=True,
+    )
+
+    dummy2 = Objective(
+        name="dummy objective, not relevant",
+        symbol="f_2",
+        func="-x + Y[1]",
+        maximize=False,
+        ideal=-1000,
+        nadir=1000,
+        is_linear=True,
+        is_convex=True,
+        is_twice_differentiable=True,
+    )
+
+    return Problem(
+        name="Mixed variable dimensions problem",
+        description="A problem with variables with mixed dimensions. For testing.",
+        variables=[x, y, z, a],
+        objectives=[dummy, dummy2],
+    )

desdeo/problem/testproblems/momip_problem.py

@@ -0,0 +1,172 @@
+from desdeo.problem.schema import (
+    Constraint,
+    ConstraintTypeEnum,
+    Objective,
+    ObjectiveTypeEnum,
+    Problem,
+    Variable,
+    VariableTypeEnum,
+)
+
+
+def momip_ti2() -> Problem:
+    """Defines the mixed-integer multiobjective optimization problem test instance 2 (TI2).
+
+    The problem has four variables, two continuous and two integer. The Pareto optimal solutions
+    hold for solutions with x_1^2 + x_2^2 = 0.25 and (x_3, x_4) = {(0, -1), (-1, 0)}.
+
+    References:
+        Eichfelder, G., Gerlach, T., & Warnow, L. (n.d.). Test Instances for
+            Multiobjective Mixed-Integer Nonlinear Optimization.
+    """
+    x_1 = Variable(name="x_1", symbol="x_1", variable_type=VariableTypeEnum.real, lowerbound=-1.0, upperbound=1.0)
+    x_2 = Variable(name="x_2", symbol="x_2", variable_type=VariableTypeEnum.real, lowerbound=-1.0, upperbound=1.0)
+    x_3 = Variable(name="x_3", symbol="x_3", variable_type=VariableTypeEnum.integer, lowerbound=-1, upperbound=1)
+    x_4 = Variable(name="x_4", symbol="x_4", variable_type=VariableTypeEnum.integer, lowerbound=-1, upperbound=1)
+
+    f_1 = Objective(
+        name="f_1",
+        symbol="f_1",
+        func="x_1 + x_3",
+        objective_type=ObjectiveTypeEnum.analytical,
+        maximize=False,
+        is_linear=True,
+        is_convex=True,
+        is_twice_differentiable=True,
+    )
+    f_2 = Objective(
+        name="f_2",
+        symbol="f_2",
+        func="x_2 + x_4",
+        objective_type=ObjectiveTypeEnum.analytical,
+        maximize=False,
+        is_linear=True,
+        is_convex=True,
+        is_twice_differentiable=True,
+    )
+
+    con_1 = Constraint(
+        name="g_1",
+        symbol="g_1",
+        cons_type=ConstraintTypeEnum.LTE,
+        func="x_1**2 + x_2**2 - 0.25",
+        is_linear=False,
+        is_convex=False,
+        is_twice_differentiable=True,
+    )
+    con_2 = Constraint(
+        name="g_2",
+        symbol="g_2",
+        cons_type=ConstraintTypeEnum.LTE,
+        func="x_3**2 + x_4**2 - 1",
+        is_linear=False,
+        is_convex=False,
+        is_twice_differentiable=True,
+    )
+
+    return Problem(
+        name="MOMIP Test Instance 2",
+        description="Test instance 2",
+        variables=[x_1, x_2, x_3, x_4],
+        constraints=[con_1, con_2],
+        objectives=[f_1, f_2],
+    )
+
+
+def momip_ti7() -> Problem:
+    r"""Defines the mixed-integer multiobjective optimization problem test instance 7 (T7).
+
+    The problem is defined as follows:
+
+    \begin{align}
+        &\min_{\mathbf{x}} & f_1(\mathbf{x}) & = x_1 + x_4 \\
+        &\max_{\mathbf{x}} & f_2(\mathbf{x}) & = -(x_2 + x_5) \\
+        &\min_{\mathbf{x}} & f_3(\mathbf{x}) & = x_3 + x_6 \\
+        &\text{s.t.,}   & x_1^2 +x_2^2 + x_3^2 & \leq 1,\\
+        &               & x_4^2 + x_5^2 + x_6^2 & \leq 1,\\
+        &               & -1 \leq x_i \leq 1&\;\text{for}\;i=\{1,2,3\},\\
+        &               & x_i \in \{-1, 0, 1\}&\;\text{for}\;i=\{4, 5, 6\}.
