gemseo-multi-fidelity 0.0.1__py3-none-any.whl

gemseo_multi_fidelity/models/rbf/kernels/gaussian.py

@@ -0,0 +1,93 @@
+# Copyright 2021 IRT Saint Exupéry, https://www.irt-saintexupery.com
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU Lesser General Public
+# License version 3 as published by the Free Software Foundation.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+# Lesser General Public License for more details.
+#
+# You should have received a copy of the GNU Lesser General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
+
+# Copyright (c) 2018 AIRBUS OPERATIONS
+
+#
+# Contributors:
+#    INITIAL AUTHORS - API and implementation and/or documentation
+#      :author: Romain Olivanti
+#    OTHER AUTHORS   - MACROSCOPIC CHANGES
+"""Gaussian kernel."""
+
+from __future__ import annotations
+
+from numpy import exp
+
+from gemseo_multi_fidelity.models.rbf.kernels.rbf_kernel import RBFKernel
+
+
+class GaussianKernel(RBFKernel):
+    """Gaussian kernel."""
+
+    def __init__(self, scaling: float = 1.0) -> None:
+        """Initialize the kernel.
+
+        Args:
+            scaling: The kernel scaling.
+        """
+        super().__init__(scaling=scaling)
+
+    def set_scaling(self, scaling: float) -> None:
+        """Set the scaling of the kernel.
+
+        Args:
+            scaling: The scaling to set.
+        """
+        # Pre-squared
+        super().set_scaling(scaling**2)
+
+    def _compute_val(self, radius: float) -> float:
+        """Compute the kernel value for radius.
+
+        Args:
+            radius: The 1D or 2D ndarray. ??? TODO
+
+        Returns:
+            The kernel value.
+        """
+        return exp(-self._scaling * radius**2)
+
+    def _compute_grad_fact(self, radius: float) -> float:
+        """Compute the kernel gradient factor.
+
+        Args:
+            radius: The 1D or 2D ndarray. ??? TODO
+
+        Returns:
+            The gradient factor.
+        """
+        scale = self._scaling
+        return -2.0 * scale * exp(-scale * radius**2)
+
+    def _compute_hess_fact(self, radius) -> tuple[str]:
+        """Compute the kernel hessians factors.
+
+        Args:
+            radius: The 1D or 2D ndarray. ??? TODO
+
+        Returns:
+            The multiplicative factor, the diagonal factor.
+        """
+        scale = self._scaling
+        return 4.0 * scale**2 * exp(-scale * radius**2), -0.5 / scale
+
+    def __str__(self) -> str:
+        """Get the string representation.
+
+        Returns:
+            The string representation.
+        """
+        return f"K(r) = exp(-{self._scaling:.6g} * r**2)"

gemseo_multi_fidelity/models/rbf/kernels/matern_3_2.py

@@ -0,0 +1,101 @@
+# Copyright 2021 IRT Saint Exupéry, https://www.irt-saintexupery.com
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU Lesser General Public
+# License version 3 as published by the Free Software Foundation.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+# Lesser General Public License for more details.
+#
+# You should have received a copy of the GNU Lesser General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
+
+# Copyright (c) 2018 AIRBUS OPERATIONS
+
+#
+# Contributors:
+#    INITIAL AUTHORS - API and implementation and/or documentation
+#      :author: Romain Olivanti
+#    OTHER AUTHORS   - MACROSCOPIC CHANGES
+"""Matern 3/2 kernel."""
+
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+
+from numpy import exp
+from numpy import sqrt
+
+from gemseo_multi_fidelity.models.rbf.kernels.rbf_kernel import RBFKernel
+
+if TYPE_CHECKING:
+    from numpy.typing import NDArray
+
+
+class Matern32Kernel(RBFKernel):
+    """Matern 3/2 kernel."""
+
+    def __init__(self, scaling: float = 1.0) -> None:
+        """Initialize the kernel.
+
+        Args:
+            scaling: The kernel scaling.
+        """
+        super().__init__(scaling=scaling)
+
+    def set_scaling(self, scaling: float) -> None:
+        """Set the scaling of the kernel.
+
+        Args:
+            scaling: The scaling to set.
+        """
+        # Pre-multiplied by the Matern 3/2 factor
+        super().set_scaling(sqrt(3.0) * scaling)
+
+    def _compute_val(self, radius: NDArray) -> float:
+        """Compute the kernel value for radius.
+
+        Args:
+            radius: The radius (1D or 2D array).
+
+        Returns:
+            The kernel value.
+        """
+        arg = self._scaling * radius
+        return (1.0 + arg) * exp(-arg)
+
+    def _compute_grad_fact(self, radius: float) -> float:
+        """Compute the kernel gradient factor.
+
+        Args:
+            radius: The radius (1D or 2D array).
+
+        Returns:
+            The kernel gradient factor.
+        """
+        scale = self._scaling
+        return -(scale**2) * exp(-scale * radius)
+
+    def _compute_hess_fact(self, radius: float) -> tuple[float, float]:
+        """Compute the kernel hessians factors.
+
+        Args:
+            radius: The radius (1D or 2D array).
+
+        Returns:
+            The tuple (multiplicative factor, diagonal factor).
+        """
+        scale = self._scaling
+        return scale**3 * exp(-scale * radius) / radius, -radius / scale
+
+    def __str__(self) -> str:
+        """Get the string representation.
+
+        Returns:
+            The string representation.
+        """
+        arg = f"{self._scaling:.6g} * r"
+        return f"K(r) = (1. + {arg}) * exp(-{arg})"

