From b2e21a64f8469c2f11b9c00e750b9dd512d9df2e Mon Sep 17 00:00:00 2001
From: Philipp Arras <parras@mpa-garching.mpg.de>
Date: Wed, 24 Jul 2019 15:47:17 +0200
Subject: [PATCH] Add Reimar's criterion to NewtonCG
---
nifty5/__init__.py | 2 +-
nifty5/minimization/descent_minimizers.py | 20 +++---
nifty5/minimization/iteration_controllers.py | 67 ++++++++++++++++++++
nifty5/probing.py | 1 +
4 files changed, 80 insertions(+), 10 deletions(-)
diff --git a/nifty5/__init__.py b/nifty5/__init__.py
index 0dfba4cc0..a9326ba4a 100644
--- a/nifty5/__init__.py
+++ b/nifty5/__init__.py
@@ -59,7 +59,7 @@ from .probing import probe_with_posterior_samples, probe_diagonal, \
from .minimization.line_search import LineSearch
from .minimization.iteration_controllers import (
IterationController, GradientNormController, DeltaEnergyController,
- GradInfNormController)
+ GradInfNormController, AbsDeltaEnergyController)
from .minimization.minimizer import Minimizer
from .minimization.conjugate_gradient import ConjugateGradient
from .minimization.nonlinear_cg import NonlinearCG
diff --git a/nifty5/minimization/descent_minimizers.py b/nifty5/minimization/descent_minimizers.py
index b25c41963..d0520ea5b 100644
--- a/nifty5/minimization/descent_minimizers.py
+++ b/nifty5/minimization/descent_minimizers.py
@@ -19,7 +19,6 @@ import numpy as np
from ..logger import logger
from .conjugate_gradient import ConjugateGradient
-from .iteration_controllers import GradientNormController
from .line_search import LineSearch
from .minimizer import Minimizer
from .quadratic_energy import QuadraticEnergy
@@ -46,7 +45,7 @@ class DescentMinimizer(Minimizer):
self._controller = controller
self.line_searcher = line_searcher
- def __call__(self, energy, preconditioner=None):
+ def __call__(self, energy):
"""Performs the minimization of the provided Energy functional.
Parameters
@@ -82,7 +81,7 @@ class DescentMinimizer(Minimizer):
# compute a step length that reduces energy.value sufficiently
new_energy, success = self.line_searcher.perform_line_search(
- energy=energy, pk=self.get_descent_direction(energy),
+ energy=energy, pk=self.get_descent_direction(energy, f_k_minus_1),
f_k_minus_1=f_k_minus_1)
if not success:
self.reset()
@@ -163,12 +162,15 @@ class NewtonCG(DescentMinimizer):
super(NewtonCG, self).__init__(controller=controller,
line_searcher=line_searcher)
- def get_descent_direction(self, energy):
- g = energy.gradient
- maggrad = abs(g).sum()
- termcond = np.min([0.5, np.sqrt(maggrad)]) * maggrad
- ic = GradientNormController(tol_abs_gradnorm=termcond, p=1)
- e = QuadraticEnergy(0*energy.position, energy.metric, g)
+ def get_descent_direction(self, energy, f_k_minus_1):
+ from .iteration_controllers import AbsDeltaEnergyController, GradientNormController
+ if f_k_minus_1 is None:
+ ic = GradientNormController(iteration_limit=1)
+ else:
+ alpha = 0.1
+ ediff = alpha*(f_k_minus_1 - energy.value)
+ ic = AbsDeltaEnergyController(ediff, iteration_limit=200, name=' Internal', convergence_level=1)
+ e = QuadraticEnergy(0*energy.position, energy.metric, energy.gradient)
e, conv = ConjugateGradient(ic, nreset=np.inf)(e)
if conv == ic.ERROR:
raise RuntimeError
diff --git a/nifty5/minimization/iteration_controllers.py b/nifty5/minimization/iteration_controllers.py
index 88142a869..efaaa14e8 100644
--- a/nifty5/minimization/iteration_controllers.py
+++ b/nifty5/minimization/iteration_controllers.py
@@ -276,3 +276,70 @@ class DeltaEnergyController(IterationController):
return self.CONVERGED
return self.CONTINUE
+
+
+class AbsDeltaEnergyController(IterationController):
+ """An iteration controller checking (mainly) the energy change from one
+ iteration to the next.
+
+ Parameters
+ ----------
+ tol_rel_deltaE : float
+ If the difference between the last and current energies divided by
+ the current energy is below this value, the convergence counter will
+ be increased in this iteration.
+ convergence_level : int, default=1
+ The number which the convergence counter must reach before the
+ iteration is considered to be converged
+ iteration_limit : int, optional
+ The maximum number of iterations that will be carried out.
+ name : str, optional
+ if supplied, this string and some diagnostic information will be
+ printed after every iteration
+ """
+
+ def __init__(self, deltaE, convergence_level=1,
+ iteration_limit=None, name=None):
+ self._deltaE = deltaE
+ self._convergence_level = convergence_level
+ self._iteration_limit = iteration_limit
+ self._name = name
+
+ def start(self, energy):
+ self._itcount = -1
+ self._ccount = 0
+ self._Eold = 0.
+ return self.check(energy)
+
+ def check(self, energy):
+ self._itcount += 1
+
+ inclvl = False
+ Eval = energy.value
+ diff = abs(self._Eold-Eval)
+ if self._itcount > 0:
+ if diff < self._deltaE:
+ inclvl = True
+ self._Eold = Eval
+ if inclvl:
+ self._ccount += 1
+ else:
+ self._ccount = max(0, self._ccount-1)
+
+ # report
+ if self._name is not None:
+ logger.info(
+ "{}: Iteration #{} energy={:.6E} diff={:.6E} crit={:.6E}"
+ .format(self._name, self._itcount, Eval, diff, self._deltaE))
+
+ # Are we done?
+ if self._iteration_limit is not None:
+ if self._itcount >= self._iteration_limit:
+ logger.warning(
+ "{} Iteration limit reached. Assuming convergence"
+ .format("" if self._name is None else self._name+": "))
+ return self.CONVERGED
+ if self._ccount >= self._convergence_level:
+ return self.CONVERGED
+
+ return self.CONTINUE
diff --git a/nifty5/probing.py b/nifty5/probing.py
index 1eca3105e..2c1ad8a3f 100644
--- a/nifty5/probing.py
+++ b/nifty5/probing.py
@@ -138,6 +138,7 @@ def probe_diagonal(op, nprobes, random_type="pm1"):
def approximation2endo(op, nsamples):
+ print('Calculate preconditioner')
sc = StatCalculator()
for _ in range(nsamples):
sc.add(op.draw_sample())
--
GitLab