"""ERT modelling operator classes.
* abstract base class providing a plotting function
* standard BERT modelling class using pygimli.core (C++) functionality
* 2.5D non-optimized totalfield forward operator for ERT (reference)
"""
import numpy as np
import pygimli as pg
from pygimli.frameworks import MeshModelling
from pygimli import pf
from .visualization import showERTData
class ERTModellingBase(MeshModelling):
"""Modelling base class for ERT modelling."""
def __init__(self, **kwargs): # actually useless def
super().__init__(**kwargs)
def drawData(self, ax, data=None, **kwargs):
"""Draw data on given axes.
Args
----
ax: MPL Axe
Axes instance to draw into.
data: iterable | :py:func:pygimli.pyhsics.ert.DataContainer`
Datavalues to draw
* data is array, taken as values and draw on self.data structure
* data is Datacontainer, draw data['rhoa'] on data structure
* data is None, draw self.data['rhoa'] on self.data structure
Keyword Args
------------
vals: iterable
Draw vals on self.data structure if no data are given.
Remaining kwargs are forwarded to
:py:func:pygimli.pyhsics.ert.showERTData`
"""
kwargs['label'] = kwargs.pop('label', pg.unit('res'))
kwargs['cMap'] = kwargs.pop('cMap', pg.utils.cMap('res'))
vals = None
if hasattr(data, '__iter__'):
# show data values from array
vals = data
data = self.data
elif isinstance(data, pg.DataContainerERT):
# data is given DataContainer
pass
else:
# data is self.DataContainer
data = self.data
if vals is None:
vals = kwargs.pop('vals', data['rhoa'])
return showERTData(data, vals=vals, ax=ax, **kwargs)
def drawModel(self, ax, model, **kwargs):
"""Draw the para domain with option model values."""
kwargs.setdefault('label', pg.unit('res'))
kwargs.setdefault('cMap', pg.utils.cMap('res'))
kwargs.setdefault('logScale', True)
return super().drawModel(ax=ax, model=model, **kwargs)
[docs]
class ERTModelling(ERTModellingBase):
"""Forward operator for Electrical Resistivity Tomography.
Note
----
Convention for complex resistiviy inversion:
We want to use logarithm transformation for the imaginary part of model
so we need the startmodel to have positive imaginary parts.
The sign is flipped back to physical correct assumption before we call
the response function.
The Jacobian is calculated with negative imaginary parts and will
be a conjugated complex block matrix for further calulations.
"""
[docs]
def __init__(self, sr=True, verbose=False):
super().__init__()
# don't use DC*fop or its regionmanager directly
#
self._core = None
if sr:
self._core = pg.core.DCSRMultiElectrodeModelling(verbose=verbose)
else:
self._core = pg.core.DCMultiElectrodeModelling(verbose=verbose)
# Its good that the core knows about the actual RM
self._core.setRegionManager(self.regionManagerRef())
self._core.initJacobian()
self.setJacobian(self._core.jacobian())
# called from the ERTManager .. needed?
self.solution = self._core.solution
self.setComplex = self._core.setComplex
self.complex = self._core.complex
self.calculate = self._core.calculate
self.calcGeometricFactor = self._core.calcGeometricFactor
self.mapERTModel = self._core.mapERTModel
self._conjImag = False # the imaginary parts are flipped for log trans
[docs]
def setVerbose(self, v):
"""Set verbosity."""
super().setVerbose(v)
self._core.setVerbose(v)
[docs]
def setDefaultBackground(self):
"""Set the default background behaviour."""
if self.complex():
self.regionManager().addRegion(3, self._baseMesh, 2)
regionIds = self.regionManager().regionIdxs()
pg.info("Found {} regions.".format(len(regionIds)))
if len(regionIds) > 1:
bk = pg.sort(regionIds)[0]
pg.info(f"(ERTModelling) Region with smallest marker ({bk}) "
"set to background.")
self.setRegionProperties(bk, background=True)
[docs]
def createStartModel(self, dataVals):
"""Create Starting model for ERT inversion."""
if self.complex():
dataC = pg.utils.toComplex(dataVals)
nModel = self.parameterCount // 2
smRe = np.ones(nModel) * np.median(np.median(dataC.real))
smIm = np.ones(nModel) * np.median(np.median(dataC.imag))
if min(smIm) < 0:
# we want positive phase model
sm = smRe - 1j * smIm
pg.info("Model imaginary part being flipped to positive.")
self._conjImag = True
else:
sm = smRe + 1j * smIm
return pg.utils.squeezeComplex(sm) # complex impedance
else:
return super().createStartModel(dataVals)
[docs]
def flipImagPart(self, v):
"""Flip imaginary port (convention)."""
z = pg.utils.toComplex(v)
pg.warn('pre min/max={0} / {1} im: {2} / {3}'.format(
pf(min(z.real)), pf(max(z.real)),
pf(min(z.imag)), pf(max(z.imag))))
v = pg.utils.squeezeComplex(pg.utils.toComplex(v),
conj=self._conjImag)
z = pg.utils.toComplex(v)
pg.warn('pos min/max={0} / {1} im: {2} / {3}'.format(
pf(min(z.real)), pf(max(z.real)),
pf(min(z.imag)), pf(max(z.imag))))
return v
[docs]
def response(self, mod):
"""Forward response (apparent resistivity)."""
