def __init__(self,fn=None,pars=None,x0=None,y0=None,r0=None,sigma=None,IMG=None,mask=None):
self.mask = mask
self.IMG = mask
self.pars = Parameters()
self.pars.add('amp', 2,min=0,max=10)
self.pars.add('bg', 1)
self.pars.add('bgexp', 4.,min=1.,max=4.)
self.pars.add('r0', 1500,min=1000,max=2000)#,vary=False)
self.pars.add('sigma', 2,min=.1, max=100)#,vary=False)
self.pars.add('x0',XCEN,min=XCEN-10,max=XCEN+10)
self.pars.add('y0',YCEN,min=YCEN-10,max=YCEN+10)
super(RingFit,self).__init__(fn=fn,pars=pars)
def histfitqrings(histstot, qx, Ix=None, PF=False):
''' fit the hists'''
histspar = np.zeros((histstot.shape[0], 2))
histsparpois = histspar * 0
histsfit = histstot * 0
histsfitpois = histstot * 0
for i in range(histstot.shape[0]):
if (PF):
print("iteration {} or {}".format(i, histstot.shape[0]))
#kbar = np.interp(i,sqx,sqy)
pars = Parameters()
pars.add('M', value=4, vary=True)
if (Ix is not None):
pars.add('kbar', value=Ix[i], vary=False)
else:
pars.add('kbar', value=2., vary=True)
xdata = np.arange(histstot.shape[1])
ydata = histstot[i]
fitfn = FitObject(negbinomialdist, pars)
fitfnpois = FitObject(poissondist, pars)
fitfn.fit(xdata, ydata)
fitfnpois.fit(xdata, ydata)
pars = fitfn.mn.params
parspois = fitfnpois.mn.params
histsfit[i] = fitfn.fn(xdata, pars)
histsfitpois[i] = fitfnpois.fn(xdata, pars)
histspar[i] = np.array([pars['kbar'].value, pars['M'].value])
histsparpois[i] = np.array(
[parspois['kbar'].value, parspois['M'].value])
return histsfit, histspar, histsfitpois, histsparpois
def __init__(self, fn=None, pars=None, IMG=None, mask=None):
self.mask = mask
self.IMG = mask
if (pars is None):
pars = {
'amp': 1,
'x0': 0.,
'y0': 0,
'sigmax': 1.,
'sigmay': 1.,
'const': 0
}
if (hasattr(pars['amp'], "value")):
self.pars = pars
else:
self.pars = Parameters()
for key in pars:
self.pars.add(key, pars[key])
super(Gauss2DFitter, self).__init__(fn=gauss2Dfitfunc, pars=pars)
def twolinefit(xdata,
ydata,
m1=None,
m2=None,
xc=None,
yc=None,
varyslopes=True,
plotwin=None):
''' Two line fit func
fits to two lines with slopes m1 and m2 intersecting at point (xc, yc)
m1= ,m2= : set m1 and m2 initial parameters (not necessary)
Default is 1
varyslopes= True : set to false to fix slopes (need to set m1 and m2 then)
set plotwin to a window number to plot to that window number
'''
if (m1 is None):
m1 = 1
if (m2 is None):
m2 = 1
if (xc is None):
xc = 0
if (yc is None):
yc = 0
pars = Parameters()
pars.add('m1', value=m1, vary=varyslopes)
pars.add('m2', value=m2, vary=varyslopes)
pars.add('xc', value=xc, vary=True)
pars.add('yc', value=yc, vary=True)
fitfn = FitObject(fn=twolinefitfunc, pars=pars)
fitfn.fit(xdata, ydata)
yfit = twolinefitfunc(xdata, fitfn.pars)
if (plotwin is not None):
plt.figure(plotwin)
plt.plot(xdata, ydata, 'ko')
#compute with finer spacing
xfit = np.linspace(np.min(xdata), np.max(xdata))
yfit2 = twolinefitfunc(xfit, fitfn.pars)
plt.plot(xfit, yfit2, 'r')
pars = fitfn.pars
return yfit, pars
def polyfit(xdata,
ydata,
a=None,
b=None,
varyexps=False,
plotwin=None,
positive=False):
''' quick lin fit func
poly fit y = sum(a[i]*x**b[i])
note : assumes that the exponents are not varied. If that's ever necessary,
set varyexps to True.
Also, it is recommended you always set a and b, don't rely on default vals
as they may change.
