def getIntegral(grid, level, index):
# create new grid
if grid.getType() in [LinearBoundary, LinearL0Boundary]:
return np.power(2., -max(1, level))
elif grid.getType() == Linear:
# # employ 4/3 rule
# if gp.isLeaf():
# q *= 4. / 3.
return np.power(2., -level)
elif grid.getType() == Poly:
return getBasis(grid).getIntegral(level, index)
elif grid.getType() == PolyBoundary:
return getBasis(grid).getIntegral(level, index)
else:
raise AttributeError('unsupported grid type %s' % grid.getType())
def getIntegral(grid, level, index):
# create new grid
if grid.getType() in [LinearBoundary, LinearL0Boundary]:
return np.power(2., -max(1, level))
elif grid.getType() == Linear:
# # employ 4/3 rule
# if gp.isLeaf():
# q *= 4. / 3.
return np.power(2., -level)
elif grid.getType() == Poly:
return getBasis(grid).getIntegral(level, index)
elif grid.getType() == PolyBoundary:
return getBasis(grid).getIntegral(level, index)
else:
raise AttributeError('unsupported grid type %s' % grid.getType())
def getIntegral(grid, level, index):
# create new grid
if grid.getType() in [
GridType_LinearBoundary, GridType_LinearTruncatedBoundary,
GridType_LinearL0Boundary
]:
return np.power(2., -max(1, level))
elif grid.getType() == GridType_Linear:
# # employ 4/3 rule
# if gp.isLeaf():
# q *= 4. / 3.
return np.power(2., -level)
elif grid.getType() in [
GridType_ModLinear, GridType_LinearClenshawCurtis,
GridType_LinearClenshawCurtisBoundary,
GridType_ModLinearClenshawCurtis, GridType_Poly,
GridType_PolyBoundary, GridType_ModPoly,
GridType_PolyClenshawCurtis, GridType_PolyClenshawCurtisBoundary,
GridType_ModPolyClenshawCurtis, GridType_Bspline,
GridType_BsplineBoundary, GridType_ModBspline,
GridType_BsplineClenshawCurtis, GridType_ModBsplineClenshawCurtis
]:
return getBasis(grid).getIntegral(level, index)
else:
raise AttributeError('unsupported grid type %s' % grid.getType())
def computeBilinearFormQuad(grid, U):
gs = grid.getStorage()
basis = getBasis(grid)
A = DataMatrix(gs.size(), gs.size())
level = DataMatrix(gs.size(), gs.dim())
index = DataMatrix(gs.size(), gs.dim())
gs.getLevelIndexArraysForEval(level, index)
s = np.ndarray(gs.dim(), dtype='float')
# run over all rows
for i in xrange(gs.size()):
gpi = gs.get(i)
# run over all columns
for j in xrange(i, gs.size()):
# print "%i/%i" % (i * gs.size() + j + 1, gs.size() ** 2)
gpj = gs.get(j)
for d in xrange(gs.dim()):
# get level index
lid, iid = level.get(i, d), index.get(i, d)
ljd, ijd = level.get(j, d), index.get(j, d)
# compute left and right boundary of the support of both
# basis functions
lb = max([(iid - 1) / lid, (ijd - 1) / ljd])
ub = min([(iid + 1) / lid, (ijd + 1) / ljd])
# same level, different index
if lid == ljd and iid != ijd:
s[d] = 0.
# the support does not overlap
elif lid != ljd and lb >= ub:
s[d] = 0.
