diff --git a/examples/greedy/squareBubbleHomog.py b/examples/greedy/squareBubbleHomog.py
index 7150daa..be15aa4 100644
--- a/examples/greedy/squareBubbleHomog.py
+++ b/examples/greedy/squareBubbleHomog.py
@@ -1,104 +1,108 @@
import numpy as np
from rrompy.hfengines.scipy import HelmholtzSquareBubbleProblemEngine as HSBPE
from rrompy.reduction_methods.lagrange_greedy import \
ApproximantLagrangePadeGreedy as Pade
from rrompy.reduction_methods.lagrange_greedy import \
ApproximantLagrangeRBGreedy as RB
from rrompy.utilities.base import squareResonances
verb = 2
-timed = False
+timed = True
algo = "Pade"
#algo = "RB"
polyBasis = "LEGENDRE"
#polyBasis = "CHEBYSHEV"
-polyBasis = "MONOMIAL"
+#polyBasis = "MONOMIAL"
+errorEstimatorKind = "SIMPLIFIED"
+errorEstimatorKind = "EXACT"
k0s = np.power(np.linspace(95, 149, 200), .5)
-k0s = np.power(np.linspace(95, 129, 200), .5)
+k0s = np.power(np.linspace(95, 129, 20), .5)
k0 = np.mean(np.power(k0s, 2.)) ** .5
kl, kr = min(k0s), max(k0s)
polesexact = np.unique(np.power(squareResonances(kl**2., kr**2., False), .5))
-rescaling = lambda x: np.power(x, 2.)
-rescalingInv = lambda x: np.power(x, .5)
params = {'muBounds':[kl, kr], 'nTrainingPoints':500, 'Delta':0,
- 'greedyTol':1e-2, 'nTestPoints':2, 'basis':polyBasis}
+ 'greedyTol':1e-2, 'nTestPoints':2, 'basis':polyBasis,
+ 'errorEstimatorKind':errorEstimatorKind}
+
+if timed:
+ verb = 0
solver = HSBPE(kappa = 12 ** .5, theta = np.pi / 3, n = 30,
verbosity = verb)
solver.omega = np.real(k0)
if algo == "Pade":
approx = Pade(solver, mu0 = k0, approxParameters = params,
verbosity = verb)
else:
approx = RB(solver, mu0 = k0, approxParameters = params, verbosity = verb)
if timed:
from time import clock
start_time = clock()
approx.greedy()
print("Time: ", clock() - start_time)
else:
approx.greedy(True)
approx.samplingEngine.verbosity = 0
approx.verbosity = 0
from matplotlib import pyplot as plt
normApp = np.zeros_like(k0s)
norm = np.zeros_like(k0s)
res = np.zeros_like(k0s)
err = np.zeros_like(k0s)
for j in range(len(k0s)):
normApp[j] = approx.normApp(k0s[j])
norm[j] = approx.normHF(k0s[j])
res[j] = (approx.estNormer.norm(approx.getRes(k0s[j]))
/ approx.estNormer.norm(approx.getRHS(k0s[j])))
err[j] = approx.normErr(k0s[j]) / approx.normHF(k0s[j])
resApp = approx.errorEstimator(k0s)
plt.figure()
plt.semilogy(k0s, norm)
plt.semilogy(k0s, normApp, '--')
plt.semilogy(polesexact,
2.*np.max(norm)*np.ones_like(polesexact, dtype = float), 'm.')
plt.semilogy(np.real(approx.mus),
4.*np.max(norm)*np.ones_like(approx.mus, dtype = float), 'rx')
plt.xlim([kl, kr])
plt.grid()
plt.show()
plt.close()
plt.figure()
plt.semilogy(k0s, res)
plt.semilogy(k0s, resApp, '--')
plt.semilogy(polesexact,
2.*np.max(resApp)*np.ones_like(polesexact, dtype = float), 'm.')
plt.semilogy(np.real(approx.mus),
4.*np.max(resApp)*np.ones_like(approx.mus, dtype = float), 'rx')
plt.xlim([kl, kr])
plt.grid()
plt.show()
plt.close()
plt.figure()
plt.semilogy(k0s, err)
plt.semilogy(polesexact,
2.*np.max(err)*np.ones_like(polesexact, dtype = float), 'm.')
plt.xlim([kl, kr])
plt.grid()
plt.show()
plt.close()
polesApp = approx.getPoles()
mask = (np.real(polesApp) < kl) | (np.real(polesApp) > kr)
print("Outliers:", polesApp[mask])
polesApp = polesApp[~mask]
plt.figure()
plt.plot(np.real(polesApp), np.imag(polesApp), 'kx')
plt.plot(np.real(polesexact), np.imag(polesexact), 'm.')
plt.axis('equal')
plt.grid()
plt.show()
plt.close()
diff --git a/examples/greedy/squareScatteringAirfoil.py b/examples/greedy/squareScatteringAirfoil.py
index 17fb055..58b9024 100644
--- a/examples/greedy/squareScatteringAirfoil.py
+++ b/examples/greedy/squareScatteringAirfoil.py
@@ -1,145 +1,145 @@
import numpy as np
import fenics as fen
import ufl
from rrompy.hfengines.scipy import HelmholtzBoxScatteringProblemEngine as HSP
from rrompy.reduction_methods.lagrange_greedy import \
ApproximantLagrangePadeGreedy as Pade
from rrompy.reduction_methods.lagrange_greedy import \
ApproximantLagrangeRBGreedy as RB
from rrompy.utilities.base.fenics import fenONE
verb = 2
timed = False
algo = "Pade"
#algo = "RB"
polyBasis = "LEGENDRE"
#polyBasis = "CHEBYSHEV"
-polyBasis = "MONOMIAL"
+#polyBasis = "MONOMIAL"
homog = True
homog = False
-k0s = np.linspace(5, 20, 50)
+k0s = np.linspace(5, 20, 25)
k0 = np.mean(k0s)
kl, kr = min(k0s), max(k0s)
params = {'muBounds':[kl, kr], 'nTrainingPoints':500, 'Delta':0,
'greedyTol':1e-2, 'nTestPoints':2, 'basis':polyBasis}
#########
PI = np.pi
R = 2
def Dboundary(x, on_boundary):
return on_boundary and (x[0]**2+x[1]**2)**.5 < .95 * R
kappa = 10
theta = PI * - 45 / 180.
mu = 1.1
epsilon = .1
mesh = fen.Mesh('../data/mesh/airfoil.xml')
c, s = np.cos(theta), np.sin(theta)
x, y = fen.SpatialCoordinate(mesh)[:]
u0R = - fen.cos(kappa * (c * x + s * y))
u0I = - fen.sin(kappa * (c * x + s * y))
checkReal = x**2-x+y**2
rhop5 = ((x**2+y**2)/((x-1)**2+y**2))**.25
phiroot1 = fen.atan(-y/(x**2-x+y**2)) / 2
phiroot2 = fen.atan(-y/(x**2-x+y**2)) / 2 - PI * ufl.sign(-y/(x**2-x+y**2)) / 2
kappam1 = (((rhop5*fen.cos(phiroot1)+.5)**2.+(rhop5*fen.sin(phiroot1))**2.)/
((rhop5*fen.cos(phiroot1)-1.)**2.+(rhop5*fen.sin(phiroot1))**2.)
)**.5 - mu
kappam2 = (((rhop5*fen.cos(phiroot2)+.5)**2.+(rhop5*fen.sin(phiroot2))**2.)/
((rhop5*fen.cos(phiroot2)-1.)**2.+(rhop5*fen.sin(phiroot2))**2.)
)**.5 - mu
Heps1 = .9 * .5 * (1 + kappam1/epsilon + fen.sin(PI*kappam1/epsilon) / PI) + .1
Heps2 = .9 * .5 * (1 + kappam2/epsilon + fen.sin(PI*kappam2/epsilon) / PI) + .1
cTT = ufl.conditional(ufl.le(kappam1, epsilon), Heps1, fenONE)
c_F = fen.Constant(.1)
cFT = ufl.conditional(ufl.le(kappam2, epsilon), Heps2, fenONE)
c_F = fen.Constant(.1)
cT = ufl.conditional(ufl.ge(kappam1, - epsilon), cTT, c_F)
cF = ufl.conditional(ufl.ge(kappam2, - epsilon), cFT, c_F)
a = ufl.conditional(ufl.ge(checkReal, 0.), cT, cF)
solver = HSP(R, np.real(k0), theta, n = 1, verbosity = verb,
degree_threshold = 8)
solver.omega = np.real(k0)
solver.V = fen.FunctionSpace(mesh, "P", 3)
solver.diffusivity = a
solver.DirichletBoundary = Dboundary
solver.RobinBoundary = "REST"
solver.DirichletDatum = [u0R, u0I]
#########
if algo == "Pade":
approx = Pade(solver, mu0 = k0, approxParameters = params,
verbosity = verb, homogeneized = homog)
else:
approx = RB(solver, mu0 = k0, approxParameters = params, verbosity = verb,
homogeneized = homog)
if timed:
from time import clock
start_time = clock()
approx.greedy()
print("Time: ", clock() - start_time)
else:
approx.greedy(True)
approx.samplingEngine.verbosity = 0
approx.verbosity = 0
kl, kr = np.real(kl), np.real(kr)
from matplotlib import pyplot as plt
normApp = np.zeros(len(k0s))
norm = np.zeros_like(normApp)
res = np.zeros_like(normApp)
err = np.zeros_like(normApp)
for j in range(len(k0s)):
normApp[j] = approx.normApp(k0s[j])
norm[j] = approx.normHF(k0s[j])
- res[j] = (approx.estNormer.norm(approx.getRes(k0s[j]))
- / approx.estNormer.norm(approx.getRHS(k0s[j])))
+ res[j] = (approx.estNormer.norm(approx.getRes(k0s[j], homogeneized=homog))
+ / approx.estNormer.norm(approx.getRHS(k0s[j], homogeneized=homog)))
err[j] = approx.normErr(k0s[j]) / approx.normHF(k0s[j])
resApp = approx.errorEstimator(k0s)
plt.figure()
plt.plot(k0s, norm)
plt.plot(k0s, normApp, '--')
plt.plot(np.real(approx.mus),
1.05*np.max(norm)*np.ones_like(approx.mus, dtype = float), 'rx')
plt.xlim([kl, kr])
plt.grid()
plt.show()
plt.close()
plt.figure()
plt.semilogy(k0s, res)
plt.semilogy(k0s, resApp, '--')
plt.semilogy(np.real(approx.mus),
4.*np.max(resApp)*np.ones_like(approx.mus, dtype = float), 'rx')
plt.xlim([kl, kr])
plt.grid()
plt.show()
plt.close()
plt.figure()
plt.semilogy(k0s, err)
plt.xlim([kl, kr])
plt.grid()
plt.show()
plt.close()
polesApp = approx.getPoles()
mask = (np.real(polesApp) < kl) | (np.real(polesApp) > kr)
print("Outliers:", polesApp[mask])
polesApp = polesApp[~mask]
plt.figure()
plt.plot(np.real(polesApp), np.imag(polesApp), 'kx')
plt.axis('equal')
plt.grid()
plt.show()
plt.close()
diff --git a/examples/greedy/squareScatteringHomog.py b/examples/greedy/squareScatteringHomog.py
index 8c9270a..d037701 100644
--- a/examples/greedy/squareScatteringHomog.py
+++ b/examples/greedy/squareScatteringHomog.py
@@ -1,91 +1,97 @@
import numpy as np
from rrompy.hfengines.scipy import HelmholtzCavityScatteringProblemEngine as \
HCSPE
from rrompy.reduction_methods.lagrange_greedy import \
ApproximantLagrangePadeGreedy as Pade
from rrompy.reduction_methods.lagrange_greedy import \
ApproximantLagrangeRBGreedy as RB
verb = 2
timed = False
algo = "Pade"
#algo = "RB"
polyBasis = "LEGENDRE"
#polyBasis = "CHEBYSHEV"
-polyBasis = "MONOMIAL"
+#polyBasis = "MONOMIAL"
+errorEstimatorKind = "SIMPLIFIED"
+#errorEstimatorKind = "EXACT"
k0s = np.linspace(5, 12, 100)
k0 = np.mean(k0s)
kl, kr = min(k0s), max(k0s)
params = {'muBounds':[kl, kr], 'nTrainingPoints':500, 'Delta':0,
- 'greedyTol':1e-2, 'nTestPoints':2, 'basis':polyBasis}
+ 'greedyTol':1e-2, 'nTestPoints':2, 'basis':polyBasis,
+ 'errorEstimatorKind':errorEstimatorKind}
+
+if timed:
+ verb = 0
solver = HCSPE(kappa = 5, n = 10, verbosity = verb)
solver.omega = np.real(k0)
if algo == "Pade":
approx = Pade(solver, mu0 = k0, approxParameters = params,
verbosity = verb)
else:
approx = RB(solver, mu0 = k0, approxParameters = params, verbosity = verb)
if timed:
from time import clock
start_time = clock()
approx.greedy()
print("Time: ", clock() - start_time)
else:
approx.greedy(True)
approx.samplingEngine.verbosity = 0
approx.verbosity = 0
from matplotlib import pyplot as plt
normApp = np.zeros(len(k0s))
norm = np.zeros_like(normApp)
res = np.zeros_like(normApp)
err = np.zeros_like(normApp)
for j in range(len(k0s)):
normApp[j] = approx.normApp(k0s[j])
norm[j] = approx.normHF(k0s[j])
res[j] = (approx.estNormer.norm(approx.getRes(k0s[j]))
/ approx.estNormer.norm(approx.getRHS(k0s[j])))
err[j] = approx.normErr(k0s[j]) / approx.normHF(k0s[j])
resApp = approx.errorEstimator(k0s)
plt.figure()
plt.plot(k0s, norm)
plt.plot(k0s, normApp, '--')
plt.plot(np.real(approx.mus),
1.25*np.max(norm)*np.ones_like(approx.mus, dtype = float), 'rx')
plt.xlim([kl, kr])
plt.grid()
plt.show()
plt.close()
plt.figure()
plt.semilogy(k0s, res)
plt.semilogy(k0s, resApp, '--')
plt.semilogy(np.real(approx.mus),
4.*np.max(resApp)*np.ones_like(approx.mus, dtype = float), 'rx')
plt.xlim([kl, kr])
plt.grid()
plt.show()
plt.close()
plt.figure()
plt.semilogy(k0s, err)
plt.xlim([kl, kr])
plt.grid()
plt.show()
plt.close()
polesApp = approx.getPoles()
mask = (np.real(polesApp) < kl) | (np.real(polesApp) > kr)
print("Outliers:", polesApp[mask])
polesApp = polesApp[~mask]
plt.figure()
plt.plot(np.real(polesApp), np.imag(polesApp), 'kx')
plt.axis('equal')
plt.grid()
plt.show()
plt.close()
diff --git a/examples/greedy/squareTransmissionNonHomog.py b/examples/greedy/squareTransmissionNonHomog.py
index e3d153a..5803a52 100644
--- a/examples/greedy/squareTransmissionNonHomog.py
+++ b/examples/greedy/squareTransmissionNonHomog.py
@@ -1,105 +1,108 @@
import numpy as np
from rrompy.hfengines.scipy import HelmholtzSquareTransmissionProblemEngine \
as HSTPE
from rrompy.reduction_methods.lagrange_greedy import \
ApproximantLagrangeRBGreedy as RB
from rrompy.reduction_methods.lagrange_greedy import \
ApproximantLagrangePadeGreedy as Pade
timed = False
verb = 2
algo = "Pade"
#algo = "RB"
polyBasis = "LEGENDRE"
-polyBasis = "CHEBYSHEV"
-polyBasis = "MONOMIAL"
+#polyBasis = "CHEBYSHEV"
+#polyBasis = "MONOMIAL"
homog = True
-homog = False
+#homog = False
+errorEstimatorKind = "SIMPLIFIED"
+errorEstimatorKind = "EXACT"
k0s = np.power(np.linspace(4, 15, 100), .5)
k0 = np.mean(np.power(k0s, 2.)) ** .5
kl, kr = min(k0s), max(k0s)
rescaling = lambda x: np.power(x, 2.)
