Page MenuHomec4science
Files F61750249

rational_interpolant.py
No OneTemporary
Actions

File Metadata

Created
Wed, May 8, 17:59

rational_interpolant.py
View Options

# 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 <http://www.gnu.org/licenses/>.
#
from copy import deepcopy as copy
import numpy as np
from scipy.linalg import eigvals
from collections.abc import Iterable
from rrompy.reduction_methods.base import checkRobustTolerance
from .generic_standard_approximant import GenericStandardApproximant
from rrompy.utilities.poly_fitting.polynomial import (
polybases as ppb, polyfitname,
polyvander as pvP, polyTimes,
polyTimesTable, vanderInvTable,
PolynomialInterpolator as PI,
PolynomialInterpolatorNodal as PIN)
from rrompy.utilities.poly_fitting.heaviside import rational2heaviside
from rrompy.utilities.poly_fitting.radial_basis import (polybases as rbpb,
RadialBasisInterpolator as RBI)
from rrompy.utilities.base.types import (Np1D, Np2D, Tuple, List, sampList,
interpEng)
from rrompy.utilities.base import verbosityManager as vbMng
from rrompy.utilities.numerical import pseudoInverse, dot, potential
from rrompy.utilities.numerical.hash_derivative import nextDerivativeIndices
from rrompy.utilities.numerical.degree import (reduceDegreeN,
degreeTotalToFull, fullDegreeMaxMask,
totalDegreeMaxMask)
from rrompy.solver import Np2DLikeGramian
from rrompy.utilities.exception_manager import (RROMPyException, RROMPyAssert,
RROMPyWarning)
__all__ = ['RationalInterpolant']
class RationalInterpolant(GenericStandardApproximant):
"""
ROM rational 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;
- 'scaleFactorDer': scaling factors for derivative computation;
defaults to 'AUTO';
- 'S': total number of samples current approximant relies upon;
- 'sampler': sample point generator;
- 'polybasis': type of polynomial basis for interpolation; defaults
to 'MONOMIAL';
- 'M': degree of rational interpolant numerator; defaults to
'AUTO', i.e. maximum allowed;
- 'N': degree of rational interpolant denominator; defaults to
'AUTO', i.e. maximum allowed;
- 'polydegreetype': type of polynomial degree; defaults to 'TOTAL';
- 'radialDirectionalWeights': radial basis weights for interpolant
numerator; defaults to 1;
- 'radialDirectionalWeightsAdapt': bounds for adaptive rescaling of
radial basis weights; defaults to [-1, -1];
- 'functionalSolve': strategy for minimization of denominator
functional; allowed values include 'NORM', 'DOMINANT', 'NODAL',
'LOEWNER', and 'BARYCENTRIC'; defaults to 'NORM';
- 'interpRcond': tolerance for interpolation; defaults to None;
- 'robustTol': tolerance for robust rational denominator
management; defaults to 0.
Defaults to empty dict.
approx_state(optional): Whether to approximate state. Defaults to
False.
verbosity(optional): Verbosity level. Defaults to 10.
Attributes:
HFEngine: HF problem solver.
mu0: Default parameter.
mus: Array of snapshot parameters.
approxParameters: Dictionary containing values for main parameters of
approximant. Recognized keys are in parameterList.
parameterListSoft: Recognized keys of soft approximant parameters:
- 'POD': whether to compute POD of snapshots;
- 'scaleFactorDer': scaling factors for derivative computation;
- 'polybasis': type of polynomial basis for interpolation;
- 'M': degree of rational interpolant numerator;
- 'N': degree of rational interpolant denominator;
- 'polydegreetype': type of polynomial degree;
- 'radialDirectionalWeights': radial basis weights for interpolant
numerator;
- 'radialDirectionalWeightsAdapt': bounds for adaptive rescaling of
radial basis weights;
- 'functionalSolve': strategy for minimization of denominator
functional;
- 'interpRcond': tolerance for interpolation via numpy.polyfit;
- 'robustTol': tolerance for robust rational denominator
management.
parameterListCritical: Recognized keys of critical approximant
parameters:
- 'S': total number of samples current approximant relies upon;
- 'sampler': sample point generator.
approx_state: Whether to approximate state.
verbosity: Verbosity level.
