# Copyright 2021-2022 Lawrence Livermore National Security, LLC and other
# MuyGPyS Project Developers. See the top-level COPYRIGHT file for details.
#
# SPDX-License-Identifier: MIT
"""
Convenience functions for optimizing :class:`MuyGPyS.gp.muygps.MuyGPS` objects
The functions
:func:`~MuyGPyS.optimize.chassis.optimize_from_indices` and
:func:`~MuyGPyS.optimize.chassis.optimize_from_tensors` wrap different
optimization packages to provide a simple interface to optimize the
hyperparameters of :class:`~MuyGPyS.gp.muygps.MuyGPS` objects.
Currently, `opt_method="scipy"` wraps :class:`scipy.optimize.opt`
multiparameter optimization using L-BFGS-B algorithm using the objective
function :func:`MuyGPyS.optimize.objective.loo_crossval`.
Currently, `opt_method="bayesian"` (also accepts `"bayes"` and `"bayes_opt"`)
wraps :class:`bayes_opt.BayesianOptimization`. Unlike the `scipy` version,
`BayesianOptimization` can be meaningfully modified by several kwargs.
`MuyGPyS` assigns reasonable defaults if no settings are passed by the user.
See the `BayesianOptimization <https://github.com/fmfn/BayesianOptimization>`_
documentation for details.
"""
import numpy as np
import warnings
from MuyGPyS import config
from MuyGPyS._src.gp.distance import _make_train_tensors
from MuyGPyS._src.optimize.chassis import (
_scipy_optimize,
_bayes_opt_optimize,
)
from MuyGPyS.gp.muygps import MuyGPS
from MuyGPyS.optimize.utils import _switch_on_opt_method
from MuyGPyS.optimize.objective import make_obj_fn
from MuyGPyS.optimize.loss import get_loss_func
[docs]def optimize_from_indices(
muygps: MuyGPS,
batch_indices: np.ndarray,
batch_nn_indices: np.ndarray,
train_features: np.ndarray,
train_targets: np.ndarray,
loss_method: str = "mse",
obj_method: str = "loo_crossval",
opt_method: str = "bayes",
verbose: bool = False,
**kwargs,
) -> MuyGPS:
"""
Find an optimal model directly from the data.
Use this method if you do not need to retain the distance matrices used for
optimization.
See the following example, where we have already created a `batch_indices`
vector and a `batch_nn_indices` matrix using
:class:`MuyGPyS.neighbors.NN_Wrapper`, and initialized a
:class:`MuyGPyS.gp.muygps.MuyGPS` model `muygps`.
Example:
>>> from MuyGPyS.optimize.chassis import optimize_from_indices
>>> muygps = optimize_from_indices(
... muygps,
... batch_indices,
... batch_nn_indices,
... train_features,
... train_features,
... train_responses,
... loss_method='mse',
... obj_method='loo_crossval',
... opt_method='scipy',
... verbose=True,
... )
parameters to be optimized: ['nu']
bounds: [[0.1 1. ]]
sampled x0: [0.8858425]
optimizer results:
fun: 0.4797763813693626
hess_inv: <1x1 LbfgsInvHessProduct with dtype=float64>
jac: array([-3.06976666e-06])
message: b'CONVERGENCE: NORM_OF_PROJECTED_GRADIENT_<=_PGTOL'
nfev: 16
nit: 5
njev: 8
status: 0
success: True
x: array([0.39963594])
Args:
muygps:
The model to be optimized.
batch_indices:
A vector of integers of shape `(batch_count,)` identifying the
training batch of observations to be approximated.
batch_nn_indices:
A matrix of integers of shape `(batch_count, nn_count)` listing the
nearest neighbor indices for all observations in the batch.
train_features:
The full floating point training data matrix of shape
`(train_count, feature_count)`.
train_targets:
A matrix of shape `(train_count, feature_count)` whose rows are
vector-valued responses for each training element.
loss_method:
Indicates the loss function to be used.
obj_method:
Indicates the objective function to be minimized. Currently
restricted to `"loo_crossval"`.
opt_method:
Indicates the optimization method to be used. Currently restricted
to `"bayesian"` (alternately `"bayes"` or `"bayes_opt"`) and
`"scipy"`.
verbose:
If True, print debug messages.
kwargs:
Additional keyword arguments to be passed to the wrapper optimizer.