If plot win set, plot results in a window'''
if (a is None):
a = [0, 1, 1]
if (b is None):
b = [0, 1, 2]
if (positive is True):
minval = 0
else:
minval = None
pars = Parameters()
for i in range(len(a)):
pars.add('a{}'.format(i), value=a[i], vary=True, min=minval)
pars.add('b{}'.format(i), value=b[i], vary=varyexps)
fitfn = FitObject(fn=polyfitfunc, pars=pars)
fitfn.fit(xdata, ydata)
yfit = polyfitfunc(xdata, fitfn.pars)
pars = fitfn.pars
if (plotwin is not None):
plt.figure(plotwin)
plt.plot(xdata, ydata, 'ko')
#compute with finer spacing
xfit = np.linspace(np.min(xdata), np.max(xdata))
yfit2 = polyfitfunc(xfit, fitfn.pars)
plt.plot(xfit, yfit2, 'r')
return yfit, pars
def linfit(xdata,ydata,a=None,b=None,plotwin=None):
''' quick lin fit func
linear fit y = a + b*x
If plot win set, plot results in a window'''
if a is None:
a = 1
if b is None:
b = 1.
pars = Parameters()
pars.add('a' , value= a, vary= True)
pars.add('b' , value = b, vary= True)
fitfn = FitObject(fn=linfitfunc,pars=pars)
fitfn.fit(xdata,ydata)
yfit = linfitfunc(xdata,fitfn.pars)
if(plotwin is not None):
plt.figure(plotwin)
plt.plot(xdata,ydata,'ko')
#compute with finer spacing
xfit = np.linspace(np.min(xdata),np.max(xdata));
yfit2 = linfitfunc(xfit,fitfn.pars)
plt.plot(xfit,yfit2,'r')
return yfit, fitfn.pars
class RingFitObject(FitObject):
'''A class to fit a ring for detector calibration.'''
def __init__(self,fn=None,pars=None,x0=None,y0=None,r0=None,sigma=None,IMG=None,mask=None):
self.mask = mask
self.IMG = mask
self.pars = Parameters()
self.pars.add('amp', 2,min=0,max=10)
self.pars.add('bg', 1)
self.pars.add('bgexp', 4.,min=1.,max=4.)
self.pars.add('r0', 1500,min=1000,max=2000)#,vary=False)
self.pars.add('sigma', 2,min=.1, max=100)#,vary=False)
self.pars.add('x0',XCEN,min=XCEN-10,max=XCEN+10)
self.pars.add('y0',YCEN,min=YCEN-10,max=YCEN+10)
super(RingFit,self).__init__(fn=fn,pars=pars)
#setting parameters (can type rngfitobj.set and hit tab to see them)
def setr0(self,r0,min=None,max=None,vary=None):
self.pars['r0'].value = r0
if(min is not None):
self.pars['r0'].min = min
if(max is not None):
self.pars['r0'].max = max
if(vary is not None):
self.pars['r0'].vary = vary
def setamp(self,amp,min=None,max=None,vary=None):
self.pars['amp'].value = amp
if(min is not None):
self.pars['amp'].min = min
if(max is not None):
self.pars['amp'].max = max
if(vary is not None):
self.pars['amp'].vary = vary
def setx0(self,x0,min=None,max=None,vary=None):
self.pars['x0'].value = x0
if(min is not None):
self.pars['x0'].min = min
if(max is not None):
self.pars['x0'].max = max
if(vary is not None):
self.pars['x0'].vary = vary
def sety0(self,y0,min=None,max=None,vary=None):
self.pars['y0'].value = y0
if(min is not None):
self.pars['y0'].min = min
if(max is not None):
self.pars['y0'].max = max
if(vary is not None):
self.pars['y0'].vary = vary
def setsigma(self,sigma,min=None,max=None,vary=None):
self.pars['sigma'].value = sigma
if(min is not None):
self.pars['sigma'].min = min
if(max is not None):
self.pars['sigma'].max = max
if(vary is not None):
self.pars['sigma'].vary = vary
def setbg(self,bg,min=None,max=None,vary=None):
self.pars['bg'].value = bg
if(min is not None):
self.pars['bg'].min = min
if(max is not None):
self.pars['bg'].max = max
if(vary is not None):
self.pars['bg'].vary = vary
def setbgexp(self,bgexp,min=None,max=None,vary=None):
self.pars['bgexp'].value = bgexp
if(min is not None):
self.pars['bgexp'].min = min
if(max is not None):
self.pars['bgexp'].max = max
if(vary is not None):
self.pars['bgexp'].vary = vary
def setmask(self,mask):
self.mask = mask
self.pixellist = np.where(mask.ravel() == 1)
def setIMG(self,IMG):
self.IMG = IMG
def fit(self,IMG=None,mask=None):
''' Fit the function. If not data is given use previous data.'''
if(IMG is None):
IMG = self.IMG
if(self.IMG is None):
print("Cannot fit, no data in fit object (please supply data)")
return
else:
self.IMG = IMG
if(mask is None):
mask = self.mask
if(self.mask is None):
print("Cannot fit, no data in fit object (please supply data)")
return
else:
self.mask = mask
pixellist = self.pixellist
x = np.arange(IMG.shape[1])
y = np.arange(IMG.shape[0])
X,Y = np.meshgrid(x,y)
xy = np.array([X.ravel()[pixellist],Y.ravel()[pixellist]])
erbs = np.ones(len(pixellist[0]))*1
fitobj.fit(xy,data,erbs=erbs)
super(RingFit,self).fit(xdata,ydata,erbs=wdata)
def getfit(self):
''' Get the current fitted image.'''
