def test_gridPointsInPolygon(self):
#
# Make a polygon, check the grid points exist where we need them
# Simple example -- no points on the trial grid are excluded
myPoly = [[0., 10.], [10., 10.], [10., 0.], [0., 0.]]
pip = su.gridPointsInPolygon(myPoly, approx_grid_spacing=[1., 1.])
# There should be 121 points in total
assert(pip.shape[0] == 121)
# The min/max x and y should be inside
assert(min(pip[:, 0]) > 0.)
assert(max(pip[:, 0]) < 10.)
assert(min(pip[:, 1]) > 0.)
assert(max(pip[:, 1]) < 10.)
# Example where some points on the trial grid would be excluded
myPoly = [[0., 10.], [10., 10.], [10., 0.]]
pip = su.gridPointsInPolygon(myPoly, approx_grid_spacing=[1., 1.])
# The min/max x and y should be inside
assert(min(pip[:, 0]) > 0.)
assert(max(pip[:, 0]) < 10.)
assert(min(pip[:, 1]) > 0.)
assert(max(pip[:, 1]) < 10.)
# x+y should always be >= 10, since the line connecting [0,10] and
# [10,0] is x+y=10
assert(all(pip[:, 0]+pip[:, 1] >= 10.))
def test_gridPointsInPolygon(self):
#
# Make a polygon, check the grid points exist where we need them
## Simple example -- no points on the trial grid are excluded
myPoly=[[0., 10.], [10., 10.], [10., 0.], [0., 0.]]
pip=su.gridPointsInPolygon(myPoly, approx_grid_spacing=[1.,1.])
# There should be 121 points in total
assert(pip.shape[0]==121)
# The min/max x and y should be inside
assert(min(pip[:,0])>0.)
assert(max(pip[:,0])<10.)
assert(min(pip[:,1])>0.)
assert(max(pip[:,1])<10.)
## Example where some points on the trial grid would be excluded
myPoly=[[0., 10.], [10., 10.], [10., 0.]]
pip=su.gridPointsInPolygon(myPoly, approx_grid_spacing=[1.,1.])
# The min/max x and y should be inside
assert(min(pip[:,0])>0.)
assert(max(pip[:,0])<10.)
assert(min(pip[:,1])>0.)
assert(max(pip[:,1])<10.)
# x+y should always be >= 10, since the line connecting [0,10] and
# [10,0] is x+y=10
assert(all(pip[:,0]+pip[:,1]>=10.))
def elevation_setter(xc, yc):
# Return scipy array of values
out = xc * 0.
# Get multiple elevation values in each triangle.
# Process triangles in chunks to reduce function call overhead
lx = len(xc)
lx_div_cs = scipy.ceil(lx * 1. / (1. * chunk_size)).astype(int)
# Crude check that xc/yc are the centroid values
#
erMess = ' Result of make_meanFun can ONLY be applied to a vector' +\
' of ALL centroid coordinates\n' +\
' (since mesh triangles are used to spatially average)'
assert scipy.all(xc == domain.centroid_coordinates[:, 0]), erMess
assert scipy.all(yc == domain.centroid_coordinates[:, 1]), erMess
# Find triangles in which we want to average
if polygons_for_averaging is not None:
averaging_flag = 0 * xc
# Need georeferenced centroid coordinates to find which
# are in the polygon
xll = domain.geo_reference.xllcorner
yll = domain.geo_reference.yllcorner
centroid_coordinates_georef = scipy.vstack([xc + xll,
yc + yll]).transpose()
for j in range(len(polygons_for_averaging)):
poly_j = polygons_for_averaging[j]
# poly_j can either be a polygon, or a filename
if type(poly_j) is str:
poly_j = su.read_polygon(poly_j)
points_in_poly_j = inside_polygon(centroid_coordinates_georef,
poly_j)
averaging_flag[points_in_poly_j] = 1
else:
averaging_flag = 1 + 0 * xc
for i in range(lx_div_cs):
# Evaluate in triangles lb:ub
lb = i * chunk_size
ub = min((i + 1) * chunk_size, lx)
if verbose:
print 'Averaging in triangles ', lb, '-', ub - 1
# Store x,y,triangleIndex
px = scipy.array([])
py = scipy.array([])
p_indices = scipy.array([])
for j in range(lb, ub):