POD: Whether to compute POD of snapshots.
scaleFactorDer: Scaling factors for derivative computation.
S: Number of solution snapshots over which current approximant is
based upon.
sampler: Sample point generator.
polybasis: type of polynomial basis for interpolation.
M: Numerator degree of approximant.
N: Denominator degree of approximant.
polydegreetype: Type of polynomial degree.
radialDirectionalWeights: Radial basis weights for interpolant
numerator.
radialDirectionalWeightsAdapt: Bounds for adaptive rescaling of radial
basis weights.
functionalSolve: Strategy for minimization of denominator functional.
interpRcond: Tolerance for interpolation via numpy.polyfit.
robustTol: Tolerance for robust rational denominator management.
muBounds: list of bounds for parameter values.
samplingEngine: Sampling engine.
uHF: High fidelity solution(s) with parameter(s) lastSolvedHF as
sampleList.
lastSolvedHF: Parameter(s) corresponding to last computed high fidelity
solution(s) as parameterList.
uApproxReduced: Reduced approximate solution(s) with parameter(s)
lastSolvedApprox as sampleList.
lastSolvedApproxReduced: Parameter(s) corresponding to last computed
reduced approximate solution(s) as parameterList.
uApprox: Approximate solution(s) with parameter(s) lastSolvedApprox as
sampleList.
lastSolvedApprox: Parameter(s) corresponding to last computed
approximate solution(s) as parameterList.
Q: Numpy 1D vector containing complex coefficients of approximant
denominator.
P: Numpy 2D vector whose columns are FE dofs of coefficients of
approximant numerator.
"""
def __init__(self, *args, **kwargs):
self._preInit()
self._addParametersToList(["polybasis", "M", "N", "polydegreetype",
"radialDirectionalWeights",
"radialDirectionalWeightsAdapt",
"functionalSolve", "interpRcond",
"robustTol"],
["MONOMIAL", "AUTO", "AUTO", "TOTAL", 1.,
[-1., -1.], "NORM", -1, 0.])
super().__init__(*args, **kwargs)
self.catchInstability = 0
self._postInit()
@property
def tModelType(self):
from .trained_model.trained_model_rational import TrainedModelRational
return TrainedModelRational
@property
def polybasis(self):
"""Value of polybasis."""
return self._polybasis
@polybasis.setter
def polybasis(self, polybasis):
try:
polybasis = polybasis.upper().strip().replace(" ","")
if polybasis not in ppb + rbpb:
raise RROMPyException("Prescribed polybasis not recognized.")
self._polybasis = polybasis
except:
RROMPyWarning(("Prescribed polybasis not recognized. Overriding "
"to 'MONOMIAL'."))
self._polybasis = "MONOMIAL"
self._approxParameters["polybasis"] = self.polybasis
@property
def polybasis0(self):
if "_" in self.polybasis:
return self.polybasis.split("_")[0]
return self.polybasis
@property
def functionalSolve(self):
"""Value of functionalSolve."""
return self._functionalSolve
@functionalSolve.setter
def functionalSolve(self, functionalSolve):
try:
functionalSolve = functionalSolve.upper().strip().replace(" ","")
if functionalSolve not in ["NORM", "DOMINANT", "NODAL", "LOEWNER",
"BARYCENTRIC"]:
raise RROMPyException(("Prescribed functionalSolve not "
"recognized."))
self._functionalSolve = functionalSolve
except:
RROMPyWarning(("Prescribed functionalSolve not recognized. "
"Overriding to 'NORM'."))
self._functionalSolve = "NORM"
self._approxParameters["functionalSolve"] = self.functionalSolve
@property
def interpRcond(self):
"""Value of interpRcond."""
return self._interpRcond
@interpRcond.setter
def interpRcond(self, interpRcond):
self._interpRcond = interpRcond
self._approxParameters["interpRcond"] = self.interpRcond
@property
def radialDirectionalWeights(self):
"""Value of radialDirectionalWeights."""