Returns:
A new MuyGPs model whose specified hyperparameters have been optimized.
"""
(
crosswise_dists,
pairwise_dists,
batch_targets,
batch_nn_targets,
) = _make_train_tensors(
muygps.kernel.metric,
batch_indices,
batch_nn_indices,
train_features,
train_targets,
)
return optimize_from_tensors(
muygps,
batch_targets,
batch_nn_targets,
crosswise_dists,
pairwise_dists,
loss_method=loss_method,
obj_method=obj_method,
opt_method=opt_method,
verbose=verbose,
**kwargs,
)
[docs]def optimize_from_tensors(
muygps: MuyGPS,
batch_targets: np.ndarray,
batch_nn_targets: np.ndarray,
crosswise_dists: np.ndarray,
pairwise_dists: np.ndarray,
loss_method: str = "mse",
obj_method: str = "loo_crossval",
opt_method: str = "bayes",
verbose: bool = False,
**kwargs,
) -> MuyGPS:
"""
Find the optimal model using existing distance matrices.
Use this method if you need to retain the distance matrices used for later
use.
See the following example, where we have already created a `batch_indices`
vector and a `batch_nn_indices` matrix using
:class:`MuyGPyS.neighbors.NN_Wrapper`, a `crosswise_dists`
matrix using :func:`MuyGPyS.gp.distance.crosswise_distances` and
`pairwise_dists` using :func:`MuyGPyS.gp.distance.pairwise_distances`, and
initialized a :class:`MuyGPyS.gp.muygps.MuyGPS` model `muygps`.
Example:
>>> from MuyGPyS.optimize.chassis import optimize_from_tensors
>>> muygps = optimize_from_tensors(
... muygps,
... batch_indices,
... batch_nn_indices,
... crosswise_dists,
... pairwise_dists,
... train_responses,
... loss_method='mse',
... obj_method='loo_crossval',
... opt_method='scipy',
... verbose=True,
... )
parameters to be optimized: ['nu']
bounds: [[0.1 1. ]]
sampled x0: [0.8858425]
optimizer results:
fun: 0.4797763813693626
hess_inv: <1x1 LbfgsInvHessProduct with dtype=float64>
jac: array([-3.06976666e-06])
message: b'CONVERGENCE: NORM_OF_PROJECTED_GRADIENT_<=_PGTOL'
nfev: 16
nit: 5
njev: 8
status: 0
success: True
x: array([0.39963594])
Args:
muygps:
The model to be optimized.
batch_targets:
Matrix of floats of shape `(batch_count, response_count)` whose rows
give the expected response for each batch element.
batch_nn_targets:
Tensor of floats of shape `(batch_count, nn_count, response_count)`
containing the expected response for each nearest neighbor of each
batch element.
crosswise_dists:
Distance matrix of floats of shape `(batch_count, nn_count)` whose
rows give the distances between each batch element and its nearest
neighbors.
pairwise_dists:
Distance tensor of floats of shape
`(batch_count, nn_count, nn_count)` whose second two dimensions give
the pairwise distances between the nearest neighbors of each batch
element.
loss_method:
Indicates the loss function to be used.
obj_method:
Indicates the objective function to be minimized. Currently
restricted to `"loo_crossval"`.
opt_method:
Indicates the optimization method to be used. Currently restricted
to `"bayesian"` (alternately `"bayes"` or `"bayes_opt"`) and
`"scipy"`.
verbose:
If True, print debug messages.
kwargs:
Additional keyword arguments to be passed to the wrapper optimizer.
Returns:
A new MuyGPs model whose specified hyperparameters have been optimized.
"""
loss_fn = get_loss_func(loss_method)
kernel_fn = muygps.kernel.get_opt_fn(opt_method)
predict_fn = muygps.get_opt_fn(opt_method)
obj_fn = make_obj_fn(
obj_method,
opt_method,
loss_fn,
kernel_fn,
predict_fn,
pairwise_dists,
crosswise_dists,
batch_nn_targets,
batch_targets,
)
return _switch_on_opt_method(
opt_method,
_bayes_opt_optimize,
_scipy_optimize,
muygps,
obj_fn,
verbose=verbose,
**kwargs,
)