else:
lid, iid = gpi.getLevel(d), int(iid)
ljd, ijd = gpj.getLevel(d), int(ijd)
# ----------------------------------------------------
# use scipy for integration
def f(x):
return basis.eval(lid, iid, x) * \
basis.eval(ljd, ijd, x) * \
U[d].pdf(x)
s[d], _ = quad(f, lb, ub, epsabs=1e-8)
# ----------------------------------------------------
A.set(i, j, float(np.prod(s)))
A.set(j, i, A.get(i, j))
return A
def computeBilinearFormQuad(grid, U):
gs = grid.getStorage()
basis = getBasis(grid)
A = DataMatrix(gs.size(), gs.size())
level = DataMatrix(gs.size(), gs.dim())
index = DataMatrix(gs.size(), gs.dim())
gs.getLevelIndexArraysForEval(level, index)
s = np.ndarray(gs.dim(), dtype='float')
# run over all rows
for i in xrange(gs.size()):
gpi = gs.get(i)
# run over all columns
for j in xrange(i, gs.size()):
# print "%i/%i" % (i * gs.size() + j + 1, gs.size() ** 2)
gpj = gs.get(j)
for d in xrange(gs.dim()):
# get level index
lid, iid = level.get(i, d), index.get(i, d)
ljd, ijd = level.get(j, d), index.get(j, d)
# compute left and right boundary of the support of both
# basis functions
lb = max([(iid - 1) / lid, (ijd - 1) / ljd])
ub = min([(iid + 1) / lid, (ijd + 1) / ljd])
# same level, different index
if lid == ljd and iid != ijd:
s[d] = 0.
# the support does not overlap
elif lid != ljd and lb >= ub:
s[d] = 0.
else:
lid, iid = gpi.getLevel(d), int(iid)
ljd, ijd = gpj.getLevel(d), int(ijd)
# ----------------------------------------------------
# use scipy for integration
def f(x):
return basis.eval(lid, iid, x) * \
basis.eval(ljd, ijd, x) * \
U[d].pdf(x)
s[d], _ = quad(f, lb, ub, epsabs=1e-8)
# ----------------------------------------------------
A.set(i, j, float(np.prod(s)))
A.set(j, i, A.get(i, j))
return A
def computeBFQuad(grid, U, admissibleSet, n=100):
"""
@param grid: Grid
@param U: list of distributions
@param admissibleSet: AdmissibleSet
@param n: int, number of MC samples
"""
gs = grid.getStorage()
basis = getBasis(grid)
A = DataMatrix(admissibleSet.getSize(), gs.size())
b = DataVector(admissibleSet.getSize())
s = np.ndarray(gs.dim(), dtype='float')
# run over all rows
for i, gpi in enumerate(admissibleSet.values()):
# run over all columns
for j in xrange(gs.size()):
# print "%i/%i" % (i * gs.size() + j + 1, gs.size() ** 2)
gpj = gs.get(j)
for d in xrange(gs.dim()):
# get level index
lid, iid = gpi.getLevel(d), gpi.getIndex(d)
ljd, ijd = gpj.getLevel(d), gpj.getIndex(d)
# compute left and right boundary of the support of both
# basis functions
xlow = max([(iid - 1) * 2 ** -lid, (ijd - 1) * 2 ** -ljd])
xhigh = min([(iid + 1) * 2 ** -lid, (ijd + 1) * 2 ** -ljd])
# same level, different index
if lid == ljd and iid != ijd:
s[d] = 0.
# the support does not overlap
elif lid != ljd and xlow >= xhigh:
s[d] = 0.
else:
# ----------------------------------------------------
# use scipy for integration
def f(x):
return basis.eval(lid, iid, x) * \
basis.eval(ljd, ijd, x) * \
U[d].pdf(x)
s[d], _ = quad(f, xlow, xhigh, epsabs=1e-8)
# ----------------------------------------------------
A.set(i, j, float(np.prod(s)))
if gs.seq(gpi) == j:
b[i] = A.get(i, j)
return A, b
def computeBFQuad(grid, U, admissibleSet, n=100):
"""
@param grid: Grid
@param U: list of distributions
@param admissibleSet: AdmissibleSet
@param n: int, number of MC samples
"""
gs = grid.getStorage()
basis = getBasis(grid)
A = DataMatrix(admissibleSet.getSize(), gs.size())
b = DataVector(admissibleSet.getSize())
s = np.ndarray(gs.getDimension(), dtype='float')
# run over all rows
for i, gpi in enumerate(admissibleSet.values()):
# run over all columns
for j in range(gs.size()):
# print "%i/%i" % (i * gs.size() + j + 1, gs.size() ** 2)
gpj = gs.getPoint(j)
for d in range(gs.getDimension()):
# get level index
lid, iid = gpi.getLevel(d), gpi.getIndex(d)
ljd, ijd = gpj.getLevel(d), gpj.getIndex(d)
# compute left and right boundary of the support of both
# basis functions
xlow = max([(iid - 1) * 2 ** -lid, (ijd - 1) * 2 ** -ljd])
xhigh = min([(iid + 1) * 2 ** -lid, (ijd + 1) * 2 ** -ljd])
# same level, different index
if lid == ljd and iid != ijd:
s[d] = 0.