rescalingInv = lambda x: np.power(x, .5)
params = {'muBounds':[kl, kr], 'nTrainingPoints':500, 'Delta':0,
- 'greedyTol':1e-2, 'nTestPoints':5, 'basis':polyBasis}
+ 'greedyTol':1e-2, 'nTestPoints':5, 'basis':polyBasis,
+ 'errorEstimatorKind':errorEstimatorKind}
solver = HSTPE(nT = 1, nB = 2, theta = np.pi * 20 / 180, kappa = 4.,
n = 20, verbosity = verb)
solver.omega = np.real(k0)
if algo == "Pade":
approx = Pade(solver, mu0 = k0, approxParameters = params,
verbosity = verb, homogeneized = homog)
else:
approx = RB(solver, mu0 = k0, approxParameters = params, verbosity = verb,
homogeneized = homog)
if timed:
from time import clock
start_time = clock()
approx.greedy()
print("Time: ", clock() - start_time)
else:
approx.greedy(True)
approx.samplingEngine.verbosity = 0
approx.verbosity = 0
from matplotlib import pyplot as plt
normApp = np.zeros(len(k0s))
norm = np.zeros_like(normApp)
res = np.zeros_like(normApp)
err = np.zeros_like(normApp)
for j in range(len(k0s)):
normApp[j] = approx.normApp(k0s[j])
norm[j] = approx.normHF(k0s[j])
- res[j] = (approx.estNormer.norm(approx.getRes(k0s[j]))
- / approx.estNormer.norm(approx.getRHS(k0s[j])))
+ res[j] = (approx.estNormer.norm(approx.getRes(k0s[j], homogeneized=homog))
+ / approx.estNormer.norm(approx.getRHS(k0s[j], homogeneized=homog)))
err[j] = approx.normErr(k0s[j]) / approx.normHF(k0s[j])
resApp = approx.errorEstimator(k0s)
polesApp = approx.getPoles()
polesApp = polesApp[np.abs(np.imag(polesApp)) < 1e-3]
plt.figure()
plt.semilogy(k0s, norm)
plt.semilogy(k0s, normApp, '--')
plt.semilogy(np.real(approx.mus),
4.*np.max(norm)*np.ones_like(approx.mus, dtype = float), 'rx')
plt.semilogy(np.real(polesApp),
2.*np.max(norm)*np.ones_like(polesApp, dtype = float), 'k.')
plt.xlim([kl, kr])
plt.grid()
plt.show()
plt.close()
plt.figure()
plt.semilogy(k0s, res)
plt.semilogy(k0s, resApp, '--')
plt.semilogy(np.real(approx.mus),
4.*np.max(resApp)*np.ones_like(approx.mus, dtype = float), 'rx')
plt.semilogy(np.real(polesApp),
2.*np.max(resApp)*np.ones_like(polesApp, dtype = float), 'k.')
plt.xlim([kl, kr])
plt.grid()
plt.show()
plt.close()
plt.figure()
plt.semilogy(k0s, err)
plt.semilogy(np.real(polesApp),
2.*np.max(err)*np.ones_like(polesApp, dtype = float), 'k.')
plt.xlim([kl, kr])
plt.grid()
plt.show()
plt.close()
polesApp = approx.getPoles()
mask = (np.real(polesApp) < kl) | (np.real(polesApp) > kr)
print("Outliers:", polesApp[mask])
polesApp = polesApp[~mask]
plt.figure()
plt.plot(np.real(polesApp), np.imag(polesApp), 'kx')
plt.axis('equal')
plt.grid()
plt.show()
plt.close()
diff --git a/rrompy/reduction_methods/lagrange/approximant_lagrange_pade.py b/rrompy/reduction_methods/lagrange/approximant_lagrange_pade.py
index 11f1846..096a0f6 100644
--- a/rrompy/reduction_methods/lagrange/approximant_lagrange_pade.py
+++ b/rrompy/reduction_methods/lagrange/approximant_lagrange_pade.py
@@ -1,479 +1,479 @@
# Copyright (C) 2018 by the RROMPy authors
#
# This file is part of RROMPy.
#
# RROMPy is free software: you can redistribute it and/or modify
# it under the terms of the GNU Lesser General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# RROMPy 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 RROMPy. If not, see .
#
from copy import copy
import numpy as np
from rrompy.reduction_methods.base import (checkRobustTolerance,
setupFitCallables)
from .generic_approximant_lagrange import GenericApproximantLagrange
from rrompy.utilities.base.types import Np1D, List, HFEng, DictAny
from rrompy.utilities.base import verbosityDepth, purgeDict, customFit
from rrompy.utilities.warning_manager import warn
__all__ = ['ApproximantLagrangePade']
class ApproximantLagrangePade(GenericApproximantLagrange):
"""
ROM Lagrange Pade' interpolant computation for parametric problems.
Args:
HFEngine: HF problem solver.
mu0(optional): Default parameter. Defaults to 0.
approxParameters(optional): Dictionary containing values for main
parameters of approximant. Recognized keys are:
- 'POD': whether to compute POD of snapshots; defaults to True;
- 'S': total number of samples current approximant relies upon;
defaults to 2;
- 'sampler': sample point generator; defaults to uniform sampler on
[0, 1];
- 'basis': type of basis for interpolation; allowed values include
'MONOMIAL', 'CHEBYSHEV' and 'LEGENDRE'; defaults to 'MONOMIAL';
- 'E': coefficient of interpolant to be minimized; defaults to
min(S, M + 1);
- 'M': degree of Pade' interpolant numerator; defaults to 0;
- 'N': degree of Pade' interpolant denominator; defaults to 0;
- 'interpRcond': tolerance for interpolation via numpy.polyfit;
defaults to None;
- 'robustTol': tolerance for robust Pade' denominator management;
defaults to 0.
Defaults to empty dict.
homogeneized: Whether to homogeneize Dirichlet BCs. Defaults to False.
verbosity(optional): Verbosity level. Defaults to 10.
Attributes:
HFEngine: HF problem solver.
mu0: Default parameter.
mus: Array of snapshot parameters.
ws: Array of snapshot weigths.
homogeneized: Whether to homogeneize Dirichlet BCs.
approxParameters: Dictionary containing values for main parameters of
approximant. Recognized keys are in parameterList.
parameterList: Recognized keys of approximant parameters:
- 'POD': whether to compute POD of snapshots;
- 'S': total number of samples current approximant relies upon;
- 'sampler': sample point generator;
- 'basis': type of basis for interpolation;
- 'E': coefficient of interpolant to be minimized;
- 'M': degree of Pade' interpolant numerator;
- 'N': degree of Pade' interpolant denominator;
- 'interpRcond': tolerance for interpolation via numpy.polyfit;
- 'robustTol': tolerance for robust Pade' denominator management.
extraApproxParameters: List of approxParameters keys in addition to
mother class's.
S: Number of solution snapshots over which current approximant is
based upon.
sampler: Sample point generator.
M: Numerator degree of approximant.
N: Denominator degree of approximant.
POD: Whether to compute POD of snapshots.
interpRcond: Tolerance for interpolation via numpy.polyfit.
robustTol: Tolerance for robust Pade' denominator management.
samplingEngine: Sampling engine.
uHF: High fidelity solution with wavenumber lastSolvedHF as numpy
complex vector.
lastSolvedHF: Wavenumber corresponding to last computed high fidelity
solution.
Q: Numpy 1D vector containing complex coefficients of approximant
denominator.
P: Numpy 2D vector whose columns are FE dofs of coefficients of
approximant numerator.
uApp: Last evaluated approximant as numpy complex vector.
lastApproxParameters: List of parameters corresponding to last
computed approximant.
"""
def __init__(self, HFEngine:HFEng, mu0 : complex = 0.,
approxParameters : DictAny = {}, homogeneized : bool = False,
verbosity : int = 10):
self._preInit()
self._addParametersToList(["basis", "E", "M", "N",
"interpRcond", "robustTol"])
super().__init__(HFEngine = HFEngine, mu0 = mu0,
approxParameters = approxParameters,
homogeneized = homogeneized,
verbosity = verbosity)
self._postInit()
@property
def approxParameters(self):
"""
Value of approximant parameters.
"""
return self._approxParameters
@approxParameters.setter
def approxParameters(self, approxParams):
approxParameters = purgeDict(approxParams, self.parameterList,
dictname = self.name() + ".approxParameters",
baselevel = 1)
approxParametersCopy = purgeDict(approxParameters, ["basis",
"E", "M", "N",
"interpRcond",
"robustTol"],
True, True, baselevel = 1)
if hasattr(self, "M"):
Mold = self.M
self._M = 0
if hasattr(self, "N"):
Nold = self.N
self._N = 0
if hasattr(self, "E"):
self._E = 0
GenericApproximantLagrange.approxParameters.fset(self,
approxParametersCopy)
keyList = list(approxParameters.keys())
if "basis" in keyList or not hasattr(self, "_val"):
if "basis" in keyList:
kind = approxParameters["basis"]
else:
kind = "MONOMIAL"
setupFit = setupFitCallables(kind)
for x in setupFit:
super().__setattr__("_" + x, setupFit[x])
if "interpRcond" in keyList:
self.interpRcond = approxParameters["interpRcond"]
elif hasattr(self, "interpRcond"):
self.interpRcond = self.interpRcond
else:
self.interpRcond = None
if "robustTol" in keyList:
self.robustTol = approxParameters["robustTol"]
elif hasattr(self, "robustTol"):
self.robustTol = self.robustTol
else:
self.robustTol = 0
if "M" in keyList:
self.M = approxParameters["M"]
elif hasattr(self, "M"):
self.M = Mold
else:
self.M = 0
if "N" in keyList:
self.N = approxParameters["N"]
elif hasattr(self, "N"):
self.N = Nold
else:
self.N = 0
if "E" in keyList:
self.E = approxParameters["E"]
else:
self.E = min(self.S - 1, self.M + 1)
@property
def M(self):
"""Value of M. Its assignment may change S."""
return self._M
@M.setter
def M(self, M):
if M < 0: raise ArithmeticError("M must be non-negative.")
self._M = M
self._approxParameters["M"] = self.M
if hasattr(self, "S") and self.S < self.M + 1:
warn("Prescribed S is too small. Updating S to M + 1.")
self.S = self.M + 1
@property
def N(self):
"""Value of N. Its assignment may change S."""
return self._N
@N.setter
def N(self, N):
if N < 0: raise ArithmeticError("N must be non-negative.")
self._N = N
self._approxParameters["N"] = self.N
if hasattr(self, "S") and self.S < self.N + 1:
warn("Prescribed S is too small. Updating S to N + 1.")
self.S = self.N + 1
@property
def E(self):
"""Value of E. Its assignment may change S."""
return self._E
@E.setter
def E(self, E):
if E < 0: raise ArithmeticError("E must be non-negative.")
self._E = E
self._approxParameters["E"] = self.E
if hasattr(self, "S") and self.S < self.E + 1:
warn("Prescribed S is too small. Updating S to E + 1.")
self.S = self.E + 1
@property
def robustTol(self):
"""Value of tolerance for robust Pade' denominator management."""
return self._robustTol
@robustTol.setter
def robustTol(self, robustTol):
if robustTol < 0.:
warn("Overriding prescribed negative robustness tolerance to 0.")
robustTol = 0.
self._robustTol = robustTol
self._approxParameters["robustTol"] = self.robustTol
@property
def S(self):
"""Value of S."""
return self._S
@S.setter
def S(self, S):
if S <= 0: raise ArithmeticError("S must be positive.")
if hasattr(self, "S"): Sold = self.S
else: Sold = -1
vals, label = [0] * 3, {0:"M", 1:"N", 2:"E"}
if hasattr(self, "M"): vals[0] = self.M
if hasattr(self, "N"): vals[1] = self.N
if hasattr(self, "E"): vals[2] = self.E
idxmax = np.argmax(vals)
if vals[idxmax] + 1 > S:
warn("Prescribed S is too small. Updating S to {} + 1."\
.format(label[idxmax]))
self.S = vals[idxmax] + 1
else:
self._S = S
self._approxParameters["S"] = self.S
if Sold != self.S:
self.resetSamples()
def setupApprox(self):
"""
Compute Pade' interpolant.
SVD-based robust eigenvalue management.
"""
if not self.checkComputedApprox():
if self.verbosity >= 5:
verbosityDepth("INIT", "Setting up {}.". format(self.name()))
self.computeScaleFactor()
self.computeSnapshots()
if self.N > 0:
if self.verbosity >= 7:
verbosityDepth("INIT", ("Starting computation of "
"denominator."))
TE = self._vander(self.radiusPade(self.mus), self.E)
while self.N > 0:
RHS = np.zeros(self.E + 1)
RHS[-1] = 1.
fitOut = customFit(TE.T, RHS, w = self.ws, full = True,
rcond = self.interpRcond)
if self.verbosity >= 5:
verbosityDepth("MAIN", ("Fitting {} samples with "
"degree {} through {}... "
"Conditioning of LS system: "
"{:.4e}.").format(
self.S, self.E, self._fitname,
fitOut[1][2][0] / fitOut[1][2][-1]))
if fitOut[1][1] < self.E + 1:
Enew = fitOut[1][1] - 1
Nnew = min(self.N, Enew)
Mnew = min(self.M, Enew)
if Nnew == self.N:
strN = ""
else:
strN = "N from {} to {} and ".format(self.N, Nnew)
if Mnew == self.M:
strM = ""
else:
strM = "M from {} to {} and ".format(self.M, Mnew)
warn(("Polyfit is poorly conditioned.\nReducing {}{}E "
"from {} to {}.").format(strN, strM,
self.E, Enew))
newParameters = {"N" : Nnew, "M" : Mnew, "E" : Enew}
self.approxParameters = newParameters
continue
G = (TE[:, : self.N + 1].T * fitOut[0]).T
if self.POD:
if self.verbosity >= 7:
verbosityDepth("INIT", ("Solving svd for square "
- "root of gramian matrix."),
- end = "")
+ "root of gramian matrix."))