gemseo_multi_fidelity/models/rbf/kernels/matern_5_2.py

@@ -0,0 +1,101 @@
+# Copyright 2021 IRT Saint Exupéry, https://www.irt-saintexupery.com
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU Lesser General Public
+# License version 3 as published by the Free Software Foundation.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+# Lesser General Public License for more details.
+#
+# You should have received a copy of the GNU Lesser General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
+
+# Copyright (c) 2018 AIRBUS OPERATIONS
+
+#
+# Contributors:
+#    INITIAL AUTHORS - API and implementation and/or documentation
+#      :author: Romain Olivanti
+#    OTHER AUTHORS   - MACROSCOPIC CHANGES
+"""Matern 5/2 kernel."""
+
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+
+from numpy import exp
+from numpy import sqrt
+
+from gemseo_multi_fidelity.models.rbf.kernels.rbf_kernel import RBFKernel
+
+if TYPE_CHECKING:
+    from numpy.typing import NDArray
+
+
+class Matern52Kernel(RBFKernel):
+    """Matern 5/2 kernel."""
+
+    def __init__(self, scaling: float = 1.0) -> None:
+        """Initialize the kernel.
+
+        Args:
+            scaling: The kernel scaling.
+        """
+        super().__init__(scaling=scaling)
+
+    def set_scaling(self, scaling: float) -> None:
+        """Set the scaling of the kernel.
+
+        Args:
+            scaling: The scaling to set.
+        """
+        # Pre-multiplied by the Matern 5/2 factor
+        super().set_scaling(sqrt(5.0) * scaling)
+
+    def _compute_val(self, radius: NDArray) -> float:
+        """Compute the kernel value for radius.
+
+        Args:
+            radius: The radius (1D or 2D array).
+
+        Returns:
+            The kernel value.
+        """
+        arg = self._scaling * radius
+        return (1.0 + arg + arg**2 / 3.0) * exp(-arg)
+
+    def _compute_grad_fact(self, radius: float) -> float:
+        """Compute the kernel gradient factor.
+
+        Args:
+            radius: The radius (1D or 2D array).
+
+        Returns:
+            The kernel gradient factor.
+        """
+        scale = self._scaling
+        return -(scale**2) / 3.0 * (1.0 + scale * radius) * exp(-scale * radius)
+
+    def _compute_hess_fact(self, radius: float) -> tuple[float, float]:
+        """Compute the kernel hessians factors.
+
+        Args:
+            radius: The radius (1D or 2D array).
+
+        Returns:
+            The tuple (multiplicative factor, diagonal factor).
+        """
+        scale = self._scaling
+        return scale**4 / 3.0 * exp(-scale * radius), -(1.0 + scale * radius) / scale**2
+
+    def __str__(self) -> str:
+        """Get the string representation.
+
+        Returns:
+            The string representation.
+        """
+        arg = f"{self._scaling:.6g} * r"
+        return f"K(r) = (1. + {arg} + {arg}**2 / 3.)* exp(-{arg})"