# the support does not overlap
elif lid != ljd and xlow >= xhigh:
s[d] = 0.
else:
# ----------------------------------------------------
# use scipy for integration
def f(x):
return basis.eval(lid, iid, x) * \
basis.eval(ljd, ijd, x) * \
U[d].pdf(x)
s[d], _ = quad(f, xlow, xhigh, epsabs=1e-8)
# ----------------------------------------------------
A.set(i, j, float(np.prod(s)))
if gs.getSequenceNumber(gpi) == j:
b[i] = A.get(i, j)
return A, b
def computeLinearForm(self, grid):
"""
Compute bilinear form for the current grid
@param grid: Grid
@return numpy array
"""
gs = grid.getStorage()
basis = getBasis(grid)
v = np.ndarray(gs.size())
err = 0.
# run over all rows
for i in range(gs.size()):
gpi = gs.getPoint(i)
# compute bilinear form for one entry
v[i], erri = self.getLinearFormEntry(gs, gpi, basis)
err += erri
return v, err
def computeLinearForm(self, grid):
"""
Compute bilinear form for the current grid
@param grid: Grid
@return DataVector
"""
gs = grid.getStorage()
basis = getBasis(grid)
v = DataVector(gs.size())
err = 0.
# run over all rows
for i in xrange(gs.size()):
gpi = gs.get(i)
# compute bilinear form for one entry
v[i], erri = self.getLinearFormEntry(gpi, basis)
err += erri
return v, err
def computeLinearForm(self, grid):
"""
Compute bilinear form for the current grid
@param grid: Grid
@return DataVector
"""
gs = grid.getStorage()
basis = getBasis(grid)
v = DataVector(gs.size())
err = 0.
# run over all rows
for i in xrange(gs.size()):
gpi = gs.get(i)
# compute bilinear form for one entry
v[i], erri = self.getLinearFormEntry(gpi, basis)
err += erri
return v, err
def computeExpectationValueEstimation(grid, U, admissibleSet):
"""
Compute
(b)_i = \int phi_i dU(x)
on measure U, which is in this case supposed to be a lebesgue measure.
@param grid: Grid, sparse grid
@param U: list of distributions, Lebeasgue measure
@param admissibleSet: AdmissibleSet
@return: DataVector
"""
gs = grid.getStorage()
basis = getBasis(grid)
# interpolate phi_i phi_j on sparse grid with piecewise polynomial SG
# the product of two piecewise linear functions is a piecewise
# polynomial one of degree 2.
b = DataVector(admissibleSet.getSize())
s = np.ndarray(gs.dim(), dtype='float')
# run over all rows
p = DataVector(gs.dim())
for i, gpi in enumerate(admissibleSet.values()):
gpi.getCoords(p)
for d in xrange(gs.dim()):
# get level index
lid, iid = gpi.getLevel(d), gpi.getIndex(d)
xlow = (iid - 1) * 2**-lid
xhigh = (iid + 1) * 2**-lid
# ----------------------------------------------------
# use scipy integration
def f(x):
return basis.eval(lid, iid, x) * U[d].pdf(x)
s[d], _ = quad(f, xlow, xhigh, epsabs=1e-8)
# ----------------------------------------------------
b[i] = float(np.prod(s))
return b
def computeExpectationValueEstimation(grid, U, admissibleSet):
"""
Compute
(b)_i = \int phi_i dU(x)
on measure U, which is in this case supposed to be a lebesgue measure.
