def test_numarray_cvx_conversions():
_1_10_ = numarray.array( range(1,11) )
test_cases = [
_1_10_,
numarray.array([ _1_10_ ]),
numarray.transpose( numarray.array([ _1_10_ ]) ),
numarray.array([ _1_10_, range(11, 21) ])
]
for t, a in enumerate(test_cases):
print "-"*80, '\nTest Case %d\n'%(t+1), "-"*80
print "NumArray matrix shaped", a.shape
print a
a_cvx = numarray_to_cvx(a)
print "CVX matrix shaped", a_cvx.size
print a_cvx
a = numarray_asmatrix(a)
a_numarray = cvx_to_numarray(a_cvx)
m, n = a.shape
assert (m,n)==a_cvx.size, \
"(m,n) = (%d,%d) != %s = a_cvx.size"%(m, n, a_cvx.size)
assert (m,n)==a_numarray.shape, \
"(m,n) = (%d,%d) != %s = a_numarray.size"%(m, n, a_numarray.shape)
for i in range(m):
for j in range(n):
assert a[i,j] == a_cvx[i,j], "a[%d,%d] = %f != %f = a_cvx[%d,%d]"%(
i, j, a[i,j], a_cvx[i,j], i, j )
assert a[i,j] == a_numarray[i,j], "a[%d,%d] = %f != %f = a_numarray[%d,%d]"%(
i, j, a[i,j], a_numarray[i,j], i, j )
print
def test_numarray_cvx_conversions():
_1_10_ = numarray.array(range(1, 11))
test_cases = [
_1_10_,
numarray.array([_1_10_]),
numarray.transpose(numarray.array([_1_10_])),
numarray.array([_1_10_, range(11, 21)])
]
for t, a in enumerate(test_cases):
print "-" * 80, '\nTest Case %d\n' % (t + 1), "-" * 80
print "NumArray matrix shaped", a.shape
print a
a_cvx = numarray_to_cvx(a)
print "CVX matrix shaped", a_cvx.size
print a_cvx
a = numarray_asmatrix(a)
a_numarray = cvx_to_numarray(a_cvx)
m, n = a.shape
assert (m,n)==a_cvx.size, \
"(m,n) = (%d,%d) != %s = a_cvx.size"%(m, n, a_cvx.size)
assert (m,n)==a_numarray.shape, \
"(m,n) = (%d,%d) != %s = a_numarray.size"%(m, n, a_numarray.shape)
for i in range(m):
for j in range(n):
assert a[i, j] == a_cvx[
i, j], "a[%d,%d] = %f != %f = a_cvx[%d,%d]" % (
i, j, a[i, j], a_cvx[i, j], i, j)
assert a[i, j] == a_numarray[
i, j], "a[%d,%d] = %f != %f = a_numarray[%d,%d]" % (
i, j, a[i, j], a_numarray[i, j], i, j)
print
def regular_xyval_to_2d_grid_values(xyval):
"""Returns (grid_values, x0, y0, deltax, deltay)"""
xyval = numarray.array(xyval)
n = len(xyval)
x = xyval[:, 0]
y = xyval[:, 1]
values = xyval[:, 2:].copy()
# print "type(values)",type(values)
valsize = numarray.size(values, 1)
x0 = x[0]
y0 = y[0]
k = 1
if x[1] == x0:
deltay = y[1] - y[0]
while x[k] == x0:
k = k + 1
deltax = x[k] - x0
ny = k
nx = n // ny
# print 'A) nx,ny:',nx,ny
values.shape = (nx, ny, valsize)
# print "A type(values)",type(values)
values = numarray.transpose(values, (1, 0, 2))
# print "B type(values)",type(values)
elif y[1] == y0:
deltax = x[1] - x[0]
while y[k] == y0:
k = k + 1
deltay = y[k] - y0
nx = k
ny = n // nx
# print 'B) nx,ny:',nx,ny
values.shape = (ny, nx, valsize)
# print "C type(values)",type(values)
values = numarray.transpose(values, (1, 0, 2))
# print "D type(values)",type(values)
else:
raise ValueError(
"Strange: x[1]!=x0 and y[1]!=y0 this doesn't look like a regular grid..."
)
print 'In regular_xyval_to_2d_grid_values: ', type(xyval), type(values)
return values, x0, y0, deltax, deltay
def regular_xyval_to_2d_grid_values(xyval):
"""Returns (grid_values, x0, y0, deltax, deltay)"""
xyval = numarray.array(xyval)
n = len(xyval)
x = xyval[:,0]
y = xyval[:,1]
values = xyval[:,2:].copy()
# print "type(values)",type(values)
valsize = numarray.size(values,1)
x0 = x[0]
y0 = y[0]
k = 1
if x[1]==x0:
deltay = y[1]-y[0]
while x[k]==x0:
k = k+1
deltax = x[k]-x0
ny = k
nx = n // ny
# print 'A) nx,ny:',nx,ny
values.shape = (nx,ny,valsize)
# print "A type(values)",type(values)
values = numarray.transpose(values,(1,0,2))
# print "B type(values)",type(values)
elif y[1]==y0:
deltax = x[1]-x[0]
while y[k]==y0:
k = k+1
deltay = y[k]-y0
nx = k
ny = n // nx
# print 'B) nx,ny:',nx,ny
values.shape = (ny,nx,valsize)
# print "C type(values)",type(values)
values = numarray.transpose(values,(1,0,2))
# print "D type(values)",type(values)
else:
raise ValueError("Strange: x[1]!=x0 and y[1]!=y0 this doesn't look like a regular grid...")
print 'In regular_xyval_to_2d_grid_values: ', type(xyval), type(values)
return values, x0, y0, deltax, deltay
def update(self, arr):
"""Update the accumulators of the StatsCollector given a complete matrix;
assume that all observations have the same weight.