return self._radialDirectionalWeights
@radialDirectionalWeights.setter
def radialDirectionalWeights(self, radialDirectionalWeights):
if isinstance(radialDirectionalWeights, Iterable):
radialDirectionalWeights = list(radialDirectionalWeights)
else:
radialDirectionalWeights = [radialDirectionalWeights]
self._radialDirectionalWeights = radialDirectionalWeights
self._approxParameters["radialDirectionalWeights"] = (
self.radialDirectionalWeights)
@property
def radialDirectionalWeightsAdapt(self):
"""Value of radialDirectionalWeightsAdapt."""
return self._radialDirectionalWeightsAdapt
@radialDirectionalWeightsAdapt.setter
def radialDirectionalWeightsAdapt(self, radialDirectionalWeightsAdapt):
self._radialDirectionalWeightsAdapt = radialDirectionalWeightsAdapt
self._approxParameters["radialDirectionalWeightsAdapt"] = (
self.radialDirectionalWeightsAdapt)
@property
def M(self):
"""Value of M."""
return self._M
@M.setter
def M(self, M):
if isinstance(M, str):
M = M.strip().replace(" ","")
if "-" not in M: M = M + "-0"
self._M_isauto, self._M_shift = True, int(M.split("-")[-1])
M = 0
if M < 0: raise RROMPyException("M must be non-negative.")
self._M = M
self._approxParameters["M"] = self.M
def _setMAuto(self):
self.M = max(0, reduceDegreeN(self.S, self.S, self.npar,
self.polydegreetype) - self._M_shift)
vbMng(self, "MAIN", "Automatically setting M to {}.".format(self.M),
25)
@property
def N(self):
"""Value of N."""
return self._N
@N.setter
def N(self, N):
if isinstance(N, str):
N = N.strip().replace(" ","")
if "-" not in N: N = N + "-0"
self._N_isauto, self._N_shift = True, int(N.split("-")[-1])
N = 0
if N < 0: raise RROMPyException("N must be non-negative.")
self._N = N
self._approxParameters["N"] = self.N
def _setNAuto(self):
self.N = max(0, reduceDegreeN(self.S, self.S, self.npar,
self.polydegreetype) - self._N_shift)
vbMng(self, "MAIN", "Automatically setting N to {}.".format(self.N),
25)
@property
def polydegreetype(self):
"""Value of polydegreetype."""
return self._polydegreetype
@polydegreetype.setter
def polydegreetype(self, polydegreetype):
try:
polydegreetype = polydegreetype.upper().strip().replace(" ","")
if polydegreetype not in ["TOTAL", "FULL"]:
raise RROMPyException(("Prescribed polydegreetype not "
"recognized."))
self._polydegreetype = polydegreetype
except:
RROMPyWarning(("Prescribed polydegreetype not recognized. "
"Overriding to 'TOTAL'."))
self._polydegreetype = "TOTAL"
self._approxParameters["polydegreetype"] = self.polydegreetype
@property
def robustTol(self):
"""Value of tolerance for robust rational denominator management."""
return self._robustTol
@robustTol.setter
def robustTol(self, robustTol):
if robustTol < 0.:
RROMPyWarning(("Overriding prescribed negative robustness "
"tolerance to 0."))
robustTol = 0.
self._robustTol = robustTol
self._approxParameters["robustTol"] = self.robustTol
def resetSamples(self):
"""Reset samples."""
super().resetSamples()
self._musUniqueCN = None
self._derIdxs = None
self._reorder = None
def _setupInterpolationIndices(self):
"""Setup parameters for polyvander."""
if self._musUniqueCN is None or len(self._reorder) != len(self.mus):
self._musUniqueCN, musIdxsTo, musIdxs, musCount = (
self.trainedModel.centerNormalize(self.mus).unique(
return_index = True, return_inverse = True,
return_counts = True))
self._musUnique = self.mus[musIdxsTo]
self._derIdxs = [None] * len(self._musUniqueCN)
self._reorder = np.empty(len(musIdxs), dtype = int)
filled = 0
for j, cnt in enumerate(musCount):
self._derIdxs[j] = nextDerivativeIndices([], self.mus.shape[1],
cnt)
jIdx = np.nonzero(musIdxs == j)[0]
self._reorder[jIdx] = np.arange(filled, filled + cnt)
filled += cnt
def _setupDenominator(self):
"""Compute rational denominator."""