G = self.samplingEngine.RPOD.dot(G)
_, s, eV = np.linalg.svd(G, full_matrices = False)
ev = s[::-1]
eV = eV[::-1, :].conj().T
if self.verbosity >= 5:
try: condev = s[0] / s[-1]
except: condev = np.inf
verbosityDepth("MAIN", ("Solved svd problem of "
"size {} x {} with "
"condition number "
"{:.4e}.").format(
self.S, self.N + 1, condev))
else:
if self.verbosity >= 10:
verbosityDepth("INIT", "Building gramian matrix.",
end = "")
G = self.samplingEngine.samples.dot(G)
G2 = self.HFEngine.innerProduct(G, G)
if self.verbosity >= 10:
verbosityDepth("DEL", "Done building gramian.",
inline = True)
if self.verbosity >= 7:
verbosityDepth("INIT", ("Solving eigenvalue "
"problem for gramian "
- "matrix."), end = "")
+ "matrix."))
ev, eV = np.linalg.eigh(G2)
if self.verbosity >= 5:
try: condev = ev[-1] / ev[0]
except: condev = np.inf
verbosityDepth("MAIN", ("Solved eigenvalue "
"problem of size {} with "
"condition number "
"{:.4e}.").format(
self.N + 1, condev))
if self.verbosity >= 7:
- verbosityDepth("DEL", " Done.", inline = True)
+ verbosityDepth("DEL", ("Done solving eigenvalue "
+ "problem."))
newParameters = checkRobustTolerance(ev, self.E,
self.robustTol)
if not newParameters:
break
self.approxParameters = newParameters
if self.N <= 0:
self._N = 0
eV = np.ones((1, 1))
self.Q = eV[:, 0]
if self.verbosity >= 7:
verbosityDepth("DEL", "Done computing denominator.")
else:
self.Q = np.ones(1, dtype = np.complex)
if self.verbosity >= 7:
verbosityDepth("INIT", "Starting computation of numerator.")
self.lastApproxParameters = copy(self.approxParameters)
Qevaldiag = np.diag(self.getQVal(self.mus))
while self.M >= 0:
fitVander = self._vander(self.radiusPade(self.mus), self.M)
fitOut = customFit(fitVander, Qevaldiag, w = self.ws,
full = True, rcond = self.interpRcond)
if self.verbosity >= 5:
verbosityDepth("MAIN", ("Fitting {} samples with degree "
"{} through {}... Conditioning of "
"LS system: {:.4e}.").format(
self.S, self.M, self._fitname,
fitOut[1][2][0] / fitOut[1][2][-1]))
if fitOut[1][1] == self.M + 1:
P = fitOut[0].T
break
warn(("Polyfit is poorly conditioned. Reducing M from {} to "
"{}. Exact snapshot interpolation not guaranteed.")\
.format(self.M, fitOut[1][1] - 1))
self.M = fitOut[1][1] - 1
if self.M <= 0:
raise Exception(("Instability in computation of numerator. "
"Aborting."))
self.P = np.atleast_2d(P)
if self.POD:
self.P = self.samplingEngine.RPOD.dot(self.P)
if self.verbosity >= 7:
verbosityDepth("DEL", "Done computing numerator.")
self.lastApproxParameters = copy(self.approxParameters)
if hasattr(self, "lastSolvedApp"): del self.lastSolvedApp
if self.verbosity >= 5:
verbosityDepth("DEL", "Done setting up approximant.\n")
def radiusPade(self, mu:Np1D, mu0 : float = None) -> float:
"""
Compute translated radius to be plugged into Pade' approximant.
Args:
mu: Parameter(s) 1.
mu0: Parameter(s) 2. If None, set to self.mu0.
Returns:
Translated radius to be plugged into Pade' approximant.
"""
if mu0 is None: mu0 = self.mu0
return ((self.HFEngine.rescaling(mu) - self.HFEngine.rescaling(mu0))
/ self.scaleFactor)
def getPVal(self, mu:List[complex]):
"""
Evaluate Pade' numerator at arbitrary parameter.
Args:
mu: Target parameter.
"""
self.setupApprox()
if self.verbosity >= 10:
verbosityDepth("INIT",
"Evaluating numerator at mu = {}.".format(mu),
end = "")
try:
len(mu)
except:
mu = [mu]
p = self._val(self.radiusPade(mu), self.P.T)
if len(mu) == 1:
p = p.flatten()
if self.verbosity >= 10:
verbosityDepth("DEL", " Done.", inline = True)
return p
def getQVal(self, mu:List[complex]):
"""
Evaluate Pade' denominator at arbitrary parameter.
Args:
mu: Target parameter.
"""
self.setupApprox()
if self.verbosity >= 10:
verbosityDepth("INIT",
"Evaluating denominator at mu = {}.".format(mu),
end = "")
q = self._val(self.radiusPade(mu), self.Q)
if self.verbosity >= 10:
verbosityDepth("DEL", " Done.", inline = True)
return q
def evalApproxReduced(self, mu:complex):
"""
Evaluate Pade' approximant at arbitrary parameter.
Args:
mu: Target parameter.
"""
self.setupApprox()
if (not hasattr(self, "lastSolvedApp")
or not np.isclose(self.lastSolvedApp, mu)):
if self.verbosity >= 5:
verbosityDepth("INIT",
"Evaluating approximant at mu = {}.".format(mu))
self.uAppReduced = self.getPVal(mu) / self.getQVal(mu)
self.lastSolvedApp = mu
if self.verbosity >= 5:
verbosityDepth("DEL", "Done evaluating approximant.")
def evalApprox(self, mu:complex):
"""
Evaluate approximant at arbitrary parameter.
Args:
mu: Target parameter.
"""
self.evalApproxReduced(mu)
self.uApp = self.samplingEngine.samples.dot(self.uAppReduced)
def getPoles(self) -> Np1D:
"""
Obtain approximant poles.
Returns:
Numpy complex vector of poles.
"""
self.setupApprox()
return self.HFEngine.rescalingInv(
self.scaleFactor * self._roots(self.Q)
+ self.HFEngine.rescaling(self.mu0))
diff --git a/rrompy/reduction_methods/lagrange_greedy/approximant_lagrange_greedy_pade.py b/rrompy/reduction_methods/lagrange_greedy/approximant_lagrange_greedy_pade.py
index 0b11da6..da631f7 100644
--- a/rrompy/reduction_methods/lagrange_greedy/approximant_lagrange_greedy_pade.py
+++ b/rrompy/reduction_methods/lagrange_greedy/approximant_lagrange_greedy_pade.py
@@ -1,402 +1,592 @@
# Copyright (C) 2018 by the RROMPy authors
#
# This file is part of RROMPy.
#
# RROMPy is free software: you can redistribute it and/or modify
# it under the terms of the GNU Lesser General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# RROMPy 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 RROMPy. If not, see .
#
from copy import copy
import numpy as np
from rrompy.reduction_methods.base import (checkRobustTolerance,
setupFitCallables)
from .generic_approximant_lagrange_greedy import (
GenericApproximantLagrangeGreedy)
from rrompy.reduction_methods.lagrange import ApproximantLagrangePade
from rrompy.utilities.base.types import DictAny, List, HFEng
from rrompy.utilities.base import purgeDict, verbosityDepth, customFit
from rrompy.utilities.warning_manager import warn
__all__ = ['ApproximantLagrangePadeGreedy']
class ApproximantLagrangePadeGreedy(GenericApproximantLagrangeGreedy,
ApproximantLagrangePade):
"""
ROM greedy Pade' interpolant computation for parametric problems.
Args:
HFEngine: HF problem solver.
mu0(optional): Default parameter. Defaults to 0.
approxParameters(optional): Dictionary containing values for main
parameters of approximant. Recognized keys are:
- 'POD': whether to compute POD of snapshots; defaults to True;
- 'muBounds': list of bounds for parameter values; defaults to
[[0], [1]];
- 'basis': type of basis for interpolation; allowed values include
'MONOMIAL', 'CHEBYSHEV' and 'LEGENDRE'; defaults to 'MONOMIAL';
- 'Delta': difference between M and N in rational approximant;
defaults to 0;
- 'greedyTol': uniform error tolerance for greedy algorithm;
defaults to 1e-2;
+ - 'errorEstimatorKind': kind of error estimator; available values
+ include 'EXACT' and 'SIMPLIFIED'; defaults to 'SIMPLIFIED';
- 'maxIter': maximum number of greedy steps; defaults to 1e2;
- 'refinementRatio': ratio of training points to be exhausted
before training set refinement; defaults to 0.2;
- 'nTrainingPoints': number of training points; defaults to
maxIter / refinementRatio;
- 'nTestPoints': number of starting test points; defaults to 1;
- 'trainingSetGenerator': training sample points generator;
defaults to uniform sampler within muBounds;
- 'interpRcond': tolerance for interpolation via numpy.polyfit;
defaults to None;
- 'robustTol': tolerance for robust Pade' denominator management;
defaults to 0.
Defaults to empty dict.
homogeneized: Whether to homogeneize Dirichlet BCs. Defaults to False.
verbosity(optional): Verbosity level. Defaults to 10.
Attributes:
HFEngine: HF problem solver.
mu0: Default parameter.
mus: Array of snapshot parameters.
homogeneized: Whether to homogeneize Dirichlet BCs.
approxParameters: Dictionary containing values for main parameters of
approximant. Recognized keys are in parameterList.
parameterList: Recognized keys of approximant parameters:
- 'POD': whether to compute POD of snapshots;
- 'muBounds': list of bounds for parameter values;
- 'basis': type of basis for interpolation;
- 'Delta': difference between M and N in rational approximant;
- 'greedyTol': uniform error tolerance for greedy algorithm;
+ - 'errorEstimatorKind': kind of error estimator;
- 'maxIter': maximum number of greedy steps;
- 'refinementRatio': ratio of training points to be exhausted
before training set refinement;
- 'nTrainingPoints': number of training points;
- 'nTestPoints': number of starting test points;
- 'trainingSetGenerator': training sample points generator;
- 'interpRcond': tolerance for interpolation via numpy.polyfit;
- 'robustTol': tolerance for robust Pade' denominator management.
extraApproxParameters: List of approxParameters keys in addition to
mother class's.
POD: whether to compute POD of snapshots.
muBounds: list of bounds for parameter values.
greedyTol: uniform error tolerance for greedy algorithm.
+ errorEstimatorKind: kind of error estimator.
maxIter: maximum number of greedy steps.
refinementRatio: ratio of training points to be exhausted before
training set refinement.
nTrainingPoints: number of training points.
nTestPoints: number of starting test points.
trainingSetGenerator: training sample points generator.
interpRcond: Tolerance for interpolation via numpy.polyfit.
robustTol: Tolerance for robust Pade' denominator management.
samplingEngine: Sampling engine.
uHF: High fidelity solution with wavenumber lastSolvedHF as numpy
complex vector.
lastSolvedHF: Wavenumber corresponding to last computed high fidelity
solution.
Q: Numpy 1D vector containing complex coefficients of approximant
denominator.
P: Numpy 2D vector whose columns are FE dofs of coefficients of
approximant numerator.
uApp: Last evaluated approximant as numpy complex vector.
lastApproxParameters: List of parameters corresponding to last
computed approximant.
"""
def __init__(self, HFEngine:HFEng, mu0 : complex = 0.,
approxParameters : DictAny = {}, homogeneized : bool = False,
verbosity : int = 10):
self._preInit()
- self._addParametersToList(["basis", "Delta", "interpRcond",
- "robustTol"])
+ self._addParametersToList(["basis", "Delta", "errorEstimatorKind",
+ "interpRcond", "robustTol"])
super().__init__(HFEngine = HFEngine, mu0 = mu0,
approxParameters = approxParameters,
homogeneized = homogeneized,
verbosity = verbosity)
self._postInit()
@property
def approxParameters(self):
"""
Value of approximant parameters. Its assignment may change robustTol.
"""
return self._approxParameters
@approxParameters.setter
def approxParameters(self, approxParams):
approxParameters = purgeDict(approxParams, self.parameterList,
dictname = self.name() + ".approxParameters",
baselevel = 1)
- approxParametersCopy = purgeDict(approxParameters, ["basis",
- "Delta",
+ approxParametersCopy = purgeDict(approxParameters, ["basis", "Delta",
+ "errorEstimatorKind",
"interpRcond",
"robustTol"],
True, True, baselevel = 1)
if "Delta" in list(approxParameters.keys()):
self._Delta = approxParameters["Delta"]
elif hasattr(self, "Delta"):
self._Delta = self.Delta
else:
self._Delta = 0
GenericApproximantLagrangeGreedy.approxParameters.fset(self,
approxParametersCopy)
keyList = list(approxParameters.keys())
self.Delta = self.Delta
if "basis" in keyList or not hasattr(self, "_val"):
if "basis" in keyList:
kind = approxParameters["basis"]
else:
kind = "MONOMIAL"
setupFit = setupFitCallables(kind)
for x in setupFit:
super().__setattr__("_" + x, setupFit[x])
+ if "errorEstimatorKind" in keyList:
+ self.errorEstimatorKind = approxParameters["errorEstimatorKind"]
+ elif hasattr(self, "errorEstimatorKind"):
+ self.errorEstimatorKind = self.errorEstimatorKind
+ else:
+ self.errorEstimatorKind = "SIMPLIFIED"
if "interpRcond" in keyList:
self.interpRcond = approxParameters["interpRcond"]
elif hasattr(self, "interpRcond"):
self.interpRcond = self.interpRcond
else:
self.interpRcond = None
if "robustTol" in keyList:
self.robustTol = approxParameters["robustTol"]
elif hasattr(self, "robustTol"):
self.robustTol = self.robustTol
else:
self.robustTol = 0
@property
def Delta(self):
"""Value of Delta."""
return self._Delta
@Delta.setter
def Delta(self, Delta):
if not np.isclose(Delta, np.floor(Delta)):
raise ArithmeticError("Delta must be an integer.")
if Delta < 0:
warn(("Error estimator unreliable for Delta < 0. Overloading of "
"errorEstimator is suggested."))
else:
Deltamin = (max(self.HFEngine.nbs,
self.HFEngine.nAs * self.homogeneized)
- 1 - 1 * (self.HFEngine.nAs > 1))
if Delta < Deltamin:
- warn(("Method unreliable for selected Delta. Suggested "
+ warn(("Method may be unreliable for selected Delta. Suggested "
"minimal value of Delta: {}.").format(Deltamin))
self._Delta = Delta
self._approxParameters["Delta"] = self.Delta
+ @property
+ def errorEstimatorKind(self):
+ """Value of errorEstimatorKind."""