Properly handle missing values.
"""
assert self.width() == arr.shape[1] #shape(arr)[1]
i = 0
## Update number of elements counters
(length, width) = arr.shape #shape(arr)
initial_n = self.n.copy(
) #self.n[:] # Keep old n for argmin/argmax
n = zeros(width) + length
missings = isnan(arr)
nnan = sum(missings, 0)
self.n += n
self.nnan += nnan
self.nnonnan += n - nnan
## Create masked version of arr and update accumulators
ma = masked_array(arr, mask=missings) # Here, mask missings only
arr_nomissings = arr[~normal_sometrue(missings,
1)] # Here, strip missing rows
self.sum = self.sum + sum(ma, 0) # += does not work...
self.sum_ssq = self.sum_ssq + sum(ma * ma, 0) # += does not work...
self.sum_xxt = self.sum_xxt + matrixmultiply(transpose(arr_nomissings),
arr_nomissings)
self.sum_nomi = self.sum_nomi + sum(arr_nomissings, 0)
self.nxxt += arr_nomissings.shape[0] #shape(arr_nomissings)[0]
## Update (arg)min / make sure old argmin is kept if not updated
ma_argmin = argmin(ma, 0)
ma_min = ma[ma_argmin, range(width)]
min_newpos = argmin(array([self.min, ma_min]), 0).astype('Bool')
self.min[min_newpos] = ma_min[min_newpos]
# XXX Argmin computation needs to be revised! Does not work, at least
# when passing array of shape (1,1).
self.argmin[min_newpos] = ma_argmin[min_newpos] + initial_n[min_newpos]
## Update (arg)max / make sure old argmax is kept if not updated
ma_argmax = argmax(ma, 0)
ma_max = ma[ma_argmax, range(width)]
max_newpos = argmax(array([self.max, ma_max]), 0).astype('Bool')
self.max[max_newpos] = ma_max[max_newpos]
# XXX Argmax computation needs to be revised! Does not work, at least
# when passing array of shape (1,1). Also, is the use of min_newpos
# correct?
self.argmax[max_newpos] = ma_argmax[max_newpos] + initial_n[min_newpos]
def numarray_to_cvx(*args):
"""Returns CVX matrices from NumArray matrices
Note that NumArray one-dimensional arrays are transformed as row vectors
in CVX. See/Run test_numarray_cvx_conversions() for details.
"""
matrices = []
for a in args:
a = numarray_asmatrix(a)
m, n = a.shape
a = numarray.ravel( numarray.transpose(a) )
matrices.append( cvx.matrix(a, (m,n), 'd') )
if len(matrices)==1:
return matrices[0]
return matrices
def numarray_to_cvx(*args):
"""Returns CVX matrices from NumArray matrices
Note that NumArray one-dimensional arrays are transformed as row vectors
in CVX. See/Run test_numarray_cvx_conversions() for details.
"""
matrices = []
for a in args:
a = numarray_asmatrix(a)
m, n = a.shape
a = numarray.ravel(numarray.transpose(a))
matrices.append(cvx.matrix(a, (m, n), 'd'))
if len(matrices) == 1:
return matrices[0]
return matrices
def update(self, arr):
"""Update the accumulators of the StatsCollector given a complete matrix;
assume that all observations have the same weight.
Properly handle missing values.
"""
assert self.width() == arr.shape[1] #shape(arr)[1]
i = 0
## Update number of elements counters
(length,width)= arr.shape #shape(arr)
initial_n = self.n.copy()#self.n[:] # Keep old n for argmin/argmax
n = zeros(width) + length
missings = isnan(arr)
nnan = sum(missings,0)
self.n += n
self.nnan += nnan
self.nnonnan += n - nnan
## Create masked version of arr and update accumulators
ma = masked_array(arr, mask=missings) # Here, mask missings only
arr_nomissings = arr[~normal_sometrue(missings,1)] # Here, strip missing rows
self.sum = self.sum + sum(ma,0) # += does not work...
self.sum_ssq = self.sum_ssq + sum(ma*ma,0) # += does not work...
self.sum_xxt = self.sum_xxt + matrixmultiply(transpose(arr_nomissings),
arr_nomissings)
self.sum_nomi= self.sum_nomi + sum(arr_nomissings,0)
self.nxxt += arr_nomissings.shape[0] #shape(arr_nomissings)[0]
## Update (arg)min / make sure old argmin is kept if not updated
ma_argmin = argmin(ma,0)
ma_min = ma[ma_argmin, range(width)]
min_newpos = argmin(array([self.min, ma_min]), 0).astype('Bool')
self.min[min_newpos] = ma_min[min_newpos]
# XXX Argmin computation needs to be revised! Does not work, at least
# when passing array of shape (1,1).
self.argmin[min_newpos] = ma_argmin[min_newpos] + initial_n[min_newpos]
## Update (arg)max / make sure old argmax is kept if not updated
ma_argmax = argmax(ma,0)
ma_max = ma[ma_argmax, range(width)]
max_newpos = argmax(array([self.max, ma_max]), 0).astype('Bool')
self.max[max_newpos] = ma_max[max_newpos]
# XXX Argmax computation needs to be revised! Does not work, at least
# when passing array of shape (1,1). Also, is the use of min_newpos
# correct?
self.argmax[max_newpos] = ma_argmax[max_newpos] + initial_n[min_newpos]