@param grid: Grid, sparse grid
@param U: list of distributions, Lebeasgue measure
@param admissibleSet: AdmissibleSet
@return: DataVector
"""
gs = grid.getStorage()
basis = getBasis(grid)
# interpolate phi_i phi_j on sparse grid with piecewise polynomial SG
# the product of two piecewise linear functions is a piecewise
# polynomial one of degree 2.
b = DataVector(admissibleSet.getSize())
s = np.ndarray(gs.dim(), dtype='float')
# run over all rows
p = DataVector(gs.dim())
for i, gpi in enumerate(admissibleSet.values()):
gpi.getCoords(p)
for d in xrange(gs.dim()):
# get level index
lid, iid = gpi.getLevel(d), gpi.getIndex(d)
xlow = (iid - 1) * 2 ** -lid
xhigh = (iid + 1) * 2 ** -lid
# ----------------------------------------------------
# use scipy integration
def f(x):
return basis.eval(lid, iid, x) * U[d].pdf(x)
s[d], _ = quad(f, xlow, xhigh, epsabs=1e-8)
# ----------------------------------------------------
b[i] = float(np.prod(s))
return b
def computeBilinearForm(grid, U):
"""
Compute bilinear form
(A)_ij = \int phi_i phi_j dU(x)
on measure U, which is in this case supposed to be a lebesgue measure.
@param grid: Grid, sparse grid
@param U: list of distributions, Lebeasgue measure
@return: DataMatrix
"""
gs = grid.getStorage()
basis = getBasis(grid)
# interpolate phi_i phi_j on sparse grid with piecewise polynomial SG
# the product of two piecewise linear functions is a piecewise
# polynomial one of degree 2.
ngrid = Grid.createPolyBoundaryGrid(1, 2)
# ngrid = Grid.createLinearBoundaryGrid(1)
ngrid.getGenerator().regular(gs.getMaxLevel() + 1)
ngs = ngrid.getStorage()
nodalValues = DataVector(ngs.size())
level = DataMatrix(gs.size(), gs.getDimension())
index = DataMatrix(gs.size(), gs.getDimension())
gs.getLevelIndexArraysForEval(level, index)
A = DataMatrix(gs.size(), gs.size())
s = np.ndarray(gs.getDimension(), dtype='float')
# run over all rows
for i in range(gs.size()):
gpi = gs.getPoint(i)
# run over all columns
for j in range(i, gs.size()):
# print "%i/%i" % (i * gs.size() + j + 1, gs.size() ** 2)
gpj = gs.getPoint(j)
# run over all dimensions
for d in range(gs.getDimension()):
# get level index
lid, iid = level.get(i, d), index.get(i, d)
ljd, ijd = level.get(j, d), index.get(j, d)
# compute left and right boundary of the support of both
# basis functions
lb = max([((iid - 1) / lid), ((ijd - 1) / ljd)])
ub = min([((iid + 1) / lid), ((ijd + 1) / ljd)])
# same level, different index
if lid == ljd and iid != ijd:
s[d] = 0.
# the support does not overlap
elif lid != ljd and lb >= ub:
s[d] = 0.
else:
# ----------------------------------------------------
# do the 1d interpolation ...
lid, iid = gpi.getLevel(d), int(iid)
ljd, ijd = gpj.getLevel(d), int(ijd)
for k in range(ngs.size()):
x = ngs.getCoordinate(ngs.getPoint(k), 0)
nodalValues[k] = max(0, basis.eval(lid, iid, x)) * \
max(0, basis.eval(ljd, ijd, x))
# ... by hierarchization
v = hierarchize(ngrid, nodalValues)
def f(x, y):
return float(y * U[d].pdf(x[0]))
g, w, _ = discretize(ngrid, v, f, refnums=0)
# compute the integral of it
s[d] = doQuadrature(g, w)
# ----------------------------------------------------
# store result in matrix
A.set(i, j, float(np.prod(s)))
A.set(j, i, A.get(i, j))
return A
def computeBF(grid, U, admissibleSet):
"""
Compute bilinear form
(A)_ij = \int phi_i phi_j dU(x)
on measure U, which is in this case supposed to be a lebesgue measure.
