def gauss_lsqfit(x,y,pcen):
"""
:param x:
:param y:
:param pcen: An estimate of the Gaussian mean
:return:
"""
from pypit import arcyarc
def gfunc(x,ampl,cent,sigm,cons,tilt):
df = (x[1:]-x[:-1])/2.0
df = np.append(df,df[-1])
dff = (x[1:]**2 - x[:-1]**2)/2.0
dff = np.append(dff,dff[-1])
sqt = sigm*np.sqrt(2.0)
return cons*df*2.0 + tilt*dff + ampl*0.5*np.sqrt(np.pi)*sqt*(erf((x+df-cent)/sqt) - erf((x-df-cent)/sqt))
#return cons + ampl*np.exp(-0.5*((x-cent)/sigm)**2)
if np.any(y<0.0):
return [0.0, 0.0, 0.0], True
# Obtain a quick first guess at the parameters
ampl, cent, sigm, good = arcyarc.fit_gauss(x, y, np.zeros(3,dtype=np.float), 0, x.size, float(pcen))
if good == 0:
return [0.0, 0.0, 0.0], True
elif np.any(np.isnan([ampl, cent, sigm])):
return [0.0, 0.0, 0.0], True
else:
# Perform a least squares fit
try:
popt, pcov = curve_fit(gfunc, x, y, p0=[ampl, cent, sigm, 0.0, 0.0], maxfev=100)
#popt, pcov = curve_fit(gfunc, x, y, p0=[0.0,ampl, cent, sigm], maxfev=100)
except:
return [0.0, 0.0, 0.0], True
return [popt[0], popt[1], popt[2]], False
def gauss_fit(x, y, pcen):
# dx = np.ones(x.size)*np.mean(x[1:]-x[:-1])
# coeffs = polyfit_integral(x, y, dx, 2)
# return poly_to_gauss(coeffs)
from pypit import arcyarc
try:
if np.any(y<0.0):
return [0.0, 0.0, 0.0], True
ampl, cent, sigm, good = arcyarc.fit_gauss(x, y, np.zeros(3,dtype=np.float), 0, x.size, float(pcen))
if good == 0:
return [0.0, 0.0, 0.0], True
elif np.any(np.isnan([ampl, cent, sigm])):
return [0.0, 0.0, 0.0], True
else:
return [ampl, cent, sigm], False
except:
return [0.0, 0.0, 0.0], True
def gauss_fit(x, y, pcen):
# dx = np.ones(x.size)*np.mean(x[1:]-x[:-1])
# coeffs = polyfit_integral(x, y, dx, 2)
# return poly_to_gauss(coeffs)
from pypit import arcyarc
try:
if np.any(y < 0.0):
return [0.0, 0.0, 0.0], True
ampl, cent, sigm, good = arcyarc.fit_gauss(x, y,
np.zeros(3, dtype=np.float),
0, x.size, float(pcen))
if good == 0:
return [0.0, 0.0, 0.0], True
elif np.any(np.isnan([ampl, cent, sigm])):
return [0.0, 0.0, 0.0], True
else:
return [ampl, cent, sigm], False
except:
return [0.0, 0.0, 0.0], True
def gauss_lsqfit(x, y, pcen):
"""
:param x:
:param y:
:param pcen: An estimate of the Gaussian mean
:return:
"""
from pypit import arcyarc
def gfunc(x, ampl, cent, sigm, cons, tilt):
df = (x[1:] - x[:-1]) / 2.0
df = np.append(df, df[-1])
dff = (x[1:]**2 - x[:-1]**2) / 2.0
dff = np.append(dff, dff[-1])
sqt = sigm * np.sqrt(2.0)
return cons * df * 2.0 + tilt * dff + ampl * 0.5 * np.sqrt(
np.pi) * sqt * (erf((x + df - cent) / sqt) - erf(
(x - df - cent) / sqt))
#return cons + ampl*np.exp(-0.5*((x-cent)/sigm)**2)
if np.any(y < 0.0):
return [0.0, 0.0, 0.0], True
# Obtain a quick first guess at the parameters
ampl, cent, sigm, good = arcyarc.fit_gauss(x, y, np.zeros(3,
dtype=np.float),
0, x.size, float(pcen))
if good == 0:
return [0.0, 0.0, 0.0], True
elif np.any(np.isnan([ampl, cent, sigm])):
return [0.0, 0.0, 0.0], True
else:
# Perform a least squares fit
try:
popt, pcov = curve_fit(gfunc,
x,
y,
p0=[ampl, cent, sigm, 0.0, 0.0],
maxfev=100)
#popt, pcov = curve_fit(gfunc, x, y, p0=[0.0,ampl, cent, sigm], maxfev=100)
except:
return [0.0, 0.0, 0.0], True
return [popt[0], popt[1], popt[2]], False