RROMPyAssert(self._mode, message = "Cannot setup denominator.")
vbMng(self, "INIT", "Starting computation of denominator.", 7)
if hasattr(self, "_N_isauto"):
self._setNAuto()
else:
N = reduceDegreeN(self.N, self.S, self.npar, self.polydegreetype)
if N < self.N:
RROMPyWarning(("N too large compared to S. Reducing N by "
"{}").format(self.N - N))
self.N = N
while self.N > 0:
if self.functionalSolve != "NORM" and self.npar > 1:
RROMPyWarning(("Strategy for functional optimization must be "
"'NORM' for more than one parameter. "
"Overriding to 'NORM'."))
self.functionalSolve = "NORM"
if self.functionalSolve == "BARYCENTRIC" and self.N + 1 < self.S:
RROMPyWarning(("Barycentric strategy cannot be applied with "
"Least Squares. Overriding to 'NORM'."))
self.functionalSolve = "NORM"
if self.functionalSolve == "BARYCENTRIC":
invD, TN = None, None
self._setupInterpolationIndices()
else:
invD, TN = self._computeInterpolantInverseBlocks()
if (self.functionalSolve in ["NODAL", "LOEWNER", "BARYCENTRIC"]
and len(self._musUnique) != len(self.mus)):
if self.functionalSolve == "BARYCENTRIC":
warnflag = "Barycentric"
else:
warnflag = "Iterative"
RROMPyWarning(("{} functional optimization cannot be applied "
"to repeated samples. Overriding to "
"'NORM'.").format(warnflag))
self.functionalSolve = "NORM"
idxSamplesEff = list(range(self.S))
if self.POD:
ev, eV = self.findeveVGQR(
self.samplingEngine.RPOD[:, idxSamplesEff], invD, TN)
else:
ev, eV = self.findeveVGExplicit(
self.samplingEngine.samples(idxSamplesEff), invD, TN)
if self.functionalSolve in ["NODAL", "LOEWNER"]: break
nevBad = checkRobustTolerance(ev, self.robustTol)
if nevBad <= (self.functionalSolve == "NORM"): break
if self.catchInstability:
raise RROMPyException(("Instability in denominator "
"computation: eigenproblem is poorly "
"conditioned."),
self.catchInstability == 1)
vbMng(self, "MAIN", ("Smallest {} eigenvalues below tolerance. "
"Reducing N by 1.").format(nevBad), 10)
self.N = self.N - 1
if self.N <= 0:
self.N = 0
eV = np.ones((1, 1))
if self.N > 0 and self.functionalSolve in ["NODAL", "LOEWNER",
"BARYCENTRIC"]:
q = PIN()
q.polybasis, q.nodes = self.polybasis0, eV
else:
q = PI()
q.npar = self.npar
q.polybasis = self.polybasis0
if self.polydegreetype == "TOTAL":
q.coeffs = degreeTotalToFull(tuple([self.N + 1] * self.npar),
self.npar, eV)
else:
q.coeffs = eV.reshape([self.N + 1] * self.npar)
vbMng(self, "DEL", "Done computing denominator.", 7)
return q
def _setupNumerator(self):
"""Compute rational numerator."""
RROMPyAssert(self._mode, message = "Cannot setup numerator.")
