def __resid(pars, ncoefs=1, knots=None, order=3, irbkg=1, nfft=2048,
kraw=None, mu=None, kout=None, ftwin=1, chi_std=None, **kws):
coefs = [getattr(pars, FMT_COEF % i) for i in range(ncoefs)]
bkg, chi = spline_eval(kraw, mu, knots, coefs, order, kout)
if chi_std is not None:
chi = chi - chi_std
return realimag(xafsft_fast(chi*ftwin, nfft=nfft)[:irbkg])
def apply(self, chi, eps_scale=False, all_kweights=True, **kws):
"""apply transform, returns real/imag components
eps_scale: scale by appropriaat epsilon_k or epsilon_r
"""
# print 'this is transform apply ', len(chi), chi[5:10], kws
for key, val in kws.items():
if key == 'kw': key = 'kweight'
setattr(self, key, val)
all_kweights = all_kweights and isinstance(self.kweight, Iterable)
# print 'fit space = ', self.fitspace
if self.fitspace == 'k':
if all_kweights:
return np.concatenate([chi * self.k_[:len(chi)]**kw for kw in self.kweight])
else:
return chi * self.k_[:len(chi)]**self.kweight
elif self.fitspace in ('r', 'q'):
self.make_karrays()
out = []
if all_kweights:
# print 'Apply -- use all kweights ', self.kweight
chir = [self.fftf(chi, kweight=kw) for kw in self.kweight]
eps_r = self.epsilon_r
else:
chir = [self.fftf(chi)]
eps_r = [self.epsilon_r]
if self.fitspace == 'r':
irmin = int(0.01 + self.rmin/self.rstep)
irmax = min(self.nfft/2, int(1.01 + self.rmax/self.rstep))
for i, chir_ in enumerate(chir):
if eps_scale:
chir_ = chir_ /(eps_r[i])
out.append( realimag(chir_[irmin:irmax]))
else:
chiq = [self.ffti(self.r_, c) for c in chir]
iqmin = int(0.01 + self.kmin/self.kstep)
iqmax = min(self.nfft/2, int(1.01 + self.kmax/self.kstep))
for chiq_ in chiq:
out.append( realimag(chiq[iqmin:iqmax]))
return np.concatenate(out)
def __resid(pars,
knots=None,
order=3,
irbkg=1,
nfft=2048,
kraw=None,
mu=None,
kout=None,
ftwin=1,
**kws):
coefs = [p.value for p in pars.values()]
bkg, chi = spline_eval(kraw, mu, knots, coefs, order, kout)
return realimag(xafsft_fast(chi * ftwin, nfft=nfft)[:irbkg])
def apply(self, k, chi, **kws):
"""apply filter"""
for key, val in kws:
if key == 'kweight': key = 'kw'
setattr(self, key, val)
if self.fitspace == 'k':
return chi * k**self.kw
elif self.fitspace in ('r', 'q'):
k_ = self.kstep * arange(self.nfft, dtype='float64')
r_ = self.rstep * arange(self.nfft, dtype='float64')
chir = self.fft(k_, chi)
if self.fitspace == 'r':
irmin = index_of(r_, self.rmin)
irmax = min(self.nfft/2, 1 + index_of(r_, self.rmax))
return realimag(chir[irmin:irmax])
else:
chiq = self.ifft(r_, chir)
iqmin = index_of(k_, self.kmin)
iqmax = min(self.nfft/2, 1 + index(k_, self.kmax))
return realimag(chiq[ikmin:ikmax])
def apply(self, k, chi, **kws):
"""apply filter"""
for key, val in kws:
if key == 'kweight': key = 'kw'
setattr(self, key, val)
if self.fitspace == 'k':
return chi * k**self.kw
elif self.fitspace in ('r', 'q'):
k_ = self.kstep * arange(self.nfft, dtype='float64')
r_ = self.rstep * arange(self.nfft, dtype='float64')
chir = self.fft(k_, chi)
if self.fitspace == 'r':
irmin = index_of(r_, self.rmin)
irmax = min(self.nfft / 2, 1 + index_of(r_, self.rmax))
return realimag(chir[irmin:irmax])
else:
chiq = self.ifft(r_, chir)
iqmin = index_of(k_, self.kmin)
iqmax = min(self.nfft / 2, 1 + index(k_, self.kmax))
return realimag(chiq[ikmin:ikmax])
def _estimate_noise(self, chi, rmin=15.0, rmax=25.0, all_kweights=True):
"""estimage noice from high r"""
# print 'Estimate Noise!! ', rmin, self.transform.rmin
self.make_karrays()
save = self.rmin, self.rmax, self.fitspace
all_kweights = all_kweights and isinstance(self.kweight, Iterable)
if all_kweights:
chir = [self.fftf(chi, kweight=kw) for kw in self.kweight]
else:
chir = [self.fftf(chi)]
irmin = int(0.01 + rmin/self.rstep)
irmax = min(self.nfft/2, int(1.01 + rmax/self.rstep))
highr = [realimag(chir_[irmin:irmax]) for chir_ in chir]
# get average of window function value, we will scale eps_r scale by this
ikmin = index_of(self.k_, self.kmin)
ikmax = index_of(self.k_, self.kmax)
kwin_ave = self.kwin[ikmin:ikmax].sum()/(ikmax-ikmin)
eps_r = [(sqrt((chi*chi).sum() / len(chi)) / kwin_ave) for chi in highr]
eps_k = []
# use Parseval's theorem to convert epsilon_r to epsilon_k,
# compensating for kweight
if all_kweights:
kweights = self.kweight[:]
else:
kweights = [self.kweight]
for i, kw in enumerate(kweights):
w = 2 * kw + 1
scale = sqrt((2*pi*w)/(self.kstep*(self.kmax**w - self.kmin**w)))
eps_k.append(scale*eps_r[i])
self.rmin, self.rmax, self.fitspace = save
self.n_idp = 2*(self.rmax-self.rmin)*(self.kmax-self.kmin)/pi
self.epsilon_k = eps_k
self.epsilon_r = eps_r
if len(eps_r) == 1:
self.epsilon_k = eps_k[0]
self.epsilon_r = eps_r[0]
def __resid(pars, knots=None, order=3, irbkg=1, nfft=2048,
kraw=None, mu=None, kout=None, ftwin=1, **kws):
coefs = [p.value for p in pars.values()]
bkg, chi = spline_eval(kraw, mu, knots, coefs, order, kout)
return realimag(xafsft_fast(chi*ftwin, nfft=nfft)[:irbkg])