# ensure the mesh is initialized
self.mesh()
if self.complex() and self._conjImag:
pg.warn('flip imaginary part for response calc')
mod = self.flipImagPart(mod)
resp = self._core.response(mod)
if self.complex() and self._conjImag:
pg.warn('backflip imaginary part after response calc')
resp = self.flipImagPart(resp)
return resp
[docs]
def createJacobian(self, mod):
"""Compute Jacobian matrix and store but not return."""
# ensure the mesh is initialized
self.mesh()
if self.complex():
if self._conjImag:
pg.warn("Flipping imaginary part for jacobian calc")
mod = self.flipImagPart(mod)
self._core.createJacobian(mod)
self._J = pg.utils.squeezeComplex(self._core.jacobian(),
conj=self._conjImag
)
self.setJacobian(self._J)
# pg._r("create Jacobian", self, self._J)
return self._J
return self._core.createJacobian(mod)
[docs]
def setDataPost(self, data):
"""Set data (at a later stage)."""
self._core.setData(data)
[docs]
def setMeshPost(self, mesh):
"""Set mesh (at a later stage)."""
self._core.setMesh(mesh, ignoreRegionManager=True)
[docs]
class ERTModellingReference(ERTModellingBase):
"""Reference implementation for 2.5D Electrical Resistivity Tomography."""
[docs]
def __init__(self, **kwargs):
super().__init__()
self.subPotentials = None
self.lastResponse = None
# only for mixed boundary hack since this need to know resistivies.
self.resistivity = None
# abscissa k and weight for 2.5 inverse cos-transform
self.k = None
self.w = None
[docs]
def response(self, model):
"""Solve forward task and return apparent resistivity for self.mesh."""
# NOTE TODO can't be MT until mixed boundary condition depends on
# self.resistivity
pg.tic()
if not self.data.allNonZero('k'):
pg.error('Need valid geometric factors: "k".')
pg.warn('Fallback "k" values to -sign("rhoa")')
self.data.set('k', -pg.math.sign(self.data['rhoa']))
mesh = self.mesh()
nDof = mesh.nodeCount()
elecs = self.data.sensorPositions()
nEle = len(elecs)
nData = self.data.size()
self.resistivity = res = self.createMappedModel(model, -1.0)
if self.verbose:
print("Calculate response for model:", min(res), max(res))
rMin = elecs[0].dist(elecs[1]) / 2.0
rMax = elecs[0].dist(elecs[-1]) * 2.0
k, w = self.getIntegrationWeights(rMin, rMax)
self.k = k
self.w = w
# pg.show(mesh, res, label='res')
# pg.wait()
rhs = self.createRHS(mesh, elecs)
# store all potential fields
u = np.zeros((nEle, nDof))
self.subPotentials = [pg.Matrix(nEle, nDof) for i in range(len(k))]
for i, ki in enumerate(k):
uE = pg.solve(mesh, a=1./res, b=-(ki * ki)/res, f=rhs,
bc={'Robin': ['*', self.mixedBC]},
userData={'sourcePos': elecs, 'k': ki},
verbose=False, stats=0, debug=False)
self.subPotentials[i] = uE
u += w[i] * uE
# collect potential matrix,
# i.e., potential for all electrodes and all injections
pM = np.zeros((nEle, nEle))
for i in range(nEle):
pM[i] = pg.interpolate(mesh, u[i, :], destPos=elecs)
# collect resistivity values for all 4 pole measurements
r = np.zeros(nData)
for i in range(nData):
iA = int(self.data['a'][i])
iB = int(self.data['b'][i])
iM = int(self.data['m'][i])
iN = int(self.data['n'][i])
uAB = pM[iA] - pM[iB]
r[i] = uAB[iM] - uAB[iN]
self.lastResponse = r * self.data['k']
if self.verbose:
print("Resp min/max: {0} {1} {2}s".format(min(self.lastResponse),
max(self.lastResponse),
pg.dur()))
return self.lastResponse
[docs]
def createJacobian(self, model):
"""Create Jacobian matrix for model and store it in self.jacobian()."""
if self.subPotentials is None:
self.response(model)
J = self.jacobian()
J.resize(self.data.size(), self.parameterCount)
cells = self.mesh().findCellByMarker(0, -1)
Si = pg.matrix.ElementMatrix()
St = pg.matrix.ElementMatrix()
u = self.subPotentials
pg.tic()
if self.verbose:
print("Calculate sensitivity matrix for model: ",
min(model), max(model))
Jt = pg.Matrix(self.data.size(), self.parameterCount)
for kIdx, w in enumerate(self.w):
k = self.k[kIdx]
w = self.w[kIdx]
Jt *= 0.