vbMng(self, "INIT", "Starting computation of numerator.", 7)
self._setupInterpolationIndices()
Qevaldiag = polyTimesTable(self.trainedModel.data.Q, self._musUniqueCN,
self._reorder, self._derIdxs,
self.scaleFactorRel)
if self.POD: Qevaldiag = Qevaldiag.dot(self.samplingEngine.RPOD.T)
if hasattr(self, "_M_isauto"):
self._setMAuto()
M = self.M
else:
M = reduceDegreeN(self.M, self.S, self.npar, self.polydegreetype)
if M < self.M:
RROMPyWarning(("M too large compared to S. Reducing M by "
"{}").format(self.M - M))
self.M = M
while self.M >= 0:
pParRest = [self.M, self.polybasis, self.verbosity >= 5,
self.polydegreetype == "TOTAL",
{"derIdxs": self._derIdxs, "reorder": self._reorder,
"scl": self.scaleFactorRel}]
if self.polybasis in ppb:
p = PI()
else:
self.computeScaleFactor()
rDWEff = np.array([w * f for w, f in zip(
self.radialDirectionalWeights,
self.scaleFactor)])
pParRest = pParRest[: 2] + [rDWEff] + pParRest[2 :]
pParRest[-1]["optimizeScalingBounds"] = (
self.radialDirectionalWeightsAdapt)
p = RBI()
if self.polybasis in ppb + rbpb:
pParRest += [{"rcond": self.interpRcond}]
wellCond, msg = p.setupByInterpolation(self._musUniqueCN,
Qevaldiag, *pParRest)
vbMng(self, "MAIN", msg, 5)
if wellCond: break
if self.catchInstability:
raise RROMPyException(("Instability in numerator computation: "
"polyfit is poorly conditioned."),
self.catchInstability == 1)
vbMng(self, "MAIN", ("Polyfit is poorly conditioned. Reducing M "
"by 1."), 10)
self.M = self.M - 1
if self.M < 0:
raise RROMPyException(("Instability in computation of numerator. "
"Aborting."))
self.M = M
vbMng(self, "DEL", "Done computing numerator.", 7)
return p
def setupApprox(self) -> int:
"""Compute rational interpolant."""
if self.checkComputedApprox(): return -1
RROMPyAssert(self._mode, message = "Cannot setup approximant.")
vbMng(self, "INIT", "Setting up {}.". format(self.name()), 5)
self.computeSnapshots()
self._setupTrainedModel(self.samplingEngine.projectionMatrix)
self._setupRational(self._setupDenominator())
self.trainedModel.data.approxParameters = copy(self.approxParameters)
vbMng(self, "DEL", "Done setting up approximant.", 5)
return 0
def _setupRational(self, Q:interpEng, P : interpEng = None):
vbMng(self, "INIT", "Starting approximant finalization.", 5)
self.trainedModel.data.Q = Q
if P is None: P = self._setupNumerator()
if self.N > 0 and self.npar == 1:
pls = Q.roots()
idxBad = self.HFEngine.flagBadPolesResidues(pls, relative = True)
plsN = self.mapParameterList(self.mapParameterList(self.mu0)(0, 0)
+ self.scaleFactor * pls, "B")(0)
idxBad = np.logical_or(self.HFEngine.flagBadPolesResidues(pls,
relative = True),
self.HFEngine.flagBadPolesResidues(plsN))
if np.any(idxBad):
vbMng(self, "MAIN",
"Removing {} spurious poles out of {} due to poles."\
.format(np.sum(idxBad), self.N), 10)
if isinstance(Q, PIN):
Q.nodes = Q.nodes[np.logical_not(idxBad)]
else:
Q = PI()
Q.npar = self.npar
Q.polybasis = self.polybasis0
Q.coeffs = np.ones(1, dtype = np.complex)
for pl in pls[np.logical_not(idxBad)]:
Q.coeffs = polyTimes(Q.coeffs, [- pl, 1.],
Pbasis = Q.polybasis, Rbasis = Q.polybasis)
Q.coeffs /= np.linalg.norm(Q.coeffs)
self.trainedModel.data.Q = Q
self.N = Q.deg[0]
P = self._setupNumerator()
if (not hasattr(self.HFEngine, "_ignoreResidues")
or not self.HFEngine._ignoreResidues):
cfs, pls, _ = rational2heaviside(P, Q)
cfs = cfs[: self.N].T
if not self.POD:
cfs = self.samplingEngine.projectionMatrix.dot(cfs)
foci = self.sampler.normalFoci()
ground = self.sampler.groundPotential()
potEff = potential(pls, foci) / ground
potEff[np.logical_or(potEff < 1., np.isinf(pls))] = 1.
cfs[:, np.isinf(pls)] = 0.