+ return self._errorEstimatorKind
+ @errorEstimatorKind.setter
+ def errorEstimatorKind(self, errorEstimatorKind):
+ errorEstimatorKind = errorEstimatorKind.upper()
+ if errorEstimatorKind not in ["EXACT", "SIMPLIFIED"]:
+ warn(("Error estimator kind not recognized. Overriding to "
+ "'SIMPLIFIED'."))
+ errorEstimatorKind = "SIMPLIFIED"
+ self._errorEstimatorKind = errorEstimatorKind
+ self._approxParameters["errorEstimatorKind"] = self.errorEstimatorKind
+
@property
def nTestPoints(self):
"""Value of nTestPoints."""
return self._nTestPoints
@nTestPoints.setter
def nTestPoints(self, nTestPoints):
if nTestPoints <= np.abs(self.Delta):
warn(("nTestPoints must be at least abs(Delta) + 1. Increasing "
"value to abs(Delta) + 1."))
nTestPoints = np.abs(self.Delta) + 1
if not np.isclose(nTestPoints, np.int(nTestPoints)):
raise ArithmeticError("nTestPoints must be an integer.")
nTestPoints = np.int(nTestPoints)
if hasattr(self, "nTestPoints"):
nTestPointsold = self.nTestPoints
else: nTestPointsold = -1
self._nTestPoints = nTestPoints
self._approxParameters["nTestPoints"] = self.nTestPoints
if nTestPointsold != self.nTestPoints:
self.resetSamples()
+ def resetSamples(self):
+ """Reset samples."""
+ super().resetSamples()
+ self.resbb = None
+ self.resAb = None
+ self.resAA = None
+ self.As = None
+ self.bs = None
+
def errorEstimator(self, mus:List[np.complex]) -> List[np.complex]:
- """
- Standard residual-based error estimator. Unreliable for unstable
- problems.
- """
+ """Standard residual-based error estimator."""
self.setupApprox()
self.initEstNormer()
- nmus = len(mus)
- if self.N + 1 == self.S:
- QSf = self._domcoeff(self.N) * self.Q[-1] * self.HFEngine.b(
- self.mu0, 1, homogeneized = self.homogeneized)
- else:
- QSf = 0
- L1P = self._domcoeff(self.M) * self.HFEngine.A(self.mu0, 1).dot(
- self.samplingEngine.samples.dot(self.P[:, -1]))
- jOpt = self.estNormer.norm(QSf - L1P)
- if np.isnan(jOpt) or np.isinf(jOpt):
- err = np.empty(nmus)
+ PM = self.P[:, -1]
+ if np.any(np.isnan(PM)) or np.any(np.isinf(PM)):
+ err = np.empty(len(mus))
err[:] = np.inf
return err
- musTile = np.tile(self.HFEngine.rescaling(mus).reshape(-1, 1),
- [1, len(self.mus)])
- smusCol = self.HFEngine.rescaling(self.mus).reshape(1, -1)
- mussmus = np.abs(musTile - smusCol)
- num = np.prod(mussmus, axis = 1)
- den = np.abs(self.getQVal(mus))
- RHSnorms = np.empty(nmus)
- if self.HFEngine.nbs == 1:
- RHS = self.getRHS(mus[0], homogeneized = self.homogeneized)
- RHSnorms[:] = self.estNormer.norm(RHS)
+ nAs = self.HFEngine.nAs - 1
+ nbs = max(self.HFEngine.nbs - 1, nAs * self.homogeneized)
+ radiusmus = self.radiusPade(mus)
+ radiussmus = self.radiusPade(self.mus)
+ musTile = np.tile(radiusmus.reshape(-1, 1), [1, self.S])
+ smusCol = radiussmus.reshape(1, -1)
+ num = np.prod(musTile[:, : self.S] - smusCol, axis = 1)
+ den = self.getQVal(mus)
+ self.assembleReducedResidualBlocks()
+ vanderBase = np.polynomial.polynomial.polyvander(radiusmus,
+ max(nAs, nbs)).T
+ radiusb0 = vanderBase[: nbs + 1, :]
+ RHSnorms = np.power(np.abs(np.einsum('ij,ij->j',
+ self.resbb.dot(radiusb0),
+ radiusb0.conj())), .5)
+ vanderBase = vanderBase[: -1, :]
+ delta = self.S - self.N - 1
+ nbsEff = max(0, nbs - delta)
+ computeff = ((self.errorEstimatorKind == "SIMPLIFIED" and delta == 0)
+ or (self.errorEstimatorKind == "EXACT" and nbsEff > 0))
+ if self.errorEstimatorKind == "SIMPLIFIED":
+ radiusA = np.tensordot(PM, vanderBase[: nAs, :], 0)
+ if delta == 0:
+ radiusb = np.abs(self.Q[-1]) * radiusb0[: -1, :]
+ else: #if self.errorEstimatorKind == "EXACT":
+ momentQ = np.zeros(nbsEff, dtype = np.complex)
+ momentQu = np.zeros((self.S, nAs), dtype = np.complex)
+ radiusbTen = np.zeros((nbsEff, len(mus), nbsEff),
+ dtype = np.complex)
+ radiusATen = np.zeros((nAs, nAs, len(mus)), dtype = np.complex)
+ if nbsEff > 0:
+ momentQ[0] = self.Q[-1]
+ radiusbTen[:, :, 0] = vanderBase[: nbsEff, :]
+ momentQu[:, 0] = self.P[:, -1]
+ radiusATen[0, :, :] = vanderBase[: nAs, :]
+ Qvals = self.getQVal(self.mus)
+ for k in range(1, max(nAs, nbs)):
+ Qvals = Qvals * radiussmus
+ if k > delta and k < nbs:
+ momentQ[k - delta] = self._fitinv.dot(Qvals)
+ radiusbTen[k :, :, k - delta] = radiusbTen[: delta - k,
+ :, 0]
+ if k < nAs:
+ momentQu[:, k] = Qvals * self._fitinv
+ radiusATen[k, k :, :] = radiusATen[0, : - k, :]
+ if self.POD and nAs > 1:
+ momentQu[:, 1 :] = self.samplingEngine.RPOD.dot(momentQu[:,
+ 1 :])
+ radiusA = np.tensordot(momentQu, radiusATen, 1)
+ if nbsEff > 0:
+ radiusb = radiusbTen.dot(momentQ)
+ if computeff:
+ ff = np.einsum('ij,ij->j',
+ self.resbb[delta + 1 :, delta + 1 :].dot(radiusb),
+ radiusb.conj())
+ Lf = np.einsum('il,il->l',
+ np.tensordot(self.resAb[delta :, :, :], radiusA, 2),
+ radiusb.conj())
else:
- for j in range(nmus):
- RHS = self.getRHS(mus[j], homogeneized = self.homogeneized)
- RHSnorms[j] = self.estNormer.norm(RHS)
- return jOpt * num / den / RHSnorms / self.scaleFactor ** (self.S - 1)
+ ff, Lf = 0., 0.
+ LL = np.einsum('ikt,kit->t', np.tensordot(self.resAA, radiusA, 2),
+ radiusA.conj())
+ jOpt = np.power(np.abs(ff - 2. * np.real(Lf) + LL), .5)
+ return self._domcoeff(self.S - 1) * jOpt * np.abs(num / den) / RHSnorms
def setupApprox(self):
"""
Compute Pade' interpolant.
SVD-based robust eigenvalue management.
"""
if not self.checkComputedApprox():
if self.verbosity >= 5:
verbosityDepth("INIT", "Setting up {}.". format(self.name()))
self.computeScaleFactor()
self.S = len(self.mus)
self._M = self.S - 1
self._N = self.S - 1
if self.Delta < 0:
self._M += self.Delta
else:
self._N -= self.Delta
if min(self.M, self.N) < 0:
if self.verbosity >= 5:
verbosityDepth("MAIN", "Minimal sample size not achieved.")
self.Q = np.empty(max(self.N, 0) + 1, dtype = np.complex)
self.P = np.empty((len(self.mus), max(self.M, 0) + 1),
dtype = np.complex)
self.Q[:] = np.nan
self.P[:] = np.nan
self.lastApproxParameters = copy(self.approxParameters)
if self.verbosity >= 5:
verbosityDepth("DEL", ("Aborting computation of "
"approximant.\n"))
return
self.greedy()
if self.N > 0:
if self.verbosity >= 7:
verbosityDepth("INIT", ("Starting computation of "
"denominator."))
TS = self._vander(self.radiusPade(self.mus), self.S - 1)
while self.N > 0:
RHS = np.zeros(self.S)
RHS[-1] = 1.
fitOut = customFit(TS.T, RHS, full = True,
rcond = self.interpRcond)
if self.verbosity >= 2:
verbosityDepth("MAIN", ("Fitting {} samples with "
"degree {} through {}... "
"Conditioning of system: "
"{:.4e}.").format(self.S,
self.S - 1, self._fitname,
fitOut[1][2][0] / fitOut[1][2][-1]))
if fitOut[1][1] < self.S:
warn(("Polyfit is poorly conditioned. Starting "
"preemptive termination of computation of "
"approximant."))
self.Q = np.empty(max(self.N, 0) + 1,
dtype = np.complex)
self.P = np.empty((len(self.mus), max(self.M, 0) + 1),
dtype = np.complex)
self.Q[:] = np.nan
self.P[:] = np.nan
self.lastApproxParameters = copy(self.approxParameters)
if hasattr(self, "lastSolvedApp"):
del self.lastSolvedApp
if self.verbosity >= 7:
verbosityDepth("DEL", ("Aborting computation of "
"denominator."))
if self.verbosity >= 5:
verbosityDepth("DEL", ("Aborting computation of "
"approximant.\n"))
return
+ self._fitinv = fitOut[0]
G = (TS[:, : self.N + 1].T * fitOut[0]).T
if self.POD:
if self.verbosity >= 7:
verbosityDepth("INIT", ("Solving svd for square "
- "root of gramian matrix."),
- end = "")
+ "root of gramian matrix."))
G = self.samplingEngine.RPOD.dot(G)
_, s, eV = np.linalg.svd(G, full_matrices = False)
ev = s[::-1]
eV = eV[::-1, :].conj().T
if self.verbosity >= 2:
try: condev = s[0] / s[-1]
except: condev = np.inf
verbosityDepth("MAIN", ("Solved svd problem of "
"size {} x {} with "
"condition number "
"{:.4e}.").format(
self.S, self.N + 1, condev))
else:
if self.verbosity >= 10:
verbosityDepth("INIT", "Building gramian matrix.",
end = "")
G = self.samplingEngine.samples.dot(G)
G2 = self.HFEngine.innerProduct(G, G)
if self.verbosity >= 10:
verbosityDepth("DEL", "Done building gramian.",
inline = True)
if self.verbosity >= 7:
verbosityDepth("INIT", ("Solving eigenvalue "
"problem for gramian "
- "matrix."), end = "")
+ "matrix."))
ev, eV = np.linalg.eigh(G2)
if self.verbosity >= 2:
try: condev = ev[-1] / ev[0]
except: condev = np.inf
verbosityDepth("MAIN", ("Solved eigenvalue "
"problem of size {} with "
"condition number "
"{:.4e}.").format(
self.N + 1, condev))
if self.verbosity >= 7:
- verbosityDepth("DEL", " Done.", inline = True)
+ verbosityDepth("DEL", ("Done solving eigenvalue "
+ "problem."))
newParameters = checkRobustTolerance(ev, self.M,
self.robustTol)
if not newParameters:
break
self._N = newParameters["N"]
self._M = newParameters["E"]
if self.N <= 0:
self._N = 0
eV = np.ones((1, 1))
self.Q = eV[:, 0]
if self.verbosity >= 7:
verbosityDepth("DEL", "Done computing denominator.")
else:
self.Q = np.ones(1, dtype = np.complex)
if self.verbosity >= 7:
verbosityDepth("INIT", "Starting computation of numerator.")
self.lastApproxParameters = copy(self.approxParameters)
Qevaldiag = np.diag(self.getQVal(self.mus))
while self.M >= 0:
fitVander = self._vander(self.radiusPade(self.mus), self.M)
fitOut = customFit(fitVander, Qevaldiag, full = True,
rcond = self.interpRcond)
if fitOut[1][1] == self.M + 1:
P = fitOut[0].T
break
warn(("Polyfit is poorly conditioned. Reducing M from {} to "
"{}. Exact snapshot interpolation not guaranteed.")\
.format(self.M, fitOut[1][1] - 1))
self._M = fitOut[1][1] - 1
if self.M <= 0:
raise Exception(("Instability in computation of numerator. "
"Aborting."))
self.P = np.atleast_2d(P)
if self.POD:
self.P = self.samplingEngine.RPOD.dot(self.P)
if self.verbosity >= 7:
verbosityDepth("DEL", "Done computing numerator.")
self.lastApproxParameters = copy(self.approxParameters)
if hasattr(self, "lastSolvedApp"): del self.lastSolvedApp
if self.verbosity >= 5:
verbosityDepth("DEL", "Done setting up approximant.\n")
+ def assembleReducedResidualBlocks(self):
+ """Build affine blocks of reduced linear system through projections."""
+ self.initEstNormer()
+ if self.As is None or self.bs is None:
+ if self.verbosity >= 7:
+ verbosityDepth("INIT", "Computing Taylor blocks of system.")
+ nAs = self.HFEngine.nAs - 1
+ nbs = max(self.HFEngine.nbs, (nAs + 1) * self.homogeneized)
+ self.As = [self.HFEngine.A(self.mu0, j + 1) for j in range(nAs)]
+ self.bs = [self.HFEngine.b(self.mu0, j, self.homogeneized)
+ for j in range(nbs)]
+ if self.verbosity >= 7:
+ verbosityDepth("DEL", "Done computing Taylor blocks.")
+ computeResbb = self.resbb is None
+ computeResAb = self.resAb is None or self.resAb.shape[1] != self.S
+ computeResAA = self.resAA is None or self.resAA.shape[1] != self.S
+ samples = self.samplingEngine.samples
+ if computeResbb or computeResAb or computeResAA:
+ if self.verbosity >= 7:
+ verbosityDepth("INIT", "Projecting Taylor terms of residual.")