@param grid: Grid, sparse grid
@param U: list of distributions, Lebeasgue measure
@param admissibleSet: AdmissibleSet
@return: DataMatrix
"""
gs = grid.getStorage()
basis = getBasis(grid)
# interpolate phi_i phi_j on sparse grid with piecewise polynomial SG
# the product of two piecewise linear functions is a piecewise
# polynomial one of degree 2.
ngrid = Grid.createPolyBoundaryGrid(1, 2)
ngrid.createGridGenerator().regular(2)
ngs = ngrid.getStorage()
nodalValues = DataVector(ngs.size())
A = DataMatrix(admissibleSet.getSize(), gs.size())
b = DataVector(admissibleSet.getSize())
s = np.ndarray(gs.dim(), dtype='float')
# # pre compute basis evaluations
# basis_eval = {}
# for li in xrange(1, gs.getMaxLevel() + 1):
# for i in xrange(1, 2 ** li + 1, 2):
# # add value with it self
# x = 2 ** -li * i
# basis_eval[(li, i, li, i, x)] = basis.eval(li, i, x) * \
# basis.eval(li, i, x)
#
# # left side
# x = 2 ** -(li + 1) * (2 * i - 1)
# basis_eval[(li, i, li, i, x)] = basis.eval(li, i, x) * \
# basis.eval(li, i, x)
# # right side
# x = 2 ** -(li + 1) * (2 * i + 1)
# basis_eval[(li, i, li, i, x)] = basis.eval(li, i, x) * \
# basis.eval(li, i, x)
#
# # add values for hierarchical lower nodes
# for lj in xrange(li + 1, gs.getMaxLevel() + 1):
# a = 2 ** (lj - li)
# j = a * i - a + 1
# while j < a * i + a:
# # center
# x = 2 ** -lj * j
# basis_eval[(li, i, lj, j, x)] = basis.eval(li, i, x) * \
# basis.eval(lj, j, x)
# basis_eval[(lj, j, li, i, x)] = basis_eval[(li, i, lj, j, x)]
# # left side
# x = 2 ** -(lj + 1) * (2 * j - 1)
# basis_eval[(li, i, lj, j, x)] = basis.eval(li, i, x) * \
# basis.eval(lj, j, x)
# basis_eval[(lj, j, li, i, x)] = basis_eval[(li, i, lj, j, x)]
# # right side
# x = 2 ** -(lj + 1) * (2 * j + 1)
# basis_eval[(li, i, lj, j, x)] = basis.eval(li, i, x) * \
# basis.eval(lj, j, x)
# basis_eval[(lj, j, li, i, x)] = basis_eval[(li, i, lj, j, x)]
# j += 2
#
# print len(basis_eval)
# run over all rows
for i, gpi in enumerate(admissibleSet.values()):
# run over all columns
for j in xrange(gs.size()):
# print "%i/%i" % (i * gs.size() + j + 1, gs.size() ** 2)
gpj = gs.get(j)
for d in xrange(gs.dim()):
# get level index
lid, iid = gpi.getLevel(d), gpi.getIndex(d)
ljd, ijd = gpj.getLevel(d), gpj.getIndex(d)
# compute left and right boundary of the support of both
# basis functions
lb = max([(iid - 1) * 2 ** -lid, (ijd - 1) * 2 ** -ljd])
ub = min([(iid + 1) * 2 ** -lid, (ijd + 1) * 2 ** -ljd])
# same level, different index
if lid == ljd and iid != ijd:
s[d] = 0.
# the support does not overlap
elif lid != ljd and lb >= ub:
s[d] = 0.
else:
# ----------------------------------------------------
# do the 1d interpolation ...
# define transformation function
T = LinearTransformation(lb, ub)
for k in xrange(ngs.size()):
x = ngs.get(k).getCoord(0)
x = T.unitToProbabilistic(x)
nodalValues[k] = basis.eval(lid, iid, x) * \
basis.eval(ljd, ijd, x)
# ... by hierarchization
v = hierarchize(ngrid, nodalValues)
# discretize the following function
def f(x, y):
xp = T.unitToProbabilistic(x)
return float(y * U[d].pdf(xp))
# sparse grid quadrature
g, w, _ = discretize(ngrid, v, f, refnums=0, level=5,
useDiscreteL2Error=False)
s[d] = doQuadrature(g, w) * (ub - lb)
# fig = plt.figure()
# plotSG1d(ngrid, v)
# x = np.linspace(xlow, ub, 100)
# plt.plot(np.linspace(0, 1, 100), U[d].pdf(x))
# fig.show()
# fig = plt.figure()
# plotSG1d(g, w)
# x = np.linspace(0, 1, 100)
# plt.plot(x,
# [evalSGFunction(ngrid, v, DataVector([xi])) * U[d].pdf(T.unitToProbabilistic(xi)) for xi in x])
# fig.show()
# plt.show()
# compute the integral of it
# ----------------------------------------------------
A.set(i, j, float(np.prod(s)))
if gs.seq(gpi) == j:
b[i] = A.get(i, j)
return A, b
def __doMarginalize(grid, alpha, dd, measure=None):
gs = grid.getStorage()
dim = gs.dim()
if dim < 2:
raise AttributeError("The grid has to be at least of dimension 2")
if dd >= dim:
raise AttributeError("The grid has only %i dimensions, so I can't \
integrate over %i" % (dim, dd))
# create new grid
n_dim = dim - 1
n_grid = createGrid(grid, n_dim)
n_gs = n_grid.getStorage()
# insert grid points
n_gp = HashGridIndex(n_dim)
for i in xrange(gs.size()):
gp = gs.get(i)
for d in range(dim):
if d == dd:
# omit marginalization direction
continue
elif d < dd:
n_gp.set(d, gp.getLevel(d), gp.getIndex(d))
else:
n_gp.set(d - 1, gp.getLevel(d), gp.getIndex(d))
# insert grid point
if not n_gs.has_key(n_gp):
n_gs.insert(n_gp)
n_gs.recalcLeafProperty()
# create coefficient vector
n_alpha = DataVector(n_gs.size())
n_alpha.setAll(0.0)
# set function values for n_alpha
for i in xrange(gs.size()):
gp = gs.get(i)
for d in range(dim):
if d == dd:
dd_level = gp.getLevel(d)
dd_index = gp.getIndex(d)
elif d < dd:
n_gp.set(d, gp.getLevel(d), gp.getIndex(d))
else:
n_gp.set(d - 1, gp.getLevel(d), gp.getIndex(d))
if not n_gs.has_key(n_gp):
raise Exception("This should not happen!")
# compute the integral of the given basis
if measure is None:
q, err = getIntegral(grid, dd_level, dd_index), 0.
else:
dist, trans = measure[0][dd], measure[1][dd]
lf = LinearGaussQuadratureStrategy([dist], [trans])
basis = getBasis(grid)
gpdd = HashGridIndex(1)
gpdd.set(0, dd_level, dd_index)
q, err = lf.computeLinearFormByList([gpdd], basis)
q = q[0]
# search for the corresponding index
j = n_gs.seq(n_gp)
n_alpha[j] += alpha[i] * q
return n_grid, n_alpha, err
def computeBilinearForm(grid, U):
"""
Compute bilinear form
(A)_ij = \int phi_i phi_j dU(x)
on measure U, which is in this case supposed to be a lebesgue measure.
@param grid: Grid, sparse grid
@param U: list of distributions, Lebeasgue measure
@return: DataMatrix
"""
gs = grid.getStorage()
basis = getBasis(grid)
# interpolate phi_i phi_j on sparse grid with piecewise polynomial SG
# the product of two piecewise linear functions is a piecewise
# polynomial one of degree 2.
ngrid = Grid.createPolyBoundaryGrid(1, 2)
# ngrid = Grid.createLinearBoundaryGrid(1)
ngrid.createGridGenerator().regular(gs.getMaxLevel() + 1)
ngs = ngrid.getStorage()
nodalValues = DataVector(ngs.size())
level = DataMatrix(gs.size(), gs.dim())
index = DataMatrix(gs.size(), gs.dim())
gs.getLevelIndexArraysForEval(level, index)
A = DataMatrix(gs.size(), gs.size())
s = np.ndarray(gs.dim(), dtype='float')
# run over all rows
for i in xrange(gs.size()):
gpi = gs.get(i)
# run over all columns
for j in xrange(i, gs.size()):
# print "%i/%i" % (i * gs.size() + j + 1, gs.size() ** 2)
gpj = gs.get(j)
# run over all dimensions
for d in xrange(gs.dim()):
# get level index
lid, iid = level.get(i, d), index.get(i, d)
ljd, ijd = level.get(j, d), index.get(j, d)
# compute left and right boundary of the support of both
# basis functions
lb = max([(iid - 1) / lid, (ijd - 1) / ljd])
ub = min([(iid + 1) / lid, (ijd + 1) / ljd])
# same level, different index
if lid == ljd and iid != ijd:
s[d] = 0.