cfs /= potEff # rescale by potential
idxBad = self.HFEngine.flagBadPolesResidues(pls, cfs)
if np.any(idxBad):
vbMng(self, "MAIN",
("Removing {} spurious poles out of {} due to "
"residues.").format(np.sum(idxBad), self.N), 10)
if isinstance(Q, PIN):
Q.nodes = Q.nodes[np.logical_not(idxBad)]
else:
Q = PI()
Q.npar = self.npar
Q.polybasis = self.polybasis0
Q.coeffs = np.ones(1, dtype = np.complex)
for pl in pls[np.logical_not(idxBad)]:
Q.coeffs = polyTimes(Q.coeffs, [- pl, 1.],
Pbasis = Q.polybasis, Rbasis = Q.polybasis)
Q.coeffs /= np.linalg.norm(Q.coeffs)
self.trainedModel.data.Q = Q
self.N = Q.deg[0]
P = self._setupNumerator()
self.trainedModel.data.P = P
vbMng(self, "DEL", "Terminated approximant finalization.", 5)
def _computeInterpolantInverseBlocks(self) -> Tuple[List[Np2D], Np2D]:
"""
Compute inverse factors for minimal interpolant target functional.
"""
RROMPyAssert(self._mode, message = "Cannot solve eigenvalue problem.")
self._setupInterpolationIndices()
pvPPar = [self.polybasis0, self._derIdxs, self._reorder,
self.scaleFactorRel]
if hasattr(self, "_M_isauto"): self._setMAuto()
E = max(self.M, self.N)
while E >= 0:
if self.polydegreetype == "TOTAL":
Eeff = E
idxsB = totalDegreeMaxMask(E, self.npar)
else: #if self.polydegreetype == "FULL":
Eeff = [E] * self.npar
idxsB = fullDegreeMaxMask(E, self.npar)
TE = pvP(self._musUniqueCN, Eeff, *pvPPar)
fitOut = pseudoInverse(TE, rcond = self.interpRcond, full = True)
vbMng(self, "MAIN",
("Fitting {} samples with degree {} through {}... "
"Conditioning of pseudoinverse system: {:.4e}.").format(
TE.shape[0], E,
polyfitname(self.polybasis0),
fitOut[1][1][0] / fitOut[1][1][-1]),
5)
if fitOut[1][0] == TE.shape[1]:
fitinv = fitOut[0][idxsB, :]
break
if self.catchInstability:
raise RROMPyException(("Instability in denominator "
"computation: polyfit is poorly "
"conditioned."),
self.catchInstability == 1)
EeqN = E == self.N
vbMng(self, "MAIN", ("Polyfit is poorly conditioned. Reducing E {}"
"by 1.").format("and N " * EeqN), 10)
if EeqN: self.N = self.N - 1
E -= 1
if self.N < 0:
raise RROMPyException(("Instability in computation of "
"denominator. Aborting."))
invD = vanderInvTable(fitinv, idxsB, self._reorder, self._derIdxs)
if self.N == E:
TN = TE
else:
if self.polydegreetype == "TOTAL":
Neff = self.N
idxsB = totalDegreeMaxMask(self.N, self.npar)
else: #if self.polydegreetype == "FULL":
Neff = [self.N] * self.npar
idxsB = fullDegreeMaxMask(self.N, self.npar)
TN = pvP(self._musUniqueCN, Neff, *pvPPar)
return invD, TN
def findeveVGExplicit(self, sampleE:sampList,
invD:List[Np2D], TN:Np2D) -> Tuple[Np1D, Np2D]:
"""
Compute explicitly eigenvalues and eigenvectors of rational denominator
matrix.
"""
RROMPyAssert(self._mode, message = "Cannot solve eigenvalue problem.")