+    \end{align}
+
+    In the problem, $x_1, x_2, x_3$ are real-valued and $x_4, x_5, x_6$ are integer-valued. The problem
+    is convex and differentiable.
+
+    The Pareto optimal integer assignments are $(x_4, x_5, x_6) \in {(0,0,-1), (0, -1, 0), (-1,0,0)}$,
+    and the real-valued assignments are $\{x_1, x_2, x_3 \in \mathbb{R}^3 |
+    x_1^2 + x_2^2 + x_3^2 = 1, x_1 \leq 0, x_2 \leq 0, x_3 \leq 0\}$. Unlike in the original definition,
+    $f_2$ is formulated to be maximized instead of minimized.
+
+    References:
+        Eichfelder, G., Gerlach, T., & Warnow, L. (n.d.). Test Instances for
+            Multiobjective Mixed-Integer Nonlinear Optimization.
+    """
+    x_1 = Variable(name="x_1", symbol="x_1", variable_type=VariableTypeEnum.real)
+    x_2 = Variable(name="x_2", symbol="x_2", variable_type=VariableTypeEnum.real)
+    x_3 = Variable(name="x_3", symbol="x_3", variable_type=VariableTypeEnum.real)
+    x_4 = Variable(name="x_4", symbol="x_4", variable_type=VariableTypeEnum.integer, lowerbound=-1, upperbound=1)
+    x_5 = Variable(name="x_5", symbol="x_5", variable_type=VariableTypeEnum.integer, lowerbound=-1, upperbound=1)
+    x_6 = Variable(name="x_6", symbol="x_6", variable_type=VariableTypeEnum.integer, lowerbound=-1, upperbound=1)
+
+    f_1 = Objective(
+        name="f_1",
+        symbol="f_1",
+        func="x_1 + x_4",
+        objective_type=ObjectiveTypeEnum.analytical,
+        ideal=-3,
+        nadir=3,
+        maximize=False,
+        is_linear=True,
+        is_convex=True,
+        is_twice_differentiable=True,
+    )
+    f_2 = Objective(
+        name="f_2",
+        symbol="f_2",
+        func="-(x_2 + x_5)",
+        objective_type=ObjectiveTypeEnum.analytical,
+        ideal=3,
+        nadir=-3,
+        maximize=True,
+        is_linear=True,
+        is_convex=True,
+        is_twice_differentiable=True,
+    )
+    f_3 = Objective(
+        name="f_3",
+        symbol="f_3",
+        func="x_3 + x_6",
+        objective_type=ObjectiveTypeEnum.analytical,
+        ideal=-3,
+        nadir=3,
+        maximize=False,
+        is_linear=True,
+        is_convex=True,
+        is_twice_differentiable=True,
+    )
+
+    con_1 = Constraint(
+        name="g_1",
+        symbol="g_1",
+        cons_type=ConstraintTypeEnum.LTE,
+        func="x_1**2 + x_2**2 + x_3**2 - 1",
+        is_linear=False,
+        is_convex=False,
+        is_twice_differentiable=True,
+    )
+    con_2 = Constraint(
+        name="g_2",
+        symbol="g_2",
+        cons_type=ConstraintTypeEnum.LTE,
+        func="x_4**2 + x_5**2 + x_6**2 - 1",
+        is_linear=False,
+        is_convex=False,
+        is_twice_differentiable=True,
+    )
+
+    return Problem(
+        name="MOMIP Test Instance 7",
+        description="Test instance 17",
+        variables=[x_1, x_2, x_3, x_4, x_5, x_6],
+        constraints=[con_1, con_2],
+        objectives=[f_1, f_2, f_3],
+    )

desdeo/problem/testproblems/multi_valued_constraints.py

@@ -0,0 +1,119 @@
+"""Defines a test problem with a constraint that is multi-valued."""
+
+from desdeo.problem import (
+    Constant,
+    Constraint,
+    ConstraintTypeEnum,
+    Objective,
+    ObjectiveTypeEnum,
+    Problem,
+    TensorConstant,
+    TensorVariable,
+    Variable,
+    VariableTypeEnum,
+)
+
+
+def multi_valued_constraint_problem() -> Problem:
+    r"""Defines a test problem with a multi-valued constraint.