gemseo_multi_fidelity/models/rbf/kernels/rbf_kernel.py

@@ -0,0 +1,172 @@
+# Copyright 2021 IRT Saint Exupéry, https://www.irt-saintexupery.com
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU Lesser General Public
+# License version 3 as published by the Free Software Foundation.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+# Lesser General Public License for more details.
+#
+# You should have received a copy of the GNU Lesser General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
+
+# Copyright (c) 2018 AIRBUS OPERATIONS
+
+#
+# Contributors:
+#    INITIAL AUTHORS - API and implementation and/or documentation
+#      :author: Romain Olivanti
+#    OTHER AUTHORS   - MACROSCOPIC CHANGES
+"""Abstract RBF Kernel."""
+
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+from typing import Any
+
+from numpy import einsum
+from numpy import fill_diagonal
+from numpy import ones
+from numpy.linalg import norm
+
+if TYPE_CHECKING:
+    from numpy.typing import NDArray
+
+
+class RBFKernel:
+    """Abstract RBF Kernel class."""
+
+    def __init__(self, scaling: float = 1.0) -> None:
+        """Initialize the kernel.
+
+        Args:
+            scaling: The kernel scaling.
+        """
+        self._scaling = 1.0
+        # Use setter to check the input
+        self.set_scaling(scaling)
+
+    @property
+    def scaling(self) -> float:
+        """Accessor to the scaling property.
+
+        Returns:
+            The scaling.
+        """
+        return self._scaling
+
+    def set_scaling(self, scaling: float) -> None:
+        """Set the scaling of the kernel.
+
+        Args:
+            scaling: The scaling to set.
+        """
+        if not isinstance(scaling, float) or scaling <= 0.0:
+            msg = "The scaling must be a float > 0"
+            raise TypeError(msg)
+        self._scaling = scaling
+
+    @staticmethod
+    def compute_radius(arg: NDArray) -> float | NDArray:
+        """Compute the input for the kernel.
+
+        Args:
+            arg: The 1D or 2D ndarray.
+
+        Returns:
+            The kernel input.
+        """
+        norm_axis = len(arg.shape) - 1
+        if norm_axis not in [0, 1]:
+            msg = "arg must be 1D or 2D ndarray"
+            raise ValueError(msg)
+        return norm(arg, axis=norm_axis)
+
+    def __call__(self, arg) -> Any:
+        """Evaluate the kernel for the specified arg.
+
+        Args:
+            arg: The 1D or 2D ndarray.
+
+        Returns:
+            The kernel value.
+        """
+        radius = self.compute_radius(arg)
+        return self._compute_val(radius)
+
+    def grad(self, arg: NDArray) -> NDArray:
+        """Evaluate the gradient of the kernel for the specified arg.
+
+        Args:
+            arg: The 1D or 2D ndarray.
+
+        Returns:
+             The gradient.
+        """
+        radius = self.compute_radius(arg)
+        grad_fact = self._compute_grad_fact(radius)
+        if len(arg.shape) == 1:
+            return grad_fact * arg
+        return einsum("i,ij->ij", grad_fact, arg)
+
+    def hess(self, arg: NDArray) -> NDArray:
+        """Evaluate the hessian of the kernel for the specified arg.
+
+        Args:
+            arg: The 1D or 2D ndarray.
+
+        Returns:
+            The hessian.
+        """
+        radius = self.compute_radius(arg)
+        fact, diag_fact = self._compute_hess_fact(radius)
+        if len(arg.shape) == 1:
+            # 1d input
+            ret = einsum("i,j->ij", arg, arg)
+            fill_diagonal(ret, ret.diagonal() + diag_fact)
+            return fact * ret
+        # 2d input
+        ret = einsum("ij,il->ijl", arg, arg)
+        n_points = ret.shape[0]
+        if isinstance(diag_fact, float):
+            diag_fact *= ones(n_points)
+
+        for i in range(n_points):
+            fill_diagonal(ret[i], ret[i].diagonal() + diag_fact[i])
+        return einsum("i, ijk->ijk", fact, ret)
+
+    def _compute_val(self, radius: NDArray):
+        """Compute the kernel value for radius.
+
+        Args:
+            radius: The 1D or 2D ndarray.
+
+        Returns:
+            The kernel value.
+        """
+        raise NotImplementedError
+
+    def _compute_grad_fact(self, radius: NDArray) -> Any:
+        """Compute the kernel gradient factor.
+
+        Args:
+            radius: The 1D or 2D ndarray.
+
+        Returns:
+            The gradient factor.
+        """
+        raise NotImplementedError
+
+    def _compute_hess_fact(self, radius: NDArray) -> tuple[Any, Any]:
+        """Compute the kernel hessians factors.
+
+        Args:
+            radius: The 1D or 2D ndarray.
+
+        Returns:
+            The tuple (multiplicative factor, diagonal factor).
+        """
+        raise NotImplementedError