# If we average this cell, then get a grid
# of points in it. Otherwise just get the centroid
# coordinates.
if averaging_flag[j] == 1:
mesh_tri = \
domain.mesh.vertex_coordinates[
range(3 * j, 3 * j + 3), :].tolist()
pts = su.gridPointsInPolygon(
mesh_tri, approx_grid_spacing=approx_grid_spacing)
else:
# Careful to keep this a 2D array
pts = domain.centroid_coordinates[j, :, None].transpose()
px = scipy.hstack([px, pts[:, 0]])
py = scipy.hstack([py, pts[:, 1]])
p_indices = scipy.hstack(
[p_indices, scipy.repeat(j, len(pts[:, 0]))])
# Get function values at all px,py
if verbose:
print ' Evaluating function at ', len(px), ' points'
allTopo = q_function(px, py)
# Set output values in lb:ub
for j in range(lb, ub):
out_indices = (p_indices == j).nonzero()[0]
assert len(out_indices) > 0
if (averaging == 'mean'):
out[j] = allTopo[out_indices].mean()
elif (averaging == 'min'):
out[j] = allTopo[out_indices].min()
elif (averaging == 'max'):
out[j] = allTopo[out_indices].max()
else:
raise Exception('Unknown value of averaging')
return (out)
def elevation_setter(xc, yc):
# Return scipy array of values
out = xc * 0.
# Get multiple elevation values in each triangle.
# Process triangles in chunks to reduce function call overhead
lx = len(xc)
lx_div_cs = scipy.ceil(lx * 1. / (1. * chunk_size)).astype(int)
# Crude check that xc/yc are the centroid values
#
erMess = ' Result of make_meanFun can ONLY be applied to a vector' +\
' of ALL centroid coordinates\n' +\
' (since mesh triangles are used to spatially average)'
assert scipy.all(xc == domain.centroid_coordinates[:, 0]), erMess
assert scipy.all(yc == domain.centroid_coordinates[:, 1]), erMess
for i in range(lx_div_cs):
# Evaluate in triangles lb:ub
lb = i * chunk_size
ub = min((i + 1) * chunk_size, lx)
if verbose:
print 'Averaging in triangles ', lb, '-', ub - 1
# Store x,y,triangleIndex
px = scipy.array([])
py = scipy.array([])
p_indices = scipy.array([])
for j in range(lb, ub):
mesh_tri = \
domain.mesh.vertex_coordinates[
range(3 * j, 3 * j + 3), :].tolist()
pts = su.gridPointsInPolygon(
mesh_tri, approx_grid_spacing=approx_grid_spacing)
px = scipy.hstack([px, pts[:, 0]])
py = scipy.hstack([py, pts[:, 1]])
p_indices = scipy.hstack(
[p_indices, scipy.repeat(j, len(pts[:, 0]))])
# Get function values at all px,py
if verbose:
print ' Evaluating function at ', len(px), ' points'
allTopo = q_function(px, py)
# Set output values in lb:ub
for j in range(lb, ub):
out_indices = (p_indices == j).nonzero()[0]
assert len(out_indices) > 0
if (averaging == 'mean'):
out[j] = allTopo[out_indices].mean()
elif (averaging == 'min'):
out[j] = allTopo[out_indices].min()
elif (averaging == 'max'):
out[j] = allTopo[out_indices].max()
elif (averaging == 'harmonic_mean'):
out[j] = 1.0 / (1.0 / allTopo[out_indices]).mean()
else:
raise Exception, 'Unknown value of averaging'
return (out)
def elevation_setter(xc, yc):
# Return scipy array of values
out = xc * 0.