+
+     The problem has two objectives, two variables, and two constraints, the other of which, is multi-valued.
+     The problem is defined as follows:
+    \[
+         \begin{aligned}
+         \text{Min} \quad
+         & f_1(x_1, x_2, y) = x_1^2 + x_2^2 + y^2, \\[4pt]
+         \text{Min} \quad
+         & f_2(x_1, x_2, y) = (x_1 - 2)^2 + (x_2 - 1)^2 + (y - 1)^2, \\[6pt]
+         \text{subject to} \quad
+         & g(x_1, x_2, y) = x_1^2 + x_2 + y - 2 \le 0, \\[4pt]
+         & G(x_1, x_2) = A
+         \begin{bmatrix}
+         x_1 \\[2pt]
+         x_2
+         \end{bmatrix}
+         \le 0,
+         \quad
+         A =
+         \begin{bmatrix}
+         1 & -1 \\[2pt]
+         -1 & -2
+         \end{bmatrix}.
+         \end{aligned}
+    \]
+
+
+    Returns:
+         Problem: the problem model.
+    """
+    xs = TensorVariable(
+        name="x",
+        symbol="X",
+        variable_type=VariableTypeEnum.real,
+        shape=[2, 1],
+        lowerbounds=-5.0,
+        upperbounds=5.0,
+        initial_values=0.1,
+    )
+
+    y = Variable(
+        name="y",
+        symbol="y",
+        variable_type=VariableTypeEnum.real,
+        lowerbound=-10.0,
+        upperbound=10.0,
+        initial_value=0.1,
+    )
+
+    a = TensorConstant(name="A", symbol="A", shape=[2, 2], values=[[1.0, -1.0], [-1.0, -2.0]])
+
+    one = Constant(name="one", symbol="one", value=1.0)
+
+    f_1_expr = "X[1, 1]**2 + X[2, 1]**2 + y**2"
+    f_2_expr = "(X[1, 1] - 2)**2 + (X[2, 1] - one)**2 + (y - one)**2"
+
+    g_1_expr = "X[1, 1]**2 + X[2, 1] + y - 2"
+    big_g_expr = "A @ X"
+
+    f_1 = Objective(
+        name="f1",
+        symbol="f_1",
+        func=f_1_expr,
+        objective_type=ObjectiveTypeEnum.analytical,
+        ideal=0.0,
+        nadir=150.0,
+        is_twice_differentiable=True,
+    )
+
+    f_2 = Objective(
+        name="f2",
+        symbol="f_2",
+        func=f_2_expr,
+        ideal=0.0,
+        nadir=206.0,
+        objective_type=ObjectiveTypeEnum.analytical,
+        is_twice_differentiable=True,
+    )
+
+    g_1 = Constraint(
+        name="g1", symbol="g_1", cons_type=ConstraintTypeEnum.LTE, func=g_1_expr, is_twice_differentiable=True
+    )
+
+    big_g = Constraint(
+        name="big_g",
+        symbol="G",
+        cons_type=ConstraintTypeEnum.LTE,
+        func=big_g_expr,
+        is_twice_differentiable=True,
+        is_linear=True,
+        is_convex=True,
+    )
+
+    return Problem(
+        name="Multi-valued-constraint problem",
+        description="Problem for testing problems with multi-valued constraints.",
+        constants=[a, one],
+        variables=[xs, y],
+        constraints=[g_1, big_g],
+        objectives=[f_1, f_2],
+    )

desdeo/problem/testproblems/nimbus_problem.py

@@ -0,0 +1,143 @@
+from desdeo.problem.schema import (
+    Objective,
+    ObjectiveTypeEnum,
+    Problem,
+    Variable,
+    VariableTypeEnum,
+)
+
+def nimbus_test_problem() -> Problem:
+    r"""Defines the test problem utilized in the article describing Synchronous NIMBUS.