vbMng(self, "INIT", "Building gramian matrix.", 10)
gramian = self.HFEngine.innerProduct(sampleE, sampleE,
is_state = self.approx_state)
if self.functionalSolve == "NODAL":
SEnd = invD[0].shape[1]
G0 = np.zeros((SEnd,) * 2, dtype = np.complex)
elif self.functionalSolve == "LOEWNER":
G0 = gramian
if self.functionalSolve == "BARYCENTRIC":
nEnd = len(gramian) - 1
else:
nEnd = TN.shape[1]
G = np.zeros((nEnd, nEnd), dtype = np.complex)
for k in range(len(invD)):
iDkN = dot(invD[k], TN)
G += dot(dot(gramian, iDkN).T, iDkN.conj()).T
if self.functionalSolve == "NODAL":
G0 += dot(dot(gramian, invD[k]).T, invD[k].conj()).T
vbMng(self, "DEL", "Done building gramian.", 10)
if self.functionalSolve == "NORM":
ev, eV = np.linalg.eigh(G)
eV = eV[:, 0]
problem = "eigenproblem"
else:
if self.functionalSolve == "BARYCENTRIC":
fitOut = pseudoInverse(gramian, rcond = self.interpRcond,
full = True)
barWeigths = fitOut[0].dot(np.ones(nEnd + 1))
eV = self.findeVBarycentric(barWeigths / np.sum(barWeigths))
else:
fitOut = pseudoInverse(G[:-1, :-1], rcond = self.interpRcond,
full = True)
eV = np.append(fitOut[0].dot(G[:-1, -1]), -1.)
ev = fitOut[1][1][::-1]
problem = "linear problem"
vbMng(self, "MAIN",
("Solved {} of size {} with condition number "
"{:.4e}.").format(problem, nEnd, ev[-1] / ev[0]), 5)
if self.functionalSolve in ["NODAL", "LOEWNER"]:
q = PI()
q.npar, q.polybasis, q.coeffs = self.npar, self.polybasis0, eV
eV, tol, niter, passed = self.findeVNewton(q.roots(), G0)
if passed:
vbMng(self, "MAIN",
("Newton algorithm for problem of size {} converged "
"(tol = {:.4e}) in {} iterations.").format(nEnd, tol,
niter), 5)
else:
RROMPyWarning(("Newton algorithm for problem of size {} did "
"not converge (tol = {:.4e}) after {} "
"iterations.").format(nEnd, tol, niter))
return ev, eV
def findeveVGQR(self, RPODE:Np2D, invD:List[Np2D],
TN:Np2D) -> Tuple[Np1D, Np2D]:
"""
Compute eigenvalues and eigenvectors of rational denominator matrix
through SVD of R factor.
"""
RROMPyAssert(self._mode, message = "Cannot solve eigenvalue problem.")
vbMng(self, "INIT", "Building half-gramian matrix stack.", 10)
if self.functionalSolve == "NODAL":
gramian = Np2DLikeGramian(None, dot(RPODE, invD[0]))
elif self.functionalSolve == "LOEWNER":
gramian = Np2DLikeGramian(None, RPODE)
if self.functionalSolve == "BARYCENTRIC":
nEnd = RPODE.shape[1] - 1
else:
S, nEnd, eWidth = RPODE.shape[0], TN.shape[1], len(invD)
Rstack = np.zeros((S * eWidth, nEnd), dtype = np.complex)
for k in range(eWidth):
Rstack[k * S : (k + 1) * S, :] = dot(RPODE, dot(invD[k], TN))
vbMng(self, "DEL", "Done building half-gramian.", 10)
if self.functionalSolve in ["NORM", "BARYCENTRIC"]:
if self.functionalSolve == "NORM":
_, s, Vh = np.linalg.svd(Rstack, full_matrices = False)
eV = Vh[-1, :].conj()
else: #if self.functionalSolve == "BARYCENTRIC":
_, s, Vh = np.linalg.svd(RPODE, full_matrices = False)
s[np.logical_not(np.isclose(s, 0.))] **= -2.
barWeigths = (Vh.T.conj() * s).dot(Vh.dot(np.ones(nEnd + 1)))
eV = self.findeVBarycentric(barWeigths / np.sum(barWeigths))
ev = s[::-1]
problem = "svd problem"
else:
fitOut = pseudoInverse(Rstack[:, :-1], rcond = self.interpRcond,
full = True)
ev = fitOut[1][1][::-1]
eV = np.append(fitOut[0].dot(Rstack[:, -1]), -1.)