# the support does not overlap
elif lid != ljd and lb >= ub:
s[d] = 0.
else:
# ----------------------------------------------------
# do the 1d interpolation ...
lid, iid = gpi.getLevel(d), int(iid)
ljd, ijd = gpj.getLevel(d), int(ijd)
for k in xrange(ngs.size()):
x = ngs.get(k).getCoord(0)
nodalValues[k] = max(0, basis.eval(lid, iid, x)) * \
max(0, basis.eval(ljd, ijd, x))
# ... by hierarchization
v = hierarchize(ngrid, nodalValues)
def f(x, y):
return float(y * U[d].pdf(x[0]))
g, w, _ = discretize(ngrid, v, f, refnums=0)
# compute the integral of it
s[d] = doQuadrature(g, w)
# ----------------------------------------------------
# store result in matrix
A.set(i, j, float(np.prod(s)))
A.set(j, i, A.get(i, j))
return A
def __doMarginalize(grid, alpha, linearForm, dd, measure=None):
gs = grid.getStorage()
dim = gs.getDimension()
if dim < 2:
raise AttributeError("The grid has to be at least of dimension 2")
if dd >= dim:
raise AttributeError("The grid has only %i dimensions, so I can't \
integrate over %i" % (dim, dd))
# create new grid
n_dim = dim - 1
n_grid = createGrid(grid, n_dim)
n_gs = n_grid.getStorage()
# insert grid points
n_gp = HashGridPoint(n_dim)
for i in range(gs.getSize()):
gp = gs.getPoint(i)
for d in range(dim):
if d == dd:
# omit marginalization direction
continue
elif d < dd:
n_gp.set(d, gp.getLevel(d), gp.getIndex(d))
else:
n_gp.set(d - 1, gp.getLevel(d), gp.getIndex(d))
# insert grid point
if not n_gs.isContaining(n_gp):
n_gs.insert(n_gp)
n_gs.recalcLeafProperty()
# create coefficient vector
n_alpha = np.zeros(n_gs.getSize())
basis = getBasis(grid)
# set function values for n_alpha
for i in range(gs.getSize()):
gp = gs.getPoint(i)
for d in range(dim):
if d == dd:
dd_level = gp.getLevel(d)
dd_index = gp.getIndex(d)
elif d < dd:
n_gp.set(d, gp.getLevel(d), gp.getIndex(d))
else:
n_gp.set(d - 1, gp.getLevel(d), gp.getIndex(d))
if not n_gs.isContaining(n_gp):
raise Exception("This should not happen!")
# compute the integral of the given basis
if measure is None:
q, err = getIntegral(grid, dd_level, dd_index), 0.
else:
dist, trans = measure[0][dd], measure[1][dd]
linearForm.setDistributionAndTransformation([dist], [trans])
gpdd = HashGridPoint(1)
gpdd.set(0, dd_level, dd_index)
q, err = linearForm.computeLinearFormByList(gs, [gpdd], basis)
q = q[0] * trans.vol()
err *= trans.vol()
# search for the corresponding index
j = n_gs.getSequenceNumber(n_gp)
n_alpha[j] += alpha[i] * q
return n_grid, n_alpha, err
def computeBF(grid, U, admissibleSet):
"""
Compute bilinear form
(A)_ij = \int phi_i phi_j dU(x)
on measure U, which is in this case supposed to be a lebesgue measure.
@param grid: Grid, sparse grid
@param U: list of distributions, Lebeasgue measure
@param admissibleSet: AdmissibleSet
@return: DataMatrix
"""
gs = grid.getStorage()
basis = getBasis(grid)
