def single_lorentzian(
self,
xdata,
ydata,
parms,
domain,
upper=None,
plot=True,
err=1e-10,
maxruns=int(1e8),
xlabel=r"$Frequency \, (Hz)$",
ylabel=r"$Energy \, (dB)$",
title=r"$Spectrum \, and \, Fit$",
):
from scipy.optimize import curve_fit
import numpy as np
import matplotlib.pyplot as plt
xpeaks, ypeaks = self.choose_domain(xdata, ydata, domain, upper=upper)
yerr = 0.001 * np.ones_like(ypeaks)
def lorentzian(x, c, A, G, w0):
return c + (A / np.pi) * (0.5 * G) / ((x - w0) ** 2 + (0.5 * G) ** 2)
popt, pcov = curve_fit(
lorentzian, xpeaks, ypeaks, parms, check_finite=True, xtol=err, maxfev=maxruns, sigma=yerr
)
chi_square = np.sum((lorentzian(xpeaks, *popt) - ypeaks) ** 2 / (yerr ** 2)) / (len(xpeaks) - len(popt) - 1)
yFit = lorentzian(xpeaks, *popt)
yuFit = lorentzian(xpeaks, *parms)
if plot == True:
plt.figure(figsize=(7, 7))
plt.plot(xdata, ydata, "o", ms=0.3, label=r"$Raw \, Data$")
plt.plot(xpeaks, yFit, linewidth=4, alpha=0.6, label=r"$Fitted \, Curve$")
plt.plot(xpeaks, yuFit, alpha=0.8, label=r"$Guesses$")
plt.xlabel(xlabel)
plt.ylabel(ylabel)
plt.title(title)
plt.legend(loc=0)
if self.save_name is not None:
plt.savefig(self.save_name + ".png")
plt.show()
else:
pass
os.chdir(cwd + "/..")
error_code(code=0)
return popt, chi_square, yFit, yuFit
def pos_multi_lorentzian(self, parms, nres, domain, upper = None, std_dev = 11, plot = True, err = 1e-10, maxruns = int(1e8),
xlabel = r"$Frequency \, (Hz)$", ylabel = r"$Energy \, (dB)$",
title = r"$Spectrum \, and \, Fit$", outside_use = False):
from scipy.optimize import curve_fit
import numpy as np
import matplotlib.pyplot as plt
##Finding the peaks##
xdata = self.data[:,0]
ydata = self.data[:,1]
xpeaks, ypeaks = self.choose_domain(xdata, ydata, domain, upper = upper)
x_peak_coord, y_peak_coord = self.find_peaks(domain, std_dev)
x_peak_coord = x_peak_coord.flatten()
y_peak_coord = y_peak_coord.flatten()
##Now we have the peaks, so we'll put them into the initial guesses##
for i in range(0, len(y_peak_coord)):
parms[1+nres+i] = parms[0]-y_peak_coord[i]
parms[2*nres+1+i] = x_peak_coord[i]
parms = parms.flatten()
##And set the error...##UNDER CONSTRUCTION
yerr = .001*np.ones_like(ypeaks)
##And run the fit##
def multi_lorentz(x, *p):
y = np.zeros_like(x) + p[0]
for i in range(1, len(p)-2*nres):
gamma = p[i]
amp = p[i+nres]
center = p[i+2*nres]
y += np.absolute(amp/(1+1.j*(x-center)/(gamma/2)))
return y.flatten()
popt, pcov = curve_fit(multi_lorentz, xpeaks, ypeaks, parms,
check_finite = True, xtol = err, maxfev = maxruns)
chi_square = np.sum((-multi_lorentz(xpeaks, *popt)+ypeaks)**2/(yerr)**2)/(len(xpeaks) - len(popt) - 1)
yFit = multi_lorentz(xpeaks, *popt)
yuFit = multi_lorentz(xpeaks, *parms)
if plot == True:
plt.figure(figsize = (7,7))
plt.plot(xdata, ydata, 'o', ms = .3, label = r"$Raw \, Data$")
plt.plot(xpeaks, yFit, linewidth = 4, alpha = .6, label = r"$Fitted \, Curve$")
plt.plot(xpeaks, yuFit, alpha = .8, label = r"$Guesses$")
plt.xlabel(xlabel)
plt.ylabel(ylabel)
plt.title(title)
plt.legend(loc = 0) ##include a save_name?
plt.show()
else:
pass
os.chdir(cwd+'/..')
error_code(code = 0)
if outside_use != False:
return {"popt":popt, "chi":chi_square, "yFit":yFit, "yuFit":yuFit, "xpeaks":xpeaks}
else:
return popt, chi_square, yFit, yuFit