+ nAs = len(self.As)
+ nbs = len(self.bs) - 1
+ if computeResbb:
+ self.resbb = np.empty((nbs + 1, nbs + 1), dtype = np.complex)
+ for i in range(nbs + 1):
+ Mbi = self.scaleFactor ** i * self.bs[i]
+ for j in range(i):
+ Mbj = self.scaleFactor ** j * self.bs[j]
+ self.resbb[i, j] = self.estNormer.innerProduct(Mbj,
+ Mbi)
+ self.resbb[i, i] = self.estNormer.innerProduct(Mbi, Mbi)
+ for i in range(nbs + 1):
+ for j in range(i + 1, nbs + 1):
+ self.resbb[i, j] = self.resbb[j][i].conj()
+ if computeResAb:
+ if self.resAb is None:
+ self.resAb = np.empty((nbs, self.S, nAs),
+ dtype = np.complex)
+ for i in range(nbs):
+ Mbi = self.scaleFactor ** (i + 1) * self.bs[i + 1]
+ for j in range(nAs):
+ MAj = (self.scaleFactor ** (j + 1)
+ * self.As[j].dot(samples))
+ self.resAb[i, :, j] = self.estNormer.innerProduct(
+ MAj, Mbi)
+ else:
+ Sold = self.resAb.shape[1]
+ if Sold > self.S:
+ self.resAb = self.resAb[:, : self.S, :]
+ else:
+ resAbNew = np.empty((nbs, self.S, nAs),
+ dtype = np.complex)
+ resAbNew[:, : Sold, :] = self.resAb
+ self.resAb = resAbNew
+ for i in range(nbs):
+ Mbi = self.scaleFactor ** (i + 1) * self.bs[i + 1]
+ for j in range(nAs):
+ MAj = (self.scaleFactor ** (j + 1)
+ * self.As[j].dot(samples[:, Sold :]))
+ self.resAb[i, Sold :, j] = (
+ self.estNormer.innerProduct(MAj, Mbi))
+ if computeResAA:
+ if self.resAA is None:
+ self.resAA = np.empty((nAs, self.S, self.S, nAs),
+ dtype = np.complex)
+ for i in range(nAs):
+ MAi = (self.scaleFactor ** (i + 1)
+ * self.As[i].dot(samples))
+ for j in range(i):
+ MAj = (self.scaleFactor ** (j + 1)
+ * self.As[j].dot(samples))
+ self.resAA[i, :, :, j] = (
+ self.estNormer.innerProduct(MAj, MAi))
+ self.resAA[i, :, :, i] = self.estNormer.innerProduct(
+ MAi, MAi)
+ for i in range(nAs):
+ for j in range(i + 1, nAs):
+ self.resAA[i, :, :, j] = (
+ self.resAA[j, :, :, i].conj())
+ else:
+ Sold = self.resAA.shape[1]
+ if Sold > self.S:
+ self.resAA = self.resAA[:, : self.S, : self.S, :]
+ else:
+ resAANew = np.empty((nAs, self.S, self.S, nAs),
+ dtype = np.complex)
+ resAANew[:, : Sold, : Sold, :] = self.resAA
+ self.resAA = resAANew
+ for i in range(nAs):
+ MAi = (self.scaleFactor ** (i + 1)
+ * self.As[i].dot(samples))
+ for j in range(i):
+ MAj = (self.scaleFactor ** (j + 1)
+ * self.As[j].dot(samples))
+ self.resAA[i, : Sold, Sold :, j] = (
+ self.estNormer.innerProduct(MAj[:, Sold :],
+ MAi[:, : Sold]))
+ self.resAA[i, Sold :, : Sold, j] = (
+ self.estNormer.innerProduct(MAj[:, : Sold],
+ MAi[:, Sold :]))
+ self.resAA[i, Sold :, Sold :, j] = (
+ self.estNormer.innerProduct(MAj[:, Sold :],
+ MAi[:, Sold :]))
+ self.resAA[i, : Sold, Sold :, i] = (
+ self.estNormer.innerProduct(MAi[:, Sold :],
+ MAi[:, : Sold]))
+ self.resAA[i, Sold :, : Sold, i] = (
+ self.resAA[i, : Sold, Sold :, i].conj().T)
+ self.resAA[i, Sold :, Sold :, i] = (
+ self.estNormer.innerProduct(MAi[:, Sold :],
+ MAi[:, Sold :]))
+ for i in range(nAs):
+ for j in range(i + 1, nAs):
+ self.resAA[i, :, :, j] = (
+ self.resAA[j, :, :, i].conj())
+ if self.verbosity >= 7:
+ verbosityDepth("DEL", ("Done setting up Taylor "
+ "decomposition of residual."))
diff --git a/rrompy/reduction_methods/lagrange_greedy/approximant_lagrange_greedy_rb.py b/rrompy/reduction_methods/lagrange_greedy/approximant_lagrange_greedy_rb.py
index 50c46d6..eba8c03 100644
--- a/rrompy/reduction_methods/lagrange_greedy/approximant_lagrange_greedy_rb.py
+++ b/rrompy/reduction_methods/lagrange_greedy/approximant_lagrange_greedy_rb.py
@@ -1,303 +1,311 @@
# Copyright (C) 2018 by the RROMPy authors
#
# This file is part of RROMPy.
#
# RROMPy is free software: you can redistribute it and/or modify
# it under the terms of the GNU Lesser General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# RROMPy 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 RROMPy. If not, see .
#
import numpy as np
from copy import copy
from .generic_approximant_lagrange_greedy import (
GenericApproximantLagrangeGreedy)
from rrompy.reduction_methods.lagrange import ApproximantLagrangeRB
from rrompy.utilities.base.types import DictAny, HFEng, List
from rrompy.utilities.base import verbosityDepth
from rrompy.utilities.warning_manager import warn
__all__ = ['ApproximantLagrangeRBGreedy']
class ApproximantLagrangeRBGreedy(GenericApproximantLagrangeGreedy,
ApproximantLagrangeRB):
"""
ROM greedy RB approximant computation for parametric problems.
Args:
HFEngine: HF problem solver.
mu0(optional): Default parameter. Defaults to 0.
approxParameters(optional): Dictionary containing values for main
parameters of approximant. Recognized keys are:
- 'POD': whether to compute POD of snapshots; defaults to True;
- 'muBounds': list of bounds for parameter values; defaults to
[[0], [1]];
- 'greedyTol': uniform error tolerance for greedy algorithm;
defaults to 1e-2;
- 'maxIter': maximum number of greedy steps; defaults to 1e2;
- 'refinementRatio': ratio of training points to be exhausted
before training set refinement; defaults to 0.2;
- 'nTrainingPoints': number of training points; defaults to
maxIter / refinementRatio;
- 'nTestPoints': number of starting test points; defaults to 1;
- 'trainingSetGenerator': training sample points generator;
defaults to uniform sampler within muBounds.
Defaults to empty dict.
homogeneized: Whether to homogeneize Dirichlet BCs. Defaults to False.
verbosity(optional): Verbosity level. Defaults to 10.
Attributes:
HFEngine: HF problem solver.
mu0: Default parameter.
mus: Array of snapshot parameters.
homogeneized: Whether to homogeneize Dirichlet BCs.
approxParameters: Dictionary containing values for main parameters of
approximant. Recognized keys are in parameterList.
parameterList: Recognized keys of approximant parameters:
- 'POD': whether to compute POD of snapshots;
- 'muBounds': list of bounds for parameter values;
- 'greedyTol': uniform error tolerance for greedy algorithm;
- 'maxIter': maximum number of greedy steps;
- 'refinementRatio': ratio of training points to be exhausted
before training set refinement;
- 'nTrainingPoints': number of training points;
- 'nTestPoints': number of starting test points;
- 'trainingSetGenerator': training sample points generator;
- 'robustTol': tolerance for robust Pade' denominator management.
extraApproxParameters: List of approxParameters keys in addition to
mother class's.
POD: whether to compute POD of snapshots.
muBounds: list of bounds for parameter values.
greedyTol: uniform error tolerance for greedy algorithm.
maxIter: maximum number of greedy steps.
refinementRatio: ratio of training points to be exhausted before
training set refinement.
nTrainingPoints: number of training points.
nTestPoints: number of starting test points.
trainingSetGenerator: training sample points generator.
samplingEngine: Sampling engine.
projMat: Projection matrix for RB system assembly.
uHF: High fidelity solution with wavenumber lastSolvedHF as numpy
complex vector.
lastSolvedHF: Wavenumber corresponding to last computed high fidelity
solution.
uApp: Last evaluated approximant as numpy complex vector.
lastApproxParameters: List of parameters corresponding to last
computed approximant.
As: List of sparse matrices (in CSC format) representing coefficients
of linear system matrix wrt theta(mu).
bs: List of numpy vectors representing coefficients of linear system
RHS wrt theta(mu).
thetaAs: List of callables representing coefficients of linear system
matrix wrt mu.
thetabs: List of callables representing coefficients of linear system
RHS wrt mu.
ARBs: List of sparse matrices (in CSC format) representing coefficients
of compressed linear system matrix wrt theta(mu).
bRBs: List of numpy vectors representing coefficients of compressed
linear system RHS wrt theta(mu).
"""
def __init__(self, HFEngine:HFEng, mu0 : complex = 0.,
approxParameters : DictAny = {}, homogeneized : bool = False,
verbosity : int = 10):
self._preInit()
super().__init__(HFEngine = HFEngine, mu0 = mu0,
approxParameters = approxParameters,
homogeneized = homogeneized,
verbosity = verbosity)
- if self.verbosity >= 10:
- verbosityDepth("INIT", "Computing affine blocks of system.")
- self.As, self.thetaAs = self.HFEngine.affineBlocksA(self.mu0)
- self.bs, self.thetabs = self.HFEngine.affineBlocksb(self.mu0,
- self.homogeneized)
- if self.verbosity >= 10:
- verbosityDepth("DEL", "Done computing affine blocks.")
self._postInit()
@property
def S(self):
"""Value of S."""
return self._S
@S.setter
def S(self, S):
self._S = S
@property
def R(self):
"""Value of R."""
return self._S
@R.setter
def R(self, R):
warn(("R is used just to simplify inheritance, and its value cannot "
"be changed from that of S."))
def resetSamples(self):
"""Reset samples."""
super().resetSamples()
self.projMat = None
self.resbb = None
self.resAb = None
self.resAA = None
+ if self.verbosity >= 10:
+ verbosityDepth("INIT", "Computing affine blocks of system.")
+ self.As, self.thetaAs = self.HFEngine.affineBlocksA(self.mu0)
+ self.bs, self.thetabs = self.HFEngine.affineBlocksb(self.mu0,
+ self.homogeneized)
+ if self.verbosity >= 10:
+ verbosityDepth("DEL", "Done computing affine blocks.")
def checkComputedApprox(self) -> bool:
"""
Check if setup of new approximant is not needed.
Returns:
True if new setup is not needed. False otherwise.
"""
return (self.projMat is not None and super().checkComputedApprox())
def errorEstimator(self, mus:List[np.complex]) -> List[np.complex]:
"""
Standard residual-based error estimator. Unreliable for unstable
problems (inf-sup constant is missing).
"""
nmus = len(mus)
err = np.empty(nmus)
self.assembleReducedSystem()
self.assembleReducedResidualBlocks()
- nAs = len(self.As)
- nbs = len(self.bs)
+ nAs = self.resAA.shape[0]
+ nbs = self.resbb.shape[0]
for j in range(nmus):
mu = mus[j]
uAppReduced = self.getAppReduced(mu)
prodbb = 0.
prodAb = 0.
prodAA = 0.
for i1 in range(nbs):
rhobi1 = self.thetabs(mu, i1)
for i2 in range(nbs):
rhobi2 = self.thetabs(mu, i2).conj()
- prodbb += rhobi1 * rhobi2 * self.resbb[i1][i2]
+ prodbb += rhobi1 * rhobi2 * self.resbb[i2, i1]
for i1 in range(nAs):
rhoAi1 = self.thetaAs(mu, i1)
for i2 in range(nbs):
rhobi2 = self.thetabs(mu, i2).conj()
- prodAb += rhoAi1 * rhobi2 * self.resAb[i1][i2]
+ prodAb += rhoAi1 * rhobi2 * self.resAb[i2, i1, :]
for i1 in range(nAs):
rhoAi1 = self.thetaAs(mu, i1)
for i2 in range(nAs):
rhoAi2 = self.thetaAs(mu, i2).conj()
- prodAA += rhoAi1 * rhoAi2 * self.resAA[i1][i2]
+ prodAA += rhoAi1 * rhoAi2 * self.resAA[i2, i1, :, :]
err[j] = np.abs(((uAppReduced.T.conj().dot(prodAA.dot(uAppReduced))
- 2. * np.real(prodAb.dot(uAppReduced))) / prodbb
+ 1.)[0]) ** .5
return err
def setupApprox(self):
"""Compute RB projection matrix."""
if not self.checkComputedApprox():
if self.verbosity >= 5:
verbosityDepth("INIT", "Setting up {}.". format(self.name()))
self.greedy()
self.S = len(self.mus)
if self.verbosity >= 7:
verbosityDepth("INIT", "Computing projection matrix.",
end = "")
self.projMat = self.samplingEngine.samples
if self.verbosity >= 7:
verbosityDepth("DEL", " Done.", inline = True)
self.lastApproxParameters = copy(self.approxParameters)
if hasattr(self, "lastSolvedApp"): del self.lastSolvedApp
if self.verbosity >= 5:
verbosityDepth("DEL", "Done setting up approximant.\n")
def assembleReducedResidualBlocks(self):
"""Build affine blocks of RB linear system through projections."""