# interpolate phi_i phi_j on sparse grid with piecewise polynomial SG
# the product of two piecewise linear functions is a piecewise
# polynomial one of degree 2.
ngrid = Grid.createPolyBoundaryGrid(1, 2)
ngrid.getGenerator().regular(2)
ngs = ngrid.getStorage()
nodalValues = DataVector(ngs.size())
A = DataMatrix(admissibleSet.getSize(), gs.size())
b = DataVector(admissibleSet.getSize())
s = np.ndarray(gs.getDimension(), dtype='float')
# # pre compute basis evaluations
# basis_eval = {}
# for li in xrange(1, gs.getMaxLevel() + 1):
# for i in xrange(1, 2 ** li + 1, 2):
# # add value with it self
# x = 2 ** -li * i
# basis_eval[(li, i, li, i, x)] = basis.eval(li, i, x) * \
# basis.eval(li, i, x)
#
# # left side
# x = 2 ** -(li + 1) * (2 * i - 1)
# basis_eval[(li, i, li, i, x)] = basis.eval(li, i, x) * \
# basis.eval(li, i, x)
# # right side
# x = 2 ** -(li + 1) * (2 * i + 1)
# basis_eval[(li, i, li, i, x)] = basis.eval(li, i, x) * \
# basis.eval(li, i, x)
#
# # add values for hierarchical lower nodes
# for lj in xrange(li + 1, gs.getMaxLevel() + 1):
# a = 2 ** (lj - li)
# j = a * i - a + 1
# while j < a * i + a:
# # center
# x = 2 ** -lj * j
# basis_eval[(li, i, lj, j, x)] = basis.eval(li, i, x) * \
# basis.eval(lj, j, x)
# basis_eval[(lj, j, li, i, x)] = basis_eval[(li, i, lj, j, x)]
# # left side
# x = 2 ** -(lj + 1) * (2 * j - 1)
# basis_eval[(li, i, lj, j, x)] = basis.eval(li, i, x) * \
# basis.eval(lj, j, x)
# basis_eval[(lj, j, li, i, x)] = basis_eval[(li, i, lj, j, x)]
# # right side
# x = 2 ** -(lj + 1) * (2 * j + 1)
# basis_eval[(li, i, lj, j, x)] = basis.eval(li, i, x) * \
# basis.eval(lj, j, x)
# basis_eval[(lj, j, li, i, x)] = basis_eval[(li, i, lj, j, x)]
# j += 2
#
# print len(basis_eval)
# run over all rows
for i, gpi in enumerate(admissibleSet.values()):
# run over all columns
for j in range(gs.size()):
# print "%i/%i" % (i * gs.size() + j + 1, gs.size() ** 2)
gpj = gs.getPoint(j)
for d in range(gs.getDimension()):
# get level index
lid, iid = gpi.getLevel(d), gpi.getIndex(d)
ljd, ijd = gpj.getLevel(d), gpj.getIndex(d)
# compute left and right boundary of the support of both
# basis functions
lb = max([(iid - 1) * 2 ** -lid, (ijd - 1) * 2 ** -ljd])
ub = min([(iid + 1) * 2 ** -lid, (ijd + 1) * 2 ** -ljd])
# same level, different index
if lid == ljd and iid != ijd:
s[d] = 0.
# the support does not overlap
elif lid != ljd and lb >= ub:
s[d] = 0.
else:
# ----------------------------------------------------
# do the 1d interpolation ...
# define transformation function
T = LinearTransformation(lb, ub)
for k in range(ngs.size()):
x = ngs.getCoordinate(ngs.getPoint(k), 0)
x = T.unitToProbabilistic(x)
nodalValues[k] = basis.eval(lid, iid, x) * \
basis.eval(ljd, ijd, x)
# ... by hierarchization
v = hierarchize(ngrid, nodalValues)
# discretize the following function
def f(x, y):
xp = T.unitToProbabilistic(x)
return float(y * U[d].pdf(xp))
# sparse grid quadrature
g, w, _ = discretize(ngrid, v, f, refnums=0, level=5,
useDiscreteL2Error=False)
s[d] = doQuadrature(g, w) * (ub - lb)
# fig = plt.figure()
# plotSG1d(ngrid, v)
# x = np.linspace(xlow, ub, 100)
# plt.plot(np.linspace(0, 1, 100), U[d].pdf(x))
# fig.show()
# fig = plt.figure()
# plotSG1d(g, w)
# x = np.linspace(0, 1, 100)
# plt.plot(x,
# [evalSGFunction(ngrid, v, DataVector([xi])) * U[d].pdf(T.unitToProbabilistic(xi)) for xi in x])
# fig.show()
# plt.show()
# compute the integral of it
# ----------------------------------------------------
A.set(i, j, float(np.prod(s)))
if gs.getSequenceNumber(gpi) == j:
b[i] = A.get(i, j)
return A, b