def doarcfitfunc(p,xdata,ydata,scidata,model,coeff): par = special_functions.build_coeff(p,coeff) scidata = scidata.astype(scipy.float64) z = special_functions.genfunc(xdata,ydata,par).astype(scipy.float64) z = z.reshape((1,scidata.size)) resample = ndimage.map_coordinates(model,z,output=scipy.float64,cval=-1) diff = (scidata - resample)/scipy.sqrt(abs(resample)) return diff
def myopt2(p, x, z, model, skyx, sky, skymodel, r, o):
par = special_functions.unpack_coeff(p)
coeff, ier = optimize.leastsq(skyarcfit,
par,
(x, z, model, skyx, sky, skymodel, p, r, o),
maxfev=100000)
par = special_functions.build_coeff(coeff, p)
return par
def myoptimize(p, x, z, model, model2=None):
par = special_functions.unpack_coeff(p)
mask = ndimage.maximum_filter(z, 3)
coeff, ier = optimize.leastsq(arcfitfunc,
par, (x, z, model, p, model2, mask),
maxfev=100000)
par = special_functions.build_coeff(coeff, p)
return par
def skyopt(p, x, data, model):
par = special_functions.unpack_coeff(p).tolist()
wave = special_functions.genfunc(x, 0, p).astype(scipy.float64)
sky = interpolate.splev(wave, model).astype(scipy.float64)
ratio = scipy.median(data) / scipy.median(sky)
offset = 0.
par.append(ratio)
par.append(offset)
coeff, ier = optimize.leastsq(skyfit,
par, (x, data, model, p),
maxfev=100000)
return special_functions.build_coeff(coeff, p), coeff[-2], coeff[-1]
def skyfit(p, x, data, model, tmp):
par = special_functions.build_coeff(p[:-2], tmp)
""" Test for increasing function... """
tmp = special_functions.genfunc(x, 0., par).astype(scipy.float32)
diff = signal.convolve(tmp, [1., -1.], mode='same')[10:-10].copy()
if diff[diff < 0].size > 0:
return scipy.ones(x.size) * 1.e9
data = data.astype(scipy.float64)
z = special_functions.genfunc(x, 0, par).astype(scipy.float64)
sky = interpolate.splev(z, model).astype(scipy.float64)
sky *= p[-2]
return (sky - data) / scipy.sqrt(abs(sky))
def arcfitfunc(p, x, data, model, tmp, model2=None, mask=None):
par = special_functions.build_coeff(p, tmp)
""" Test for increasing function... """
tmp = special_functions.genfunc(x, 0., par).astype(scipy.float32)
diff = signal.convolve(tmp, [1., -1.], mode='same')[10:-10].copy()
if diff[diff < 0].size > 0:
return scipy.ones(x.size) * 1.e9
data = data.astype(scipy.float64)
z = special_functions.genfunc(x, 0, par).astype(scipy.float64)
mod = interpolate.splev(z, model).astype(scipy.float64)
if model2 is not None:
mod += interpolate.splev(z, model2).astype(scipy.float64)
mod = mod * data.max() / mod.max() + scipy.median(data)
diff = (data - mod) / scipy.sqrt(abs(mod))
return diff
def skyarcfit(p, x, data, model, skyx, sky, skymodel, tmp, r, o):
par = special_functions.build_coeff(p, tmp)
""" Test for increasing function... """
tmp = special_functions.genfunc(x, 0., par).astype(scipy.float32)
diff = signal.convolve(tmp, [1., -1.], mode='same')[10:-10].copy()
if diff[diff < 0].size > 0:
return scipy.ones(x.size) * 1.e9
data = data.astype(scipy.float64)
z = special_functions.genfunc(skyx, 0, par).astype(scipy.float64)
skymod = interpolate.splev(z, skymodel).astype(scipy.float64)
skymod = (skymod) * r
# skymod = skymod*scipy.median(sky)/scipy.median(skymod)
z = special_functions.genfunc(x, 0, par).astype(scipy.float64)
mod = interpolate.splev(z, model).astype(scipy.float64)
mod = mod * data.max() / mod.max() + scipy.median(data)
diff1 = (data - mod) / scipy.sqrt(abs(mod))
# diff2 = (sky-skymod)/scipy.sqrt(abs(sky))**0.25
diff2 = (sky - skymod)
return scipy.concatenate((diff1, diff2))
def arcfitfunc(p,x,data,scale,model,tmp,model2=None):
par = special_functions.build_coeff(p,tmp)
""" Test for increasing function... """
tmp = par['coeff'][1:].copy()
for i in range(tmp.size):
tmp[i] *= (i+1.)
tmp = {'coeff':tmp.copy(),'type':par['type']}
t = special_functions.genfunc(x,0,par).astype(scipy.float32)
if t[t<0].size>0:
return scipy.ones(x.size)*1.e9
data = data.astype(scipy.float64)
z = special_functions.genfunc(x,0,par).astype(scipy.float64)
mod = interpolate.splev(z,model).astype(scipy.float64)
if model2 is not None:
mod += interpolate.splev(z,model2).astype(scipy.float64)
mod = mod*data.max()/mod.max() + scipy.median(data)
diff = (data-mod)/scipy.sqrt(abs(mod))
return diff
def myoptimize(p,x,z,scale,model,model2=None): par = special_functions.unpack_coeff(p) coeff,ier = optimize.leastsq(arcfitfunc,par,(x,z,scale,model,p,model2),maxfev=100000) par = special_functions.build_coeff(coeff,p) return par
def arcfitfunc2(p,x,y,z,model): par = special_functions.unpack_coeff(p) coeff,ier = optimize.leastsq(doarcfitfunc,par,(x,y,z,model,p),maxfev=100000) return special_functions.build_coeff(coeff,p)