self.initEstNormer()
self.assembleReducedSystem()
computeResbb = self.resbb is None
- computeResAb = (self.resAb is None
- or self.resAb[0][0].shape[0] != self.S)
- computeResAA = (self.resAA is None
- or self.resAA[0][0].shape[0] != self.S)
+ computeResAb = self.resAb is None or self.resAb.shape[2] != self.S
+ computeResAA = self.resAA is None or self.resAA.shape[2] != self.S
+ samples = self.projMat
if computeResbb or computeResAb or computeResAA:
if self.verbosity >= 7:
verbosityDepth("INIT", "Projecting affine terms of residual.",
end = "")
nAs = len(self.As)
- nbs = len(self.bs)
+ nbs = max(len(self.bs), nAs * self.homogeneized)
if computeResbb:
- self.resbb = []
+ self.resbb = np.empty((nbs, nbs), dtype = np.complex)
for i in range(nbs):
- resbbi = []
Mbi = self.bs[i]
for j in range(i):
Mbj = self.bs[j]
- resbbi += [self.estNormer.innerProduct(Mbi, Mbj)]
- resbbi += [self.estNormer.innerProduct(Mbi, Mbi)]
- self.resbb += [resbbi]
+ self.resbb[i, j] = self.estNormer.innerProduct(Mbj,
+ Mbi)
+ self.resbb[i, i] = self.estNormer.innerProduct(Mbi, Mbi)
for i in range(nbs):
for j in range(i + 1, nbs):
- self.resbb[i] += [self.resbb[j][i].conj()]
+ self.resbb[i, j] = self.resbb[j][i].conj()
if computeResAb:
if self.resAb is None:
- self.resAb = []
- for i in range(nAs):
- resAbi = []
- MAi = self.As[i].dot(self.projMat)
- for j in range(nbs):
- Mbj = self.bs[j]
- resAbi += [self.estNormer.innerProduct(MAi, Mbj)]
- self.resAb += [resAbi]
+ self.resAb = np.empty((nbs, nAs, self.S),
+ dtype = np.complex)
+ for i in range(nbs):
+ Mbi = self.bs[i]
+ for j in range(nAs):
+ MAj = self.As[j].dot(samples)
+ self.resAb[i, j, :] = self.estNormer.innerProduct(
+ MAj, Mbi)
else:
- Sold = self.resAb[0][0].shape[0]
- for i in range(nAs):
- MAi = self.As[i].dot(self.projMat[:, Sold :])
- for j in range(nbs):
- Mbj = self.bs[j]
- self.resAb[i][j] = np.append(self.resAb[i][j],
- self.estNormer.innerProduct(MAi, Mbj))
+ resAbNew = np.empty((nbs, nAs, self.S), dtype = np.complex)
+ Sold = self.resAb.shape[2]
+ resAbNew[:, :, : Sold] = self.resAb
+ self.resAb = resAbNew
+ for i in range(nbs):
+ Mbi = self.bs[i]
+ for j in range(nAs):
+ MAj = self.As[j].dot(samples[:, Sold :])
+ self.resAb[i, j, Sold :] = (
+ self.estNormer.innerProduct(MAj, Mbi))
if computeResAA:
if self.resAA is None:
- self.resAA = []
+ self.resAA = np.empty((nAs, nAs, self.S, self.S),
+ dtype = np.complex)
for i in range(nAs):
- resAAi = []
- MAi = self.As[i].dot(self.projMat)
+ MAi = self.As[i].dot(samples)
for j in range(i):
- MAj = self.As[j].dot(self.projMat)
- resAAi += [self.estNormer.innerProduct(MAi, MAj)]
- resAAi += [self.estNormer.innerProduct(MAi, MAi)]
- self.resAA += [resAAi]
+ MAj = self.As[j].dot(samples)
+ self.resAA[i, j, :, :] = (
+ self.estNormer.innerProduct(MAj, MAi))
+ self.resAA[i, i, :, :] = self.estNormer.innerProduct(
+ MAi, MAi)
for i in range(nAs):
for j in range(i + 1, nAs):
- self.resAA[i] += [self.resAA[j][i].conj()]
+ self.resAA[i, j, :, :] = (
+ self.resAA[j, i, :, :].conj())
else:
- Sold = self.resAA[0][0].shape[0]
+ resAANew = np.empty((nAs, nAs, self.S, self.S),
+ dtype = np.complex)
+ Sold = self.resAA.shape[2]
+ resAANew[:, :, : Sold, : Sold] = self.resAA
+ self.resAA = resAANew
for i in range(nAs):
- MAi = self.As[i].dot(self.projMat)
+ MAi = self.As[i].dot(samples)
for j in range(i):
- MAj = self.As[j].dot(self.projMat)
- resAAcross1 = self.estNormer.innerProduct(
- MAi[:, Sold :], MAj[:, : Sold])
- resAAcross2 = self.estNormer.innerProduct(
- MAi[:, : Sold], MAj[:, Sold :])
- resAAdiag = self.estNormer.innerProduct(
- MAi[:, Sold :], MAj[:, Sold :])
- self.resAA[i][j] = np.block(
- [[self.resAA[i][j], resAAcross1],
- [ resAAcross2, resAAdiag]])
- resAAcross = self.estNormer.innerProduct(
- MAi[:, Sold :],
- MAi[:, : Sold])
- resAAdiag = self.estNormer.innerProduct(MAi[:, Sold :],
- MAi[:, Sold :])
- self.resAA[i][i] = np.block(
- [[ self.resAA[i][i], resAAcross],
- [resAAcross.T.conj(), resAAdiag]])
+ MAj = self.As[j].dot(samples)
+ self.resAA[i, j, : Sold, Sold :] = (
+ self.estNormer.innerProduct(MAj[:, Sold :],
+ MAi[:, : Sold]))
+ self.resAA[i, j, Sold :, : Sold] = (
+ self.estNormer.innerProduct(MAj[:, : Sold],
+ MAi[:, Sold :]))
+ self.resAA[i, j, Sold :, Sold :] = (
+ self.estNormer.innerProduct(MAj[:, Sold :],
+ MAi[:, Sold :]))
+ self.resAA[i, i, : Sold, Sold :] = (
+ self.estNormer.innerProduct(MAi[:, Sold :],
+ MAi[:, : Sold]))
+ self.resAA[i, i, Sold :, : Sold] = (
+ self.resAA[i, i, : Sold, Sold :].conj().T)
+ self.resAA[i, i, Sold :, Sold :] = (
+ self.estNormer.innerProduct(MAi[:, Sold :],
+ MAi[:, Sold :]))
for i in range(nAs):
for j in range(i + 1, nAs):
- self.resAA[i][j] = self.resAA[j][i].conj()
+ self.resAA[i, j, :, :] = (
+ self.resAA[j, i, :, :].conj())
if self.verbosity >= 7:
verbosityDepth("DEL", ("Done setting up affine "
"decomposition of residual."))
diff --git a/rrompy/reduction_methods/lagrange_greedy/generic_approximant_lagrange_greedy.py b/rrompy/reduction_methods/lagrange_greedy/generic_approximant_lagrange_greedy.py
index 9ed10f3..fa7936d 100644
--- a/rrompy/reduction_methods/lagrange_greedy/generic_approximant_lagrange_greedy.py
+++ b/rrompy/reduction_methods/lagrange_greedy/generic_approximant_lagrange_greedy.py
@@ -1,488 +1,489 @@
# Copyright (C) 2018 by the RROMPy authors
#
# This file is part of RROMPy.
#
# RROMPy is free software: you can redistribute it and/or modify
# it under the terms of the GNU Lesser General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# RROMPy 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 RROMPy. If not, see .
#
import numpy as np
from rrompy.reduction_methods.base.generic_approximant import (
GenericApproximant)
from rrompy.reduction_methods.lagrange.generic_approximant_lagrange import (
GenericApproximantLagrange)
from rrompy.utilities.base.types import DictAny, HFEng, Tuple, List
from rrompy.utilities.base import purgeDict, verbosityDepth
from rrompy.utilities.warning_manager import warn
__all__ = ['GenericApproximantLagrangeGreedy']
class GenericApproximantLagrangeGreedy(GenericApproximantLagrange):
"""
ROM greedy Lagrange interpolant computation for parametric problems
(ABSTRACT).
Args:
HFEngine: HF problem solver.
mu0(optional): Default parameter. Defaults to 0.
approxParameters(optional): Dictionary containing values for main
parameters of approximant. Recognized keys are:
- 'POD': whether to compute POD of snapshots; defaults to True;
- 'muBounds': list of bounds for parameter values; defaults to
[[0], [1]];
- 'greedyTol': uniform error tolerance for greedy algorithm;
defaults to 1e-2;
- 'maxIter': maximum number of greedy steps; defaults to 1e2;
- 'refinementRatio': ratio of training points to be exhausted
before training set refinement; defaults to 0.2;
- 'nTrainingPoints': number of training points; defaults to
maxIter / refinementRatio;
- 'nTestPoints': number of starting test points; defaults to 1;
- 'trainingSetGenerator': training sample points generator;
defaults to uniform sampler within muBounds;
- 'robustTol': tolerance for robust Pade' denominator management;
defaults to 0.
Defaults to empty dict.
homogeneized: Whether to homogeneize Dirichlet BCs. Defaults to False.
verbosity(optional): Verbosity level. Defaults to 10.
Attributes:
HFEngine: HF problem solver.
mu0: Default parameter.
mus: Array of snapshot parameters.
homogeneized: Whether to homogeneize Dirichlet BCs.
approxParameters: Dictionary containing values for main parameters of
approximant. Recognized keys are in parameterList.
parameterList: Recognized keys of approximant parameters:
- 'POD': whether to compute POD of snapshots;
- 'muBounds': list of bounds for parameter values;
- 'greedyTol': uniform error tolerance for greedy algorithm;
- 'maxIter': maximum number of greedy steps;
- 'refinementRatio': ratio of training points to be exhausted
before training set refinement;
- 'nTrainingPoints': number of training points;
- 'nTestPoints': number of starting test points;
- 'trainingSetGenerator': training sample points generator.
- 'robustTol': tolerance for robust Pade' denominator management.
extraApproxParameters: List of approxParameters keys in addition to
mother class's.
POD: whether to compute POD of snapshots.
muBounds: list of bounds for parameter values.
greedyTol: uniform error tolerance for greedy algorithm.
maxIter: maximum number of greedy steps.
refinementRatio: ratio of training points to be exhausted before
training set refinement.
nTrainingPoints: number of training points.
nTestPoints: number of starting test points.
trainingSetGenerator: training sample points generator.
robustTol: tolerance for robust Pade' denominator management.
samplingEngine: Sampling engine.
uHF: High fidelity solution with wavenumber lastSolvedHF as numpy
complex vector.
lastSolvedHF: Wavenumber corresponding to last computed high fidelity
solution.
uApp: Last evaluated approximant as numpy complex vector.
lastApproxParameters: List of parameters corresponding to last
computed approximant.
"""
TOL_INSTABILITY = 1e-6
def __init__(self, HFEngine:HFEng, mu0 : complex = 0.,
approxParameters : DictAny = {}, homogeneized : bool = False,
verbosity : int = 10):
self._preInit()
self._addParametersToList(["muBounds","greedyTol", "maxIter",
"refinementRatio", "nTrainingPoints",
"nTestPoints", "trainingSetGenerator"])
super(GenericApproximantLagrange, self).__init__(
HFEngine = HFEngine, mu0 = mu0,
approxParameters = approxParameters,
homogeneized = homogeneized,
verbosity = verbosity)
self._postInit()
@property
def approxParameters(self):
"""Value of approximant parameters. Its assignment may change S."""
return self._approxParameters
@approxParameters.setter
def approxParameters(self, approxParams):
approxParameters = purgeDict(approxParams, self.parameterList,
dictname = self.name() + ".approxParameters",
baselevel = 1)
approxParametersCopy = purgeDict(approxParameters,
["muBounds","greedyTol", "maxIter",
"refinementRatio", "nTrainingPoints",
"nTestPoints",
"trainingSetGenerator"],
True, True, baselevel = 1)
GenericApproximant.approxParameters.fset(self, approxParametersCopy)
keyList = list(approxParameters.keys())
if "muBounds" in keyList:
self.muBounds = approxParameters["muBounds"]
elif hasattr(self, "muBounds"):
self.muBounds = self.muBounds
else:
self.muBounds = [[0.], [1.]]
if "greedyTol" in keyList:
self.greedyTol = approxParameters["greedyTol"]
elif hasattr(self, "greedyTol"):
self.greedyTol = self.greedyTol
else:
self.greedyTol = 1e-2
if "maxIter" in keyList:
self.maxIter = approxParameters["maxIter"]
elif hasattr(self, "maxIter"):
self.maxIter = self.maxIter
else:
self.maxIter = 1e2
if "refinementRatio" in keyList:
self.refinementRatio = approxParameters["refinementRatio"]
elif hasattr(self, "refinementRatio"):
self.refinementRatio = self.refinementRatio
else:
self.refinementRatio = 0.2
if "nTrainingPoints" in keyList:
self.nTrainingPoints = approxParameters["nTrainingPoints"]
elif hasattr(self, "nTrainingPoints"):
self.nTrainingPoints = self.nTrainingPoints
else:
self.nTrainingPoints = np.int(np.ceil(self.maxIter
/ self.refinementRatio))
if "nTestPoints" in keyList:
self.nTestPoints = approxParameters["nTestPoints"]
elif hasattr(self, "nTestPoints"):
self.nTestPoints = self.nTestPoints
else:
self.nTestPoints = 1
if "trainingSetGenerator" in keyList:
self.trainingSetGenerator = (
approxParameters["trainingSetGenerator"])
elif hasattr(self, "trainingSetGenerator"):
self.trainingSetGenerator = self.trainingSetGenerator
else:
from rrompy.utilities.parameter_sampling import QuadratureSampler
self.trainingSetGenerator = QuadratureSampler(self.muBounds,
"UNIFORM")
del QuadratureSampler
@property
def S(self):
"""Value of S."""
return self._S
@S.setter
def S(self, S):
self._S = S
@property
def mus(self):
"""Value of mus."""
return self._mus
@mus.setter
def mus(self, mus):
self._mus = np.array(mus, dtype = np.complex)
@property
def muBounds(self):
"""Value of muBounds."""
return self._muBounds
@muBounds.setter
def muBounds(self, muBounds):
if len(muBounds) != 2:
raise Exception("2 limits must be specified.")
try:
muBounds = muBounds.tolist()
except:
muBounds = list(muBounds)
for j in range(2):
try:
len(muBounds[j])
except:
muBounds[j] = np.array([muBounds[j]])
if len(muBounds[0]) != len(muBounds[1]):
raise Exception("The bounds must have the same length.")
self._muBounds = muBounds
@property
def greedyTol(self):
"""Value of greedyTol."""
return self._greedyTol
@greedyTol.setter
def greedyTol(self, greedyTol):
if greedyTol < 0:
raise ArithmeticError("greedyTol must be non-negative.")
if hasattr(self, "greedyTol"): greedyTolold = self.greedyTol
else: greedyTolold = -1
self._greedyTol = greedyTol
self._approxParameters["greedyTol"] = self.greedyTol
if greedyTolold != self.greedyTol:
self.resetSamples()
@property
def maxIter(self):
"""Value of maxIter."""
return self._maxIter
@maxIter.setter
def maxIter(self, maxIter):
if maxIter <= 0: raise ArithmeticError("maxIter must be positive.")
if hasattr(self, "maxIter"): maxIterold = self.maxIter
else: maxIterold = -1
self._maxIter = maxIter
self._approxParameters["maxIter"] = self.maxIter
if maxIterold != self.maxIter:
self.resetSamples()
@property
def refinementRatio(self):
"""Value of refinementRatio."""