A = pg.matrix.ElementMatrixMap()
for i, c in enumerate(cells):
modelIdx = c.marker()
# 2.5D
Si.u2(c)
Si *= k * k
Si += St.ux2uy2uz2(c)
# 3D
# Si.ux2uy2uz2(c); w = w* 2
A.add(modelIdx, Si)
for dataIdx in range(self.data.size()):
a = int(self.data['a'][dataIdx])
b = int(self.data['b'][dataIdx])
m = int(self.data['m'][dataIdx])
n = int(self.data['n'][dataIdx])
Jt[dataIdx] = A.mult(u[kIdx][a] - u[kIdx][b],
u[kIdx][m] - u[kIdx][n])
J += w * Jt
m2 = model*model
k = self.data['k']
for i in range(J.rows()):
J[i] /= (m2 / k[i])
if self.verbose:
sumsens = np.zeros(J.rows())
for i in range(J.rows()):
sumsens[i] = pg.sum(J[i])
print("sens sum: median = ", pg.math.median(sumsens),
" min = ", pg.min(sumsens),
" max = ", pg.max(sumsens))
[docs]
def calcGeometricFactor(self, data):
"""Calculate geometry factors for a given dataset."""
if pg.y(data.sensorPositions()) == pg.z(data.sensorPositions()):
k = np.zeros(data.size())
for i in range(data.size()):
a = data.sensorPosition(data['a'][i])
b = data.sensorPosition(data['b'][i])
m = data.sensorPosition(data['m'][i])
n = data.sensorPosition(data['n'][i])
k[i] = 1./(2.*np.pi) * (1./a.dist(m) - 1./a.dist(n) -
1./b.dist(m) + 1./b.dist(n))
return k
else:
raise BaseException("Please use BERT for non-standard "
"data sets" + str(data))
[docs]
def uAnalytical(self, p, sourcePos, k):
"""
Calculate analytical potential for homogeneous halfspace.
For sigma = 1 [S m]
"""
r1A = (p - sourcePos).abs()
# Mirror on surface at depth=0
r2A = (p - pg.RVector3(1.0, -1.0, 1.0) * sourcePos).abs()
if r1A > 1e-12 and r2A > 1e-12:
return (pg.math.besselK0(r1A * k) + pg.math.besselK0(r2A * k)) / \
(2.0 * np.pi)
else:
return 0.
[docs]
def getIntegrationWeights(self, rMin, rMax):
"""TODO WRITEME."""
nGauLegendre = max(int((6.0 * np.log10(rMax / rMin))), 4)
nGauLaguerre = 4
k = pg.Vector()
w = pg.Vector()
k0 = 1.0 / (2.0 * rMin)
pg.GaussLegendre(0.0, 1.0, nGauLegendre, k, w)
kLeg = k0 * k * k
wLeg = 2.0 * k0 * k * w / np.pi
pg.GaussLaguerre(nGauLaguerre, k, w)
kLag = k0 * (k + 1.0)
wLag = k0 * np.exp(k) * w / np.pi
return pg.cat(kLeg, kLag), pg.cat(wLeg, wLag)
[docs]
def mixedBC(self, boundary, userData):
"""Apply mixed boundary conditions."""
if boundary.marker() != pg.core.MARKER_BOUND_MIXED:
return 0
sourcePos = pg.center(userData['sourcePos'])
k = userData['k']
r1 = boundary.center() - sourcePos
# Mirror on surface at depth=0
r2 = boundary.center() - pg.RVector3(1.0, -1.0, 1.0) * sourcePos
r1A = r1.abs()
r2A = r2.abs()
rho = 1.
if self.resistivity is not None:
rho = self.resistivity[boundary.leftCell().id()]
n = boundary.norm()
if r1A > 1e-12 and r2A > 1e-12:
# see mod-dc-2d example for robin like BC and the negative sign
if (pg.math.besselK0(r1A * k) + pg.math.besselK0(r2A * k)) > 1e-12:
return k / rho * (r1.dot(n) / r1A * pg.math.besselK1(r1A * k) +
r2.dot(n) / r2A * pg.math.besselK1(r2A * k))\
/ (pg.math.besselK0(r1A * k) + pg.math.besselK0(r2A * k))
else:
return 0.
else:
return 0.
[docs]
def pointSource(self, cell, f, userData):
r"""
Define function for the current source term.
:math:`\delta(x-pos), \int f(x) \delta(x-pos)=f(pos)=N(pos)`
Right hand side entries will be shape functions(pos)
"""
i = userData['i']
sourcePos = userData['sourcePos'][i]
if cell.shape().isInside(sourcePos):
f.setVal(cell.N(cell.shape().rst(sourcePos)), cell.ids())
[docs]
def createRHS(self, mesh, elecs):
"""Create right-hand-side vector."""
rhs = np.zeros((len(elecs), mesh.nodeCount()))
for i, e in enumerate(elecs):
c = mesh.findCell(e)
rhs[i][c.ids()] = c.N(c.shape().rst(e))
return rhs
if __name__ == "__main__":
pass