+
+    Defines the following multiobjective optimization problem:
+
+    \begin{align}
+        &\max_{\mathbf{x}} & f_1(\mathbf{x}) &= x_1 x_2\\
+        &\min_{\mathbf{x}} & f_2(\mathbf{x}) &= (x_1 - 4)^2 + x_2^2\\
+        &\min_{\mathbf{x}} & f_3(\mathbf{x}) &= -x_1 - x_2\\
+        &\min_{\mathbf{x}} & f_4(\mathbf{x}) &= x_1 - x_2\\
+        &\min_{\mathbf{x}} & f_5(\mathbf{x}) &= 50 x_1^4 + 10 x_2^4 \\
+        &\min_{\mathbf{x}} & f_6(\mathbf{x}) &= 30 (x_1 - 5)^4 + 100 (x_2 - 3)^4\\
+        &\text{s.t.,}     && 1 \leq x_i \leq 3\quad i=\{1,2\},
+    \end{align}
+
+    with the following ideal point
+    $\mathbf{z}^\star = \left[9.0, 2.0, -6.0, -2.0, 60.0, 480.0 \right]$ and nadir point
+    $\mathbf{z}^\text{nad} = \left[ 1.0, 18.0, -2.0, 2.0, 4860.0, 9280.0 \right]$.
+
+    References:
+        Miettinen, K., & Mäkelä, M. M. (2006). Synchronous approach in
+            interactive multiobjective optimization. European Journal of Operational
+            Research, 170(3), 909–922. https://doi.org/10.1016/j.ejor.2004.07.052
+
+
+    Returns:
+        Problem: the NIMBUS test problem.
+    """
+    variables = [
+        Variable(
+            name="x_1",
+            symbol="x_1",
+            variable_type=VariableTypeEnum.real,
+            initial_value=1.0,
+            lowerbound=1.0,
+            upperbound=3.0,
+        ),
+        Variable(
+            name="x_2",
+            symbol="x_2",
+            variable_type=VariableTypeEnum.real,
+            initial_value=1.0,
+            lowerbound=1.0,
+            upperbound=3.0,
+        ),
+    ]
+
+    f_1_expr = "x_1 * x_2"
+    f_2_expr = "(x_1 - 4)**2 + x_2**2"
+    f_3_expr = "-x_1 - x_2"
+    f_4_expr = "x_1 - x_2"
+    f_5_expr = "50 * x_1**4 + 10 * x_2**4"
+    f_6_expr = "30 * (x_1 - 5)**4 + 100*(x_2 - 3)**4"
+
+    objectives = [
+        Objective(
+            name="Objective 1",
+            symbol="f_1",
+            func=f_1_expr,
+            maximize=True,
+            objective_type=ObjectiveTypeEnum.analytical,
+            ideal=9.0,
+            nadir=1.0,
+            is_convex=False,
+            is_linear=False,
+            is_twice_differentiable=True,
+        ),
+        Objective(
+            name="Objective 2",
+            symbol="f_2",
+            func=f_2_expr,
+            maximize=False,
+            objective_type=ObjectiveTypeEnum.analytical,
+            ideal=2.0,
+            nadir=18.0,
+            is_convex=False,
+            is_linear=False,
+            is_twice_differentiable=True,
+        ),
+        Objective(
+            name="Objective 3",
+            symbol="f_3",
+            func=f_3_expr,
+            maximize=False,
+            objective_type=ObjectiveTypeEnum.analytical,
+            ideal=-6.0,
+            nadir=-2.0,
+            is_convex=True,
+            is_linear=True,
+            is_twice_differentiable=True,
+        ),
+        Objective(
+            name="Objective 4",
+            symbol="f_4",
+            func=f_4_expr,
+            maximize=False,
+            objective_type=ObjectiveTypeEnum.analytical,
+            ideal=-2.0,
+            nadir=2.0,
+            is_convex=True,
+            is_linear=True,
+            is_twice_differentiable=True,
+        ),
+        Objective(
+            name="Objective 5",
+            symbol="f_5",
+            func=f_5_expr,
+            maximize=False,
+            objective_type=ObjectiveTypeEnum.analytical,
+            ideal=60.0,
+            nadir=4860.0,
+            is_convex=False,
+            is_linear=False,
+            is_twice_differentiable=True,
+        ),
+        Objective(
+            name="Objective 6",
+            symbol="f_6",
+            func=f_6_expr,
+            maximize=False,
+            objective_type=ObjectiveTypeEnum.analytical,
+            ideal=480.0,
+            nadir=9280.0,
+            is_convex=False,
+            is_linear=False,
+            is_twice_differentiable=True,
+        ),
+    ]
+
+    return Problem(
+        name="NIMBUS test problem",
+        description="The test problem used in the Synchronous NIMBUS article",
+        variables=variables,
+        objectives=objectives,
+    )

desdeo/problem/testproblems/pareto_navigator_problem.py

@@ -0,0 +1,89 @@
+from desdeo.problem.schema import (
+    Constraint,
+    ConstraintTypeEnum,
+    DiscreteRepresentation,
+    Objective,
+    ObjectiveTypeEnum,
+    Problem,
+    Variable,
+    VariableTypeEnum,
+)
+
+def pareto_navigator_test_problem() -> Problem:
+    r"""Defines the problem utilized in the (convex) Pareto navigator article.