return self._refinementRatio
@refinementRatio.setter
def refinementRatio(self, refinementRatio):
if refinementRatio <= 0. or refinementRatio > 1.:
raise ArithmeticError(("refinementRatio must be between 0 "
"(excluded) and 1."))
if hasattr(self, "refinementRatio"):
refinementRatioold = self.refinementRatio
else: refinementRatioold = -1
self._refinementRatio = refinementRatio
self._approxParameters["refinementRatio"] = self.refinementRatio
if refinementRatioold != self.refinementRatio:
self.resetSamples()
@property
def nTrainingPoints(self):
"""Value of nTrainingPoints."""
return self._nTrainingPoints
@nTrainingPoints.setter
def nTrainingPoints(self, nTrainingPoints):
if nTrainingPoints <= 1:
raise ArithmeticError("nTrainingPoints must be greater than 1.")
if not np.isclose(nTrainingPoints, np.int(nTrainingPoints)):
raise ArithmeticError("nTrainingPoints must be an integer.")
nTrainingPoints = np.int(nTrainingPoints)
if hasattr(self, "nTrainingPoints"):
nTrainingPointsold = self.nTrainingPoints
else: nTrainingPointsold = -1
self._nTrainingPoints = nTrainingPoints
self._approxParameters["nTrainingPoints"] = self.nTrainingPoints
if nTrainingPointsold != self.nTrainingPoints:
self.resetSamples()
@property
def nTestPoints(self):
"""Value of nTestPoints."""
return self._nTestPoints
@nTestPoints.setter
def nTestPoints(self, nTestPoints):
if nTestPoints <= 0:
raise ArithmeticError("nTestPoints must be positive.")
if not np.isclose(nTestPoints, np.int(nTestPoints)):
raise ArithmeticError("nTestPoints must be an integer.")
nTestPoints = np.int(nTestPoints)
if hasattr(self, "nTestPoints"):
nTestPointsold = self.nTestPoints
else: nTestPointsold = -1
self._nTestPoints = nTestPoints
self._approxParameters["nTestPoints"] = self.nTestPoints
if nTestPointsold != self.nTestPoints:
self.resetSamples()
@property
def trainingSetGenerator(self):
"""Value of trainingSetGenerator."""
return self._trainingSetGenerator
@trainingSetGenerator.setter
def trainingSetGenerator(self, trainingSetGenerator):
if 'generatePoints' not in dir(trainingSetGenerator):
raise Exception("trainingSetGenerator type not recognized.")
if hasattr(self, '_trainingSetGenerator'):
trainingSetGeneratorOld = self.trainingSetGenerator
self._trainingSetGenerator = trainingSetGenerator
self._approxParameters["trainingSetGenerator"] = (
self.trainingSetGenerator)
if (not 'trainingSetGeneratorOld' in locals()
or trainingSetGeneratorOld != self.trainingSetGenerator):
self.resetSamples()
def resetSamples(self):
"""Reset samples."""
super().resetSamples()
self._mus = []
def initEstNormer(self):
"""Initialize estimator norm engine."""
if not hasattr(self, "estNormer"):
from rrompy.hfengines.base import ProblemEngineBase as PEB
# class L2InverseNormer(PEB):
# def innerProduct(self, u:Np1D, v:Np1D) -> float:
# if not hasattr(self, "energyNormMatrix"):
# self.buildEnergyNormForm()
# return v.conj().T.dot(self.energyNormMatrix.solve(u))
# def buildEnergyNormForm(self):
# super().buildEnergyNormForm()
# from scipy.sparse import csc_matrix, linalg as sla
# Mass = csc_matrix(self.energyNormMatrix,
# dtype = np.complex)
# self.energyNormMatrix = sla.spilu(Mass)
# self.estNormer = L2InverseNormer() # inverse L2 norm
self.estNormer = PEB() # L2 norm
self.estNormer.V = self.HFEngine.V
self.estNormer.buildEnergyNormForm()
def errorEstimator(self, mus:List[np.complex]) -> List[np.complex]:
"""
Standard residual-based error estimator with explicit residual
computation.
"""
self.setupApprox()
nmus = len(mus)
err = np.empty(nmus)
if self.HFEngine.nbs == 1:
RHS = self.getRHS(mus[0], homogeneized = self.homogeneized)
RHSNorm = self.estNormer.norm(RHS)
for j in range(nmus):
res = self.getRes(mus[j], homogeneized = self.homogeneized)
err[j] = self.estNormer.norm(res) / RHSNorm
else:
for j in range(nmus):
res = self.getRes(mus[j], homogeneized = self.homogeneized)
RHS = self.getRHS(mus[j], homogeneized = self.homogeneized)
err[j] = self.estNormer.norm(res) / self.estNormer.norm(RHS)
return np.abs(err)
def getMaxErrorEstimator(self) -> Tuple[float, int]:
"""
Compute maximum of (and index of maximum of) error estimator over
training set.
"""
errorEstTrain = self.errorEstimator(self.muTrain)
idxMaxEst = np.argmax(errorEstTrain)
maxEst = errorEstTrain[idxMaxEst]
return maxEst, idxMaxEst
def greedyNextSample(self, muidx:int, plotEst : bool = False):
"""Compute next greedy snapshot of solution map."""
mu = self.muTrain[muidx]
if self.verbosity >= 2:
verbosityDepth("MAIN", ("Adding {}-th sample point at {} to test "
"set.").format(
self.samplingEngine.nsamples + 1, mu))
self.mus = np.append(self.mus, mu)
idxs = np.arange(len(self.muTrain))
mask = np.ones_like(idxs, dtype = bool)
mask[muidx] = False
idxs = idxs[mask]
self.muTrain = self.muTrain[idxs]
self.samplingEngine.nextSample(mu,
homogeneized = self.homogeneized)
errorEstTrain = self.errorEstimator(self.muTrain)
muidx = np.argmax(errorEstTrain)
maxErrorEst = errorEstTrain[muidx]
mu = self.muTrain[muidx]
if plotEst and not np.all(np.isinf(errorEstTrain)):
from matplotlib import pyplot as plt
plt.figure()
plt.semilogy(np.real(self.muTrain), errorEstTrain, 'k')
plt.semilogy(np.real(self.muTrain),
self.greedyTol * np.ones(len(self.muTrain)), 'r--')
plt.semilogy(np.real(self.mus),
2. * self.greedyTol * np.ones(len(self.mus)), '*m')
plt.semilogy(np.real(mu), maxErrorEst, 'xr')
plt.grid()
plt.show()
plt.close()
return errorEstTrain, muidx, maxErrorEst, mu
def greedy(self, plotEst : bool = False):
"""Compute greedy snapshots of solution map."""
if self.samplingEngine.samples is None:
if self.verbosity >= 2:
verbosityDepth("INIT", "Starting computation of snapshots.")
self.resetSamples()
self.mus, _ = self.trainingSetGenerator.generatePoints(
self.nTestPoints)
self.mus = np.array([x[0] for x in self.mus], dtype = np.complex)
nTrain = self.nTrainingPoints
muTrainBase, _ = self.trainingSetGenerator.generatePoints(nTrain)
self.muTrain = np.empty(len(muTrainBase) + 1, dtype = np.complex)
j = 0
for mu in muTrainBase:
if not np.any(np.isclose(self.mus, mu)):
self.muTrain[j] = mu[0]
j += 1
self.muTrain[j] = self.mus[-1]
self.muTrain = self.muTrain[: j + 1]
self.mus = self.mus[:-1]
for j in range(len(self.mus)):
if self.verbosity >= 2:
verbosityDepth("MAIN", ("Adding {}-th sample point at {} "
"to test set.").format(
self.samplingEngine.nsamples + 1,
self.mus[j]))
self.samplingEngine.nextSample(self.mus[j],
homogeneized = self.homogeneized)
errorEstTrain, muidx, maxErrorEst, mu = self.greedyNextSample(-1,
plotEst)
if self.verbosity >= 2:
verbosityDepth("MAIN", ("Uniform error estimate {:.4e}.")\
.format(maxErrorEst))
while (self.samplingEngine.nsamples < self.maxIter
and (maxErrorEst > self.greedyTol
or self.samplingEngine.nsamples < self.nTestPoints)):
if (1. - self.refinementRatio) * nTrain > len(self.muTrain):
muTrainExtra, _ = self.trainingSetGenerator.refine(nTrain)
self.muTrain = np.sort(np.append(self.muTrain,
muTrainExtra))
nTrain += len(muTrainExtra)
if self.verbosity >= 5:
verbosityDepth("MAIN", ("Enriching training set by {} "
"elements.").format(
len(muTrainExtra)))
muTrainOld, maxErrorEstOld = self.muTrain, maxErrorEst
errorEstTrain, muidx, maxErrorEst, mu = self.greedyNextSample(
muidx, plotEst)
if self.verbosity >= 2:
verbosityDepth("MAIN", ("Uniform error estimate {:.4e}.")\
.format(maxErrorEst))
- if maxErrorEstOld < maxErrorEst * self.TOL_INSTABILITY:
+ if (np.isnan(maxErrorEst) or np.isinf(maxErrorEst)
+ or maxErrorEstOld < maxErrorEst * self.TOL_INSTABILITY):
warn(("Instability in a posteriori estimator. Starting "
"preemptive greedy loop termination."))
maxErrorEst = maxErrorEstOld
self.muTrain = muTrainOld
self.mus = self.mus[:-1]
self.samplingEngine.popSample()
self.setupApprox()
break
if self.verbosity >= 2:
verbosityDepth("DEL", ("Done computing snapshots (final "
"snapshot count: {}).").format(
self.samplingEngine.nsamples))
def checkComputedApprox(self) -> bool:
"""
Check if setup of new approximant is not needed.
Returns:
True if new setup is not needed. False otherwise.
"""
return (hasattr(self, "_S") and self.S == len(self.mus)
and super().checkComputedApprox())
def computeScaleFactor(self):
"""Compute parameter rescaling factor."""
self.scaleFactor = .5 * np.abs(self.HFEngine.rescaling(
self.trainingSetGenerator.lims[0][0])
- self.HFEngine.rescaling(
self.trainingSetGenerator.lims[1][0]))
diff --git a/rrompy/reduction_methods/taylor/approximant_taylor_pade.py b/rrompy/reduction_methods/taylor/approximant_taylor_pade.py
index 7a2a162..647a1d5 100644
--- a/rrompy/reduction_methods/taylor/approximant_taylor_pade.py
+++ b/rrompy/reduction_methods/taylor/approximant_taylor_pade.py
@@ -1,546 +1,546 @@
# Copyright (C) 2018 by the RROMPy authors
#
# This file is part of RROMPy.
#
# RROMPy is free software: you can redistribute it and/or modify
# it under the terms of the GNU Lesser General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# RROMPy 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 RROMPy. If not, see .
#
from copy import copy
import numpy as np
from rrompy.reduction_methods.base import checkRobustTolerance
from .generic_approximant_taylor import GenericApproximantTaylor
from rrompy.sampling.base.pod_engine import PODEngine
from rrompy.utilities.base.types import Np1D, Np2D, List, Tuple, DictAny
from rrompy.utilities.base.types import HFEng
from rrompy.utilities.base import purgeDict, verbosityDepth
from rrompy.utilities.warning_manager import warn
__all__ = ['ApproximantTaylorPade']
class ApproximantTaylorPade(GenericApproximantTaylor):
"""
ROM single-point fast Pade' approximant computation for parametric
problems.
Args:
HFEngine: HF problem solver.
mu0(optional): Default parameter. Defaults to 0.
approxParameters(optional): Dictionary containing values for main
parameters of approximant. Recognized keys are:
- 'POD': whether to compute POD of snapshots; defaults to True;
- 'rho': weight for computation of original Pade' approximant;
defaults to np.inf, i.e. fast approximant;
- 'M': degree of Pade' approximant numerator; defaults to 0;
- 'N': degree of Pade' approximant denominator; defaults to 0;
- 'E': total number of derivatives current approximant relies upon;
defaults to Emax;
- 'Emax': total number of derivatives of solution map to be
computed; defaults to E;
- 'robustTol': tolerance for robust Pade' denominator management;
defaults to 0;
- 'sampleType': label of sampling type; available values are:
- 'ARNOLDI': orthogonalization of solution derivatives through
Arnoldi algorithm;
- 'KRYLOV': standard computation of solution derivatives.
Defaults to 'KRYLOV'.
Defaults to empty dict.
homogeneized: Whether to homogeneize Dirichlet BCs. Defaults to False.
verbosity(optional): Verbosity level. Defaults to 10.
Attributes:
HFEngine: HF problem solver.
mu0: Default parameter.
homogeneized: Whether to homogeneize Dirichlet BCs.
approxParameters: Dictionary containing values for main parameters of
approximant. Recognized keys are in parameterList.
parameterList: Recognized keys of approximant parameters:
- 'POD': whether to compute POD of snapshots;
- 'rho': weight for computation of original Pade' approximant;
- 'M': degree of Pade' approximant numerator;
- 'N': degree of Pade' approximant denominator;
- 'E': total number of derivatives current approximant relies upon;
- 'Emax': total number of derivatives of solution map to be
computed;
- 'robustTol': tolerance for robust Pade' denominator management;
- 'sampleType': label of sampling type.
POD: Whether to compute QR factorization of derivatives.
rho: Weight of approximant.
M: Numerator degree of approximant.
N: Denominator degree of approximant.
E: Number of solution derivatives over which current approximant is
based upon.
Emax: Total number of solution derivatives to be computed.
robustTol: Tolerance for robust Pade' denominator management.
sampleType: Label of sampling type.
initialHFData: HF problem initial data.
uHF: High fidelity solution with wavenumber lastSolvedHF as numpy
complex vector.
lastSolvedHF: Wavenumber corresponding to last computed high fidelity
solution.
G: Square Numpy 2D vector of size (N+1) corresponding to Pade'
denominator matrix (see paper).
Q: Numpy 1D vector containing complex coefficients of approximant
denominator.
P: Numpy 2D vector whose columns are FE dofs of coefficients of
approximant numerator.
uApp: Last evaluated approximant as numpy complex vector.
lastApproxParameters: List of parameters corresponding to last
computed approximant.