# Get multiple elevation values in each triangle.
# Process triangles in chunks to reduce function call overhead
lx = len(xc)
lx_div_cs = scipy.ceil(lx * 1. / (1. * chunk_size)).astype(int)
# Crude check that xc/yc are the centroid values
#
erMess = ' Result of make_meanFun can ONLY be applied to a vector' +\
' of ALL centroid coordinates\n' +\
' (since mesh triangles are used to spatially average)'
assert scipy.all(xc == domain.centroid_coordinates[:, 0]), erMess
assert scipy.all(yc == domain.centroid_coordinates[:, 1]), erMess
# Find triangles in which we want to average
if polygons_for_averaging is not None:
averaging_flag = 0*xc
# Need georeferenced centroid coordinates to find which
# are in the polygon
xll = domain.geo_reference.xllcorner
yll = domain.geo_reference.yllcorner
centroid_coordinates_georef = scipy.vstack([xc + xll, yc + yll]).transpose()
for j in range(len(polygons_for_averaging)):
poly_j = polygons_for_averaging[j]
# poly_j can either be a polygon, or a filename
if type(poly_j) is str:
poly_j = su.read_polygon(poly_j)
points_in_poly_j = inside_polygon(centroid_coordinates_georef,
poly_j)
averaging_flag[points_in_poly_j] = 1
else:
averaging_flag = 1 + 0*xc
for i in range(lx_div_cs):
# Evaluate in triangles lb:ub
lb = i * chunk_size
ub = min((i + 1) * chunk_size, lx)
if verbose:
print 'Averaging in triangles ', lb, '-', ub - 1
# Store x,y,triangleIndex
px = scipy.array([])
py = scipy.array([])
p_indices = scipy.array([])
for j in range(lb, ub):
# If we average this cell, then get a grid
# of points in it. Otherwise just get the centroid
# coordinates.
if averaging_flag[j] == 1:
mesh_tri = \
domain.mesh.vertex_coordinates[
range(3 * j, 3 * j + 3), :].tolist()
pts = su.gridPointsInPolygon(
mesh_tri,
approx_grid_spacing=approx_grid_spacing)
else:
# Careful to keep this a 2D array
pts = domain.centroid_coordinates[j,:, None].transpose()
px = scipy.hstack([px, pts[:, 0]])
py = scipy.hstack([py, pts[:, 1]])
p_indices = scipy.hstack([p_indices,
scipy.repeat(j, len(pts[:, 0]))])
# Get function values at all px,py
if verbose:
print ' Evaluating function at ', len(px), ' points'
allTopo = q_function(px, py)
# Set output values in lb:ub
for j in range(lb, ub):
out_indices = (p_indices == j).nonzero()[0]
assert len(out_indices) > 0
if(averaging == 'mean'):
out[j] = allTopo[out_indices].mean()
elif(averaging == 'min'):
out[j] = allTopo[out_indices].min()
elif(averaging == 'max'):
out[j] = allTopo[out_indices].max()
else:
raise Exception('Unknown value of averaging')
return(out)
def elevation_setter(xc, yc):
# Return scipy array of values
out = xc * 0.
# Get multiple elevation values in each triangle.
# Process triangles in chunks to reduce function call overhead
lx = len(xc)
lx_div_cs = scipy.ceil(lx * 1. / (1. * chunk_size)).astype(int)
# Crude check that xc/yc are the centroid values
#
erMess = ' Result of make_meanFun can ONLY be applied to a vector' +\
' of ALL centroid coordinates\n' +\
' (since mesh triangles are used to spatially average)'
assert scipy.all(xc == domain.centroid_coordinates[:, 0]), erMess
assert scipy.all(yc == domain.centroid_coordinates[:, 1]), erMess
for i in range(lx_div_cs):
# Evaluate in triangles lb:ub
lb = i * chunk_size
ub = min((i + 1) * chunk_size, lx)
if verbose:
print 'Averaging in triangles ', lb, '-', ub - 1
# Store x,y,triangleIndex
px = scipy.array([])
py = scipy.array([])
p_indices = scipy.array([])
for j in range(lb, ub):
mesh_tri = \
domain.mesh.vertex_coordinates[
range(3 * j, 3 * j + 3), :].tolist()
pts = su.gridPointsInPolygon(
mesh_tri,
approx_grid_spacing=approx_grid_spacing)
px = scipy.hstack([px, pts[:, 0]])
py = scipy.hstack([py, pts[:, 1]])
p_indices = scipy.hstack([p_indices,
scipy.repeat(j, len(pts[:, 0]))])
# Get function values at all px,py
if verbose:
print ' Evaluating function at ', len(px), ' points'
allTopo = q_function(px, py)
# Set output values in lb:ub
for j in range(lb, ub):
out_indices = (p_indices == j).nonzero()[0]
assert len(out_indices) > 0
if(averaging == 'mean'):
out[j] = allTopo[out_indices].mean()
elif(averaging == 'min'):
out[j] = allTopo[out_indices].min()
elif(averaging == 'max'):
out[j] = allTopo[out_indices].max()
elif(averaging == 'harmonic_mean'):
out[j] = 1.0 / (1.0 / allTopo[out_indices]).mean()
else:
raise Exception, 'Unknown value of averaging'
return(out)