+
+    The problem is defined as follows:
+
+    \begin{align}
+        &\min_{\mathbf{x}} & f_1(\mathbf{x}) & = -x_1 - x_2 + 5 \\
+        &\min_{\mathbf{x}} & f_2(\mathbf{x}) & = 0.2(x_1^2 -10x_1 + x_2^2 - 4x_2 + 11) \\
+        &\min_{\mathbf{x}} & f_3(\mathbf{x}) & = (5 - x_1)(x_2 - 11)\\
+        &\text{s.t.,}   & 3x_1 + x_2 - 12 & \leq 0,\\
+        &               & 2x_1 + x_2 - 9 & \leq 0,\\
+        &               & x_1 + 2x_2 - 12 & \leq 0,\\
+        &               & 0 \leq x_1 & \leq 4,\\
+        &               & 0 \leq x_2 & \leq 6.
+    \end{align}
+
+    The problem comes with seven pre-defined Pareto optimal solutions that were
+    utilized in the original article as well. From these, the ideal and nadir
+    points of the problem are approximated also.
+
+    References:
+        Eskelinen, P., Miettinen, K., Klamroth, K., & Hakanen, J. (2010). Pareto
+            navigator for interactive nonlinear multiobjective optimization. OR
+            Spectrum, 32(1), 211-227.
+    """
+    x_1 = Variable(name="x_1", symbol="x_1", variable_type=VariableTypeEnum.real, lowerbound=0, upperbound=4)
+    x_2 = Variable(name="x_2", symbol="x_2", variable_type=VariableTypeEnum.real, lowerbound=0, upperbound=6)
+
+    f_1 = Objective(
+        name="f_1",
+        symbol="f_1",
+        func="-x_1 - x_2 + 5",
+        objective_type=ObjectiveTypeEnum.analytical,
+        ideal=-2.0,
+        nadir=5.0,
+        maximize=False,
+    )
+    f_2 = Objective(
+        name="f_2",
+        symbol="f_2",
+        func="0.2*(x_1**2 - 10*x_1 + x_2**2 - 4*x_2 + 11)",
+        objective_type=ObjectiveTypeEnum.analytical,
+        ideal=-3.1,
+        nadir=4.60,
+        maximize=False,
+    )
+    f_3 = Objective(
+        name="f_3",
+        symbol="f_3",
+        func="(5 - x_1) * (x_2 - 11)",
+        objective_type=ObjectiveTypeEnum.analytical,
+        ideal=-55.0,
+        nadir=-14.25,
+        maximize=False,
+    )
+
+    con_1 = Constraint(name="g_1", symbol="g_1", func="3*x_1 + x_2 - 12", cons_type=ConstraintTypeEnum.LTE)
+    con_2 = Constraint(name="g_2", symbol="g_2", func="2*x_1 + x_2 - 9", cons_type=ConstraintTypeEnum.LTE)
+    con_3 = Constraint(name="g_3", symbol="g_3", func="x_1 + 2*x_2 - 12", cons_type=ConstraintTypeEnum.LTE)
+
+    representation = DiscreteRepresentation(
+        variable_values={"x_1": [0, 0, 0, 0, 0, 0, 0], "x_2": [0, 0, 0, 0, 0, 0, 0]},
+        objective_values={
+            "f_1": [-2.0, -1.0, 0.0, 1.38, 1.73, 2.48, 5.0],
+            "f_2": [0.0, 4.6, -3.1, 0.62, 1.72, 1.45, 2.2],
+            "f_3": [-18.0, -25.0, -14.25, -35.33, -38.64, -42.41, -55.0],
+        },
+        non_dominated=True,
+    )
+
+    return Problem(
+        name="The (convex) Pareto navigator problem.",
+        description="The test problem used in the (convex) Pareto navigator paper.",
+        variables=[x_1, x_2],
+        objectives=[f_1, f_2, f_3],
+        constraints=[con_1, con_2, con_3],
+        discrete_representation=representation,
+    )