"""
def __init__(self, HFEngine:HFEng, mu0 : complex = 0,
approxParameters : DictAny = {}, homogeneized : bool = False,
verbosity : int = 10):
self._preInit()
self._addParametersToList(["M", "N", "robustTol", "rho"])
super().__init__(HFEngine = HFEngine, mu0 = mu0,
approxParameters = approxParameters,
homogeneized = homogeneized,
verbosity = verbosity)
self._postInit()
@property
def approxParameters(self):
"""Value of approximant parameters."""
return self._approxParameters
@approxParameters.setter
def approxParameters(self, approxParams):
approxParameters = purgeDict(approxParams, self.parameterList,
dictname = self.name() + ".approxParameters",
baselevel = 1)
approxParametersCopy = purgeDict(approxParameters,
["M", "N", "robustTol", "rho"],
True, True, baselevel = 1)
keyList = list(approxParameters.keys())
if "rho" in keyList:
self._rho = approxParameters["rho"]
elif hasattr(self, "rho"):
self._rho = self.rho
else:
self._rho = np.inf
GenericApproximantTaylor.approxParameters.fset(self,
approxParametersCopy)
self.rho = self._rho
if "robustTol" in keyList:
self.robustTol = approxParameters["robustTol"]
elif hasattr(self, "robustTol"):
self.robustTol = self.robustTol
else:
self.robustTol = 0
self._ignoreParWarnings = True
if "M" in keyList:
self.M = approxParameters["M"]
elif hasattr(self, "M"):
self.M = self.M
else:
self.M = 0
del self._ignoreParWarnings
if "N" in keyList:
self.N = approxParameters["N"]
elif hasattr(self, "N"):
self.N = self.N
else:
self.N = 0
@property
def rho(self):
"""Value of rho."""
return self._rho
@rho.setter
def rho(self, rho):
self._rho = np.abs(rho)
self._approxParameters["rho"] = self.rho
@property
def M(self):
"""Value of M. Its assignment may change Emax and E."""
return self._M
@M.setter
def M(self, M):
if M < 0: raise ArithmeticError("M must be non-negative.")
self._M = M
self._approxParameters["M"] = self.M
if not hasattr(self, "_ignoreParWarnings"):
self.checkMNEEmax()
@property
def N(self):
"""Value of N. Its assignment may change Emax."""
return self._N
@N.setter
def N(self, N):
if N < 0: raise ArithmeticError("N must be non-negative.")
self._N = N
self._approxParameters["N"] = self.N
self.checkMNEEmax()
def checkMNEEmax(self):
"""Check consistency of M, N, E, and Emax."""
M = self.M if hasattr(self, "_M") else 0
N = self.N if hasattr(self, "_N") else 0
E = self.E if hasattr(self, "_E") else M + N
Emax = self.Emax if hasattr(self, "_Emax") else M + N
if self.rho == np.inf:
if Emax < max(N, M):
warn(("Prescribed Emax is too small. Updating Emax to "
"max(M, N)."))
self.Emax = max(N, M)
if E < max(N, M):
warn("Prescribed E is too small. Updating E to max(M, N).")
self.E = max(N, M)
else:
if Emax < N + M:
warn("Prescribed Emax is too small. Updating Emax to M + N.")
self.Emax = self.N + M
if E < N + M:
warn("Prescribed E is too small. Updating E to M + N.")
self.E = self.N + M
@property
def robustTol(self):
"""Value of tolerance for robust Pade' denominator management."""
return self._robustTol
@robustTol.setter
def robustTol(self, robustTol):
if robustTol < 0.:
warn("Overriding prescribed negative robustness tolerance to 0.")
robustTol = 0.
self._robustTol = robustTol
self._approxParameters["robustTol"] = self.robustTol
@property
def E(self):
"""Value of E. Its assignment may change Emax."""
return self._E
@E.setter
def E(self, E):
if E < 0: raise ArithmeticError("E must be non-negative.")
self._E = E
self.checkMNEEmax()
self._approxParameters["E"] = self.E
if hasattr(self, "Emax") and self.Emax < self.E:
warn("Prescribed Emax is too small. Updating Emax to E.")
self.Emax = self.E
@property
def Emax(self):
"""Value of Emax. Its assignment may reset computed derivatives."""
return self._Emax
@Emax.setter
def Emax(self, Emax):
if Emax < 0: raise ArithmeticError("Emax must be non-negative.")
if hasattr(self, "Emax"): EmaxOld = self.Emax
else: EmaxOld = -1
if hasattr(self, "_N"): N = self.N
else: N = 0
if hasattr(self, "_M"): M = self.M
else: M = 0
if hasattr(self, "_E"): E = self.E
else: E = 0
if self.rho == np.inf:
if max(N, M, E) > Emax:
warn(("Prescribed Emax is too small. Updating Emax to "
"max(N, M, E)."))
Emax = max(N, M, E)
else:
if max(N + M, E) > Emax:
warn(("Prescribed Emax is too small. Updating Emax to "
"max(N + M, E)."))
Emax = max(N + M, E)
self._Emax = Emax
self._approxParameters["Emax"] = Emax
if EmaxOld >= self.Emax and self.samplingEngine.samples is not None:
self.samplingEngine.samples = self.samplingEngine.samples[:,
: self.Emax + 1]
if (self.sampleType == "ARNOLDI"
and self.samplingEngine.HArnoldi is not None):
self.samplingEngine.HArnoldi = self.samplingEngine.HArnoldi[
: self.Emax + 1,
: self.Emax + 1]
self.samplingEngine.RArnoldi = self.samplingEngine.RArnoldi[
: self.Emax + 1,
: self.Emax + 1]
def setupApprox(self):
"""
Compute Pade' approximant. SVD-based robust eigenvalue management.
"""
if not self.checkComputedApprox():
if self.verbosity >= 5:
verbosityDepth("INIT", "Setting up {}.". format(self.name()))
self.computeDerivatives()
if self.N > 0:
if self.verbosity >= 7:
verbosityDepth("INIT", ("Starting computation of "
"denominator."))
while self.N > 0:
if self.POD:
ev, eV = self.findeveVGQR()
else:
ev, eV = self.findeveVGExplicit()
newParameters = checkRobustTolerance(ev, self.E,
self.robustTol)
if not newParameters:
break
self.approxParameters = newParameters
if self.N <= 0:
eV = np.ones((1, 1))
self.Q = np.poly1d(eV[:, 0])
if self.verbosity >= 7:
verbosityDepth("DEL", "Done computing denominator.")
else:
self.Q = np.poly1d([1])
if self.verbosity >= 7:
verbosityDepth("INIT", "Starting computation of numerator.")
self.P = np.zeros((self.Emax + 1, self.M + 1), dtype = np.complex)
for i in range(self.Q.order):
rng = np.arange(self.M + 1 - i)
self.P[rng, - 1 - rng - i] = self.Q[i]
if self.sampleType == "ARNOLDI":
self.P = self.samplingEngine.RArnoldi.dot(self.P)
if self.verbosity >= 7:
verbosityDepth("DEL", "Done computing numerator.")
self.lastApproxParameters = copy(self.approxParameters)
if hasattr(self, "lastSolvedApp"): del self.lastSolvedApp
if self.verbosity >= 5:
verbosityDepth("DEL", "Done setting up approximant.\n")
def rescaleParameter(self, R:Np2D, A : Np2D = None,
exponent : float = 1.) -> Np2D:
"""
Prepare parameter rescaling.
Args:
R: Matrix whose columns need rescaling.
A(optional): Matrix whose diagonal defines scaling factor. If None,
previous value of scaleFactor is used. Defaults to None.
exponent(optional): Exponent of scaling factor in matrix diagonal.
Defaults to 1.
Returns:
Rescaled matrix.
"""
if A is not None:
aDiag = np.diag(A)
scaleCoeffs = np.polyfit(np.arange(A.shape[1]),
np.log(aDiag), 1)
self.scaleFactor = np.exp(- scaleCoeffs[0] / exponent)
return np.multiply(R, np.power(self.scaleFactor,np.arange(R.shape[1])))
def buildG(self):
"""Assemble Pade' denominator matrix."""
self.computeDerivatives()
if self.rho == np.inf:
Nmin = self.E - self.N
else:
Nmin = self.M - self.N + 1
if self.sampleType == "KRYLOV":
DerE = self.samplingEngine.samples[:, Nmin : self.E + 1]
G = self.HFEngine.innerProduct(DerE, DerE)
DerE = self.rescaleParameter(DerE, G, 2.)
G = self.HFEngine.innerProduct(DerE, DerE)
else:
RArnE = self.samplingEngine.RArnoldi[: self.E + 1,
Nmin : self.E + 1]
RArnE = self.rescaleParameter(RArnE, RArnE[Nmin :, :])
G = RArnE.conj().T.dot(RArnE)
if self.rho == np.inf:
self.G = G
else:
Gbig = G
self.G = np.zeros((self.N + 1, self.N + 1), dtype = np.complex)
for k in range(self.E - self.M):
self.G += self.rho ** (2 * k) * Gbig[k : k + self.N + 1,
k : k + self.N + 1]
def findeveVGExplicit(self) -> Tuple[Np1D, Np2D]:
"""
Compute explicitly eigenvalues and eigenvectors of Pade' denominator
matrix.
"""
if self.verbosity >= 10:
verbosityDepth("INIT", "Building gramian matrix.")
self.buildG()
if self.verbosity >= 10:
verbosityDepth("DEL", "Done building gramian.")
if self.verbosity >= 7:
- verbosityDepth("INIT",
+ verbosityDepth("INIT",
"Solving eigenvalue problem for gramian matrix.")
ev, eV = np.linalg.eigh(self.G)
eV = self.rescaleParameter(eV.T).T
if self.verbosity >= 5:
try: condev = ev[-1] / ev[0]
except: condev = np.inf
verbosityDepth("MAIN", ("Solved eigenvalue problem of size {} "
"with condition number {:.4e}.").format(
self.N + 1, condev))
if self.verbosity >= 7:
verbosityDepth("DEL", "Done solving eigenvalue problem.")
return ev, eV
def findeveVGQR(self) -> Tuple[Np1D, Np2D]:
"""
Compute eigenvalues and eigenvectors of Pade' denominator matrix
through SVD of R factor. See ``Householder triangularization of a
quasimatrix'', L.Trefethen, 2008 for QR algorithm.
Returns:
Eigenvalues in ascending order and corresponding eigenvector
matrix.
"""
self.computeDerivatives()
if self.rho == np.inf:
Nmin = self.E - self.N
else:
Nmin = self.M - self.N + 1
if self.sampleType == "KRYLOV":
A = copy(self.samplingEngine.samples[:, Nmin : self.E + 1])
self.PODEngine = PODEngine(self.HFEngine)
if self.verbosity >= 10:
verbosityDepth("INIT", "Orthogonalizing samples.", end = "")
R = self.PODEngine.QRHouseholder(A, only_R = True)
if self.verbosity >= 10:
verbosityDepth("DEL", " Done.", inline = True)
else:
R = self.samplingEngine.RArnoldi[: self.E + 1, Nmin : self.E + 1]
R = self.rescaleParameter(R, R[Nmin :, :])
if self.rho == np.inf:
- if self.verbosity >= 10:
+ if self.verbosity >= 7:
verbosityDepth("INIT", ("Solving svd for square root of "
- "gramian matrix."), end = "")
+ "gramian matrix."))
sizeI = R.shape[0]
_, s, V = np.linalg.svd(R, full_matrices = False)
else:
if self.verbosity >= 10:
verbosityDepth("INIT", ("Building matrix stack for square "
"root of gramian."), end = "")
Rtower = np.zeros((R.shape[0] * (self.E - self.M), self.N + 1),
dtype = np.complex)
for k in range(self.E - self.M):
Rtower[k * R.shape[0] : (k + 1) * R.shape[0], :] = (
self.rho ** k
* R[:, self.M - self.N + 1 + k : self.M + 1 + k])
if self.verbosity >= 10:
verbosityDepth("DEL", " Done.", inline = True)
if self.verbosity >= 7:
verbosityDepth("INIT", ("Solving svd for square root of "
- "gramian matrix."), end = "")
+ "gramian matrix."))
sizeI = Rtower.shape[0]
_, s, V = np.linalg.svd(Rtower, full_matrices = False)
eV = V.conj().T[:, ::-1]
eV = self.rescaleParameter(eV.T).T
if self.verbosity >= 5:
try: condev = s[0] / s[-1]
except: condev = np.inf
verbosityDepth("MAIN", ("Solved svd problem of size {} x {} with "
"condition number {:.4e}.").format(
sizeI, self.N + 1, condev))
if self.verbosity >= 7:
- verbosityDepth("DEL", " Done.", inline = True)
+ verbosityDepth("DEL", "Done solving eigenvalue problem.")
return s[::-1], eV
def radiusPade(self, mu:Np1D, mu0 : float = None) -> float:
"""
Compute translated radius to be plugged into Pade' approximant.
Args:
mu: Parameter(s) 1.
mu0: Parameter(s) 2. If None, set to self.mu0.
Returns:
Translated radius to be plugged into Pade' approximant.
"""
if mu0 is None: mu0 = self.mu0
return self.HFEngine.rescaling(mu) - self.HFEngine.rescaling(mu0)
def getPVal(self, mu:List[complex]):
"""
Evaluate Pade' numerator at arbitrary parameter.
Args:
mu: Target parameter.
"""
self.setupApprox()
if self.verbosity >= 10:
verbosityDepth("INIT",
"Evaluating numerator at mu = {}.".format(mu),
end = "")
try:
len(mu)
except:
mu = [mu]
powerlist = np.vander(self.radiusPade(mu), self.M + 1).T
p = self.P.dot(powerlist)
if len(mu) == 1:
p = p.flatten()
if self.verbosity >= 10:
verbosityDepth("DEL", " Done.", inline = True)
return p
def getQVal(self, mu:List[complex]):
"""
Evaluate Pade' denominator at arbitrary parameter.
Args:
mu: Target parameter.
"""
self.setupApprox()
if self.verbosity >= 10:
verbosityDepth("INIT",
"Evaluating denominator at mu = {}.".format(mu),
end = "")
q = self.Q(self.radiusPade(mu))
if self.verbosity >= 10:
verbosityDepth("DEL", " Done.", inline = True)
return q
def evalApproxReduced(self, mu:complex):
"""
Evaluate Pade' approximant at arbitrary parameter.
Args:
mu: Target parameter.
"""
self.setupApprox()
if (not hasattr(self, "lastSolvedApp")
or not np.isclose(self.lastSolvedApp, mu)):
if self.verbosity >= 5:
verbosityDepth("INIT",
"Evaluating approximant at mu = {}.".format(mu))
self.uAppReduced = self.getPVal(mu) / self.getQVal(mu)
self.lastSolvedApp = mu
if self.verbosity >= 5:
verbosityDepth("DEL", "Done evaluating approximant.")
def evalApprox(self, mu:complex):
"""
Evaluate approximant at arbitrary parameter.
Args:
mu: Target parameter.
"""
self.evalApproxReduced(mu)
self.uApp = self.samplingEngine.samples.dot(self.uAppReduced)
def getPoles(self) -> Np1D:
"""
Obtain approximant poles.
Returns:
Numpy complex vector of poles.
"""
self.setupApprox()
return self.HFEngine.rescalingInv(self.Q.r
+ self.HFEngine.rescaling(self.mu0))