Source code for netket.callbacks.convergence_stopping

# Copyright 2020, 2021 The NetKet Authors - All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

from collections import deque

import numpy as np

from netket.utils import struct




class ConvergenceStopping(struct.Pytree, mutable=True):
    """A simple callback to stop the optimisation if the loss gets below a certain threshold.
    based on `driver._loss_name`."""

    target: float
    """Target value for the monitored quantity. Training will stop if the driver hits the baseline."""
    monitor: str
    """Loss statistic to monitor. Should be one of 'mean', 'variance', 'sigma'."""

    smoothing_window: int
    """
    The loss is smoothed over the last `smoothing_window` iterations to reduce statistical fluctuations
    """
    patience: int
    """
    The loss must be consistently below this value for this number of iterations in order to stop the optimisation.
    """

    # caches
    _loss_window: deque
    _patience_counter: int

    def __init__(
        self,
        target: float,
        monitor: str = "mean",
        *,
        smoothing_window: int = 10,
        patience: int = 10,
    ):
        self.target = target
        self.monitor = monitor
        self.smoothing_window = smoothing_window
        self.patience = patience

        self._loss_window: deque = deque([], maxlen=self.smoothing_window)
        self._patience_counter: int = 0



    def __call__(self, step, log_data, driver):
        """
        A boolean function that determines whether or not to stop training.

        Args:
            step: An integer corresponding to the step (iteration or epoch) in training.
            log_data: A dictionary containing log data for training.
            driver: A NetKet variational driver.

        Returns:
            A boolean. If True, training continues, else, it does not.
        """
        loss = np.real(getattr(log_data[driver._loss_name], self.monitor))

        self._loss_window.append(loss)
        loss_smooth = np.mean(self._loss_window)

        if loss_smooth <= self.target:
            self._patience_counter += 1
        else:
            self._patience_counter = 0

        if self._patience_counter > self.patience:
            return False

        return True