problem = "linear problem"
vbMng(self, "MAIN",
("Solved {} of size {} with condition number "
"{:.4e}.").format(problem, nEnd, ev[-1] / ev[0]), 5)
if self.functionalSolve in ["NODAL", "LOEWNER"]:
q = PI()
q.npar, q.polybasis, q.coeffs = self.npar, self.polybasis0, eV
eV, tol, niter, passed = self.findeVNewton(q.roots(), gramian)
if passed:
vbMng(self, "MAIN",
("Newton algorithm for problem of size {} converged "
"(tol = {:.4e}) in {} iterations.").format(nEnd, tol,
niter), 5)
else:
RROMPyWarning(("Newton algorithm for problem of size {} did "
"not converge (tol = {:.2e}) after {} "
"iterations.").format(nEnd, tol, niter))
return ev, eV
def findeVBarycentric(self, baryWeights:Np1D) -> Np1D:
RROMPyAssert(self._mode,
message = "Cannot solve optimization problem.")
arrow = np.pad(np.diag(self._musUniqueCN.data[
self._reorder].flatten()),
(1, 0), "constant", constant_values = 1.) + 0.j
arrow[0, 0] = 0.
arrow[0, 1:] = baryWeights
active = np.pad(np.eye(len(baryWeights)), (1, 0), "constant")
eV = eigvals(arrow, active)
return eV[np.logical_not(np.isinf(eV))]
def findeVNewton(self, nodes0:Np1D, gram:Np2D, maxiter : int = 25,
tolNewton : float = 1e-10) \
-> Tuple[Np1D, float, int, bool]:
RROMPyAssert(self._mode,
message = "Cannot solve optimization problem.")
algo = self.functionalSolve
N = len(nodes0)
nodes = nodes0
grad = np.zeros(N, dtype = np.complex)
hess = np.zeros((N, N), dtype = np.complex)
mu = np.repeat(self._musUniqueCN.data[self._reorder], N, axis = 1)
for niter in range(maxiter):
if algo == "NODAL":
plDist = mu - nodes.reshape(1, -1)
q0, q = np.prod(plDist, axis = 1), []
elif algo == "LOEWNER":
loew = np.pad(np.power(mu - nodes.reshape(1, -1), -1.),
[(0, 0), (1, 0)], 'constant',
constant_values = 1.)
loewI = pseudoInverse(loew)
Ids = []
for jS in range(N):
if algo == "NODAL":
mask = np.arange(N) != jS
q += [np.prod(plDist[:, mask], axis = 1)]
grad[jS] = q[-1].conj().dot(gram.dot(q0))
elif algo == "LOEWNER":
Ids += [loewI.dot(np.power(loew[:, 1 + jS], 2.))]
zIj, jI = Ids[-1][0], loewI[1 + jS]
grad[jS] = (zIj * jI).conj().dot(gram.dot(loewI[0]))
for iS in range(jS + 1):
if algo == "NODAL":
if iS == jS:
hij = 0.
sij = q[-1].conj().dot(gram.dot(q[-1]))
else:
mask = np.logical_and(np.arange(N) != iS,
np.arange(N) != jS)
qij = np.prod(plDist[:, mask], axis = 1)
hij = qij.conj().dot(gram.dot(q0))
sij = q[-1].conj().dot(gram.dot(q[iS]))
elif algo == "LOEWNER":
zIi, iIj = Ids[iS][0], Ids[-1][1 + iS]
hij = (zIi * iIj * jI).conj().dot(gram.dot(loewI[0]))
if iS == jS:
iI = jI
zIdd = loewI[0].dot(np.power(loew[:, 1 + jS], 3.))
hij += (zIdd * jI).conj().dot(gram.dot(loewI[0]))
hij *= 2.
else:
jIi, iI = Ids[iS][1 + jS], loewI[1 + iS]
hij += (zIj * jIi * iI).conj().dot(
gram.dot(loewI[0]))
sij = (zIj * jI).conj().dot(gram.dot(zIi * iI))
hess[jS, iS] = hij + sij
if iS < jS: hess[iS, jS] = hij + sij.conj()
dnodes = np.linalg.solve(hess, grad)
nodes += dnodes
tol = np.linalg.norm(dnodes) / np.linalg.norm(nodes)
if tol < tolNewton: break
return nodes, tol, niter, niter < maxiter
def getResidues(self, *args, **kwargs) -> Np2D:
"""
Obtain approximant residues.
Returns:
Matrix with residues as columns.
"""
return self.trainedModel.getResidues(*args, **kwargs)

Event Timeline

c4science · Help