def chi2PolarFunc(params,nmax,im,nm):
"""Function which is to be minimized in the chi^2 analysis for Polar shapelets
params = [beta, xc, yc]
beta: characteristic size of shapelets, fit parameter
xc: x centroid of shapelets, fit parameter
yc: y centroid of shapelets, fit parameter
nmax: number of coefficents to use in the Laguerre polynomials
im: observed image
nm: noise map
"""
import sys
beta=params[0]
xc=params[1]
yc=params[2]
if beta<0.:
print 'warning: beta going negative, setting to 0.0'
beta=0.
print 'beta: %f\txc: (%f,%f)'%(beta,xc,yc)
size=im.shape
r,th=shapelet.polarArray([xc,yc],size)
bvals=fgenPolarBasisMatrix(beta,nmax,r,th)
coeffs=solveCoeffs(bvals,im)
mdl=np.abs(img.constructModel(bvals,coeffs,[xc,yc],size))
return np.sum((im-mdl)**2 / nm**2)/(size[0]*size[1])
def chi2nmaxPolarFunc(params,im,nm,beta,xc):
"""
params = [nmax]
nmax: number of coefficents
im: observed image
nm: noise map
beta: fit beta value
xc: fit centroid position
"""
print params
nmax=params
size=im.shape
r,th=shapelet.polarArray(xc,size)
bvals=genPolarBasisMatrix(beta,nmax,r,th)
coeffs=solveCoeffs(bvals,im)
mdl=np.abs(img.constructModel(bvals,coeffs,xc,size))
return np.sum((im-mdl)**2 / nm**2)/(size[0]*size[1])
def chi2nmaxPolarFunc(params,im,nm,beta0,beta1,phi,xc):
"""
params = [nmax]
nmax: number of coefficents
im: observed image
nm: noise map
beta0: fit beta value
beta1: fit beta value
phi: rotation angle
xc: fit centroid position
"""
#nmax=params[0]
nmax=[params, params]
size=im.shape
r,th=shapelet.polarArray(xc,size)
bvals=genPolarBasisMatrix([beta0,beta1],nmax,phi,r,th)
coeffs=solveCoeffs(bvals,im)
mdl=np.abs(img.constructModel(bvals,coeffs,size))
return np.sum((im-mdl)**2 / nm**2)/(size[0]*size[1])
def chi2nmaxPolarFunc(params, im, nm, beta0, beta1, phi, xc):
"""
params = [nmax]
nmax: number of coefficents
im: observed image
nm: noise map
beta0: fit beta value
beta1: fit beta value
phi: rotation angle
xc: fit centroid position
"""
#nmax=params[0]
nmax = [params, params]
size = im.shape
r, th = shapelet.polarArray(xc, size)
bvals = genPolarBasisMatrix([beta0, beta1], nmax, phi, r, th)
coeffs = solveCoeffs(bvals, im)
mdl = np.abs(img.constructModel(bvals, coeffs, size))
return np.sum((im - mdl)**2 / nm**2) / (size[0] * size[1])
def chi2PolarFunc(params,nmax,im,nm,order=['beta0','beta1','phi','yc','xc'],set_beta=[None,None],set_phi=None,set_xc=[None,None],r=None,th=None):
"""Function which is to be minimized in the chi^2 analysis for Polar shapelets
params = [beta0, beta1, phi, xc, yc] or some subset
beta0: characteristic size of shapelets, fit parameter
beta1: characteristic size of shapelets, fit parameter
phi: rotation angle of shapelets, fit parameter
yc: y centroid of shapelets, fit parameter
xc: x centroid of shapelets, fit parameter
nmax: number of coefficents to use in the Laguerre polynomials
im: observed image
nm: noise map
order: order of parameters
fixed parameters: set_beta, set_phi, set_xc
r: radius from centroid, array of im.shape, not required if xc and yc being fit
th: angle from centroid, array of im.shape, not required if xc and yc being fit
"""
#determine which parameters are being fit for, and which are not
beta0=set_beta[0]
beta1=set_beta[1]
phi=set_phi
yc=set_xc[0]
xc=set_xc[1]
fitParams={'beta':False,'phi':False,'xc':False}
for pid,paramName in enumerate(order):
if paramName=='beta0':
beta0=params[pid]
fitParams['beta']=True
elif paramName=='beta1':
beta1=params[pid]
fitParams['beta']=True
elif paramName=='phi':
phi=params[pid]
fitParams['phi']=True
elif paramName=='xc':
xc=params[pid]
fitParams['xc']=True
elif paramName=='yc':
yc=params[pid]
fitParams['xc']=True
#if beta0<0.:
# print 'warning: beta going negative, setting to 0.0'
# beta0=0.
#if beta1<0.:
# print 'warning: beta going negative, setting to 0.0'
# beta1=0.
if beta0<0.:
print 'warning: beta going negative, taking absolute value'
beta0 = np.abs(beta0)
if beta1<0.:
print 'warning: beta going negative, taking absolute value'
beta1 = np.abs(beta1)
print 'beta: (%f,%f)\t phi: %f\txc: (%f,%f)'%(beta0,beta1,phi,xc,yc)
#update noise map
nm=img.makeNoiseMap(nm.shape,np.mean(nm),np.std(nm))
size=im.shape
if fitParams['xc'] or r is None:
r,th=shapelet.polarArray([yc,xc],size) #the radius,theta pairs need to updated if fitting for the xc centre or if not using the r,th inputs
bvals=genPolarBasisMatrix([beta0,beta1],nmax,phi,r,th)
coeffs=solveCoeffs(bvals,im)
mdl=np.abs(img.constructModel(bvals,coeffs,size))
return np.sum((im-mdl)**2 / nm**2)/(size[0]*size[1])
try:
beta0,phi0=initBetaPhi(subim,mode='basic')
print beta0, phi0
beta0,phi0=initBetaPhi(subim,mode='fit')
print beta0, phi0
except:
print 'Test failed (%i):'%tc, sys.exc_info()[0]
te+=1
#genPolarBasisMatrix(beta,nmax,r,th):
#solveCoeffs(m,im):
tc+=1
try:
beta0,phi0=initBetaPhi(subim,mode='fit')
xc=img.maxPos(subim)
r0,th0=shapelet.polarArray(xc,subim.shape)
mPolar=genPolarBasisMatrix(beta0,5,0.,r0,th0)
coeffs=solveCoeffs(mPolar,subim)
print coeffs
if write_files: fileio.writeLageurreCoeffs('testLageurre.pkl',coeffs,xc,subim.shape,beta0,0.,[5,5],pos=[hdr['ra'],hdr['dec'],hdr['dra'],hdr['ddec']],info='Test Lageurre coeff file')
except:
print 'Test failed (%i):'%tc, sys.exc_info()[0]
te+=1
#genBasisMatrix(beta,nmax,rx,ry):
#solveCoeffs(m,im):
tc+=1
try:
beta0,phi0=initBetaPhi(subim,mode='fit')
xc=img.maxPos(subim)
rx=np.array(range(0,subim.shape[0]),dtype=float)-xc[0]
from time import time
import cshapelet as csh
import fshapelet as fsh
import shapelet as sh
import decomp
import cProfile, pstats
USE_PROFILER = True
n = 16
m = 2
PB_DIM = (1000, 500)
xc = (800, 1300)
beta0 = 100.
mask = None
r, th = sh.polarArray(xc, PB_DIM)
print 'Problem dim:', PB_DIM
backend_list = {'original': decomp,
'opt + numexpr': fsh,
'opt + cython': csh
}
pr = {}
for backend, mod in backend_list.iteritems():
if USE_PROFILER:
pr[backend] = cProfile.Profile()
pr[backend].enable()
t = time()
res0 = mod.genPolarBasisMatrix(beta0, n, r, th),
def chi2PolarFunc(params,
nmax,
im,
nm,
order=['beta0', 'beta1', 'phi', 'yc', 'xc'],
set_beta=[None, None],
set_phi=None,
set_xc=[None, None],
r=None,
th=None):
"""Function which is to be minimized in the chi^2 analysis for Polar shapelets
params = [beta0, beta1, phi, xc, yc] or some subset
beta0: characteristic size of shapelets, fit parameter
beta1: characteristic size of shapelets, fit parameter
phi: rotation angle of shapelets, fit parameter
yc: y centroid of shapelets, fit parameter
xc: x centroid of shapelets, fit parameter
nmax: number of coefficents to use in the Laguerre polynomials
im: observed image
nm: noise map
order: order of parameters
fixed parameters: set_beta, set_phi, set_xc
r: radius from centroid, array of im.shape, not required if xc and yc being fit
th: angle from centroid, array of im.shape, not required if xc and yc being fit
"""
#determine which parameters are being fit for, and which are not
beta0 = set_beta[0]
beta1 = set_beta[1]
phi = set_phi
yc = set_xc[0]
xc = set_xc[1]
fitParams = {'beta': False, 'phi': False, 'xc': False}
for pid, paramName in enumerate(order):
if paramName == 'beta0':
beta0 = params[pid]
fitParams['beta'] = True
elif paramName == 'beta1':
beta1 = params[pid]
fitParams['beta'] = True
elif paramName == 'phi':
phi = params[pid]
fitParams['phi'] = True
elif paramName == 'xc':
xc = params[pid]
fitParams['xc'] = True
elif paramName == 'yc':
yc = params[pid]
fitParams['xc'] = True
#if beta0<0.:
# print 'warning: beta going negative, setting to 0.0'
# beta0=0.
#if beta1<0.:
# print 'warning: beta going negative, setting to 0.0'
# beta1=0.
if beta0 < 0.:
print 'warning: beta going negative, taking absolute value'
beta0 = np.abs(beta0)
if beta1 < 0.:
print 'warning: beta going negative, taking absolute value'
beta1 = np.abs(beta1)
print 'beta: (%f,%f)\t phi: %f\txc: (%f,%f)' % (beta0, beta1, phi, xc, yc)
#update noise map
nm = img.makeNoiseMap(nm.shape, np.mean(nm), np.std(nm))
size = im.shape
if fitParams['xc'] or r is None:
r, th = shapelet.polarArray(
[yc, xc], size
) #the radius,theta pairs need to updated if fitting for the xc centre or if not using the r,th inputs
bvals = genPolarBasisMatrix([beta0, beta1], nmax, phi, r, th)
coeffs = solveCoeffs(bvals, im)
mdl = np.abs(img.constructModel(bvals, coeffs, size))
return np.sum((im - mdl)**2 / nm**2) / (size[0] * size[1])
try:
beta0, phi0 = initBetaPhi(subim, mode='basic')
print beta0, phi0
beta0, phi0 = initBetaPhi(subim, mode='fit')
print beta0, phi0
except:
print 'Test failed (%i):' % tc, sys.exc_info()[0]
te += 1
#genPolarBasisMatrix(beta,nmax,r,th):
#solveCoeffs(m,im):
tc += 1
try:
beta0, phi0 = initBetaPhi(subim, mode='fit')
xc = img.maxPos(subim)
r0, th0 = shapelet.polarArray(xc, subim.shape)
mPolar = genPolarBasisMatrix(beta0, 5, 0., r0, th0)
coeffs = solveCoeffs(mPolar, subim)
print coeffs
if write_files:
fileio.writeLageurreCoeffs(
'testLageurre.pkl',
coeffs,
xc,
subim.shape,
beta0,
0., [5, 5],
pos=[hdr['ra'], hdr['dec'], hdr['dra'], hdr['ddec']],
info='Test Lageurre coeff file')
except:
print 'Test failed (%i):' % tc, sys.exc_info()[0]
from time import time
import cshapelet as csh
import fshapelet as fsh
import shapelet as sh
import decomp
import cProfile, pstats
USE_PROFILER = True
n = 16
m = 2
PB_DIM = (1000, 500)
xc = (800, 1300)
beta0 = 100.
mask = None
r, th = sh.polarArray(xc, PB_DIM)
print('Problem dim:', PB_DIM)
backend_list = {'original': decomp, 'opt + numexpr': fsh, 'opt + cython': csh}
pr = {}
for backend, mod in backend_list.items():
if USE_PROFILER:
pr[backend] = cProfile.Profile()
pr[backend].enable()
t = time()
res0 = mod.genPolarBasisMatrix(beta0, n, r, th),
t = time() - t
if USE_PROFILER:
pr[backend].disable()
def test_polarArray():
xc = (100., 200.)
res0 = fsh.polarArray(xc, PB_DIM)
res1 = sh.polarArray(xc, PB_DIM)
yield assert_allclose, res0, res1, 1e-7, 1e-9
def setup():
global r, th, xc, beta, PB_DIM
xc = (100, 130)
beta = 100.
PB_DIM = (300, 200)
r, th = sh.polarArray(xc, PB_DIM)