pars = self.pars
data = np.zeros(self.IMG.shape)
data.ravel()[pixellist] = self.fn(xy,pars)
data.ravel()[pixellist] = self.fn(xy,fitobj.pars)
return data
def getfitparams(self):
return self.pars
#sample fit to the fit object
#include a fit function
from fit.fitfns import spheresqfunc, sspheresqfunc, savitzky_golay, spherepolygauss
import numpy as np
from fit.FitObject import FitObject, Parameters
import matplotlib.pyplot as plt
plt.figure(0)
#initialize the parameters
pars = Parameters()
pars.add('amp', value=1, vary=False)
pars.add('radius', value=100, vary=True)
pars.add('window', value=21, vary=True)
ndata = 400
xdata = np.linspace(1e-4, .1, ndata)
ydata = spheresqfunc(xdata, pars) + np.random.random(ndata) * .01
erbs = xdata * 0 + .01
fitfn = FitObject(fn=sspheresqfunc, pars=pars)
fitfn.fit(xdata, ydata, erbs)
fitfn.plot(xdata, ydata, logxy=[1, 1])
fitfn.printpars()
plt.figure(1)
#Try a different distribution
pars = Parameters()
pars.add('amp', value=1, vary=True)
pars.add('radius', value=100, vary=True)
pars.add('FWHM', value=10, vary=True)
pars.add('parasitic', value=1e-3, vary=True)
pars.add('bg', value=0, vary=False)
ndata = 400
class Gauss2DFitter(FitObject):
'''A class to fit a Gaussian peak.
'''
def __init__(self, fn=None, pars=None, IMG=None, mask=None):
self.mask = mask
self.IMG = mask
if (pars is None):
pars = {
'amp': 1,
'x0': 0.,
'y0': 0,
'sigmax': 1.,
'sigmay': 1.,
'const': 0
}
if (hasattr(pars['amp'], "value")):
self.pars = pars
else:
self.pars = Parameters()
for key in pars:
self.pars.add(key, pars[key])
super(Gauss2DFitter, self).__init__(fn=gauss2Dfitfunc, pars=pars)
def fit(self, IMG=None, mask=None):
''' Fit the function. If not data is given use previous data.'''
if (IMG is None):
IMG = self.IMG
if (self.IMG is None):
print("Cannot fit, no data in fit object (please supply data)")
return
else:
self.IMG = IMG
if (mask is None):
mask = self.mask
if (self.mask is None):
print("Cannot fit, no data in fit object (please supply data)")
return
else:
self.mask = mask
pixellist = self.pixellist
x = np.arange(IMG.shape[1])
y = np.arange(IMG.shape[0])
X, Y = np.meshgrid(x, y)
xy = np.array([X.ravel()[pixellist], Y.ravel()[pixellist]])
erbs = np.ones(len(pixellist[0])) * 1
fitobj.fit(xy, data, erbs=erbs)
super(Gauss2DFitter, self).fit(xdata, ydata, erbs=wdata)
def setmask(self, mask):
self.mask = mask
self.pixellist = np.where(mask.ravel() == 1)
def setIMG(self, IMG):
self.IMG = IMG
def fit(self, IMG=None, mask=None):
''' Fit the function. If not data is given use previous data.'''
if (IMG is None):
IMG = self.IMG
if (self.IMG is None):
print("Cannot fit, no data in fit object (please supply data)")
return
else:
self.IMG = IMG
if (mask is None):
mask = self.mask
if (self.mask is None):
print("Cannot fit, no data in fit object (please supply data)")
return
else:
self.mask = mask
pixellist = self.pixellist
x = np.arange(IMG.shape[1])
y = np.arange(IMG.shape[0])
X, Y = np.meshgrid(x, y)
xy = np.array([X.ravel()[pixellist], Y.ravel()[pixellist]])
erbs = np.ones(len(pixellist[0])) * 1
fitobj.fit(xy, data, erbs=erbs)
super(RingFit, self).fit(xdata, ydata, erbs=wdata)
def getfit(self):
''' Get the current fitted image.'''
pars = self.pars
data = np.zeros(self.IMG.shape)
data.ravel()[pixellist] = self.fn(xy, pars)
data.ravel()[pixellist] = self.fn(xy, fitobj.pars)
return data
def getfitparams(self):
return self.pars
def fitAgBHring(IMG, mask=None, pars=None, plotwin=None, clims=None):
''' Fit an AgBH (silver behenate) ring
plotwin : if set, plot to the window desired.
pars : a Parameters object of the initial guess Parameters. If not set
the program will ask you for a best guess.
mask : a mask
clims : if specified, these will be the color limits of the image plot windows
default is to take min and max of *masked* image
'''
qAGBH = 2 * np.pi / CONST_AgBHpeak
if (plotwin is not None):
plt.figure(plotwin)
if (mask is None):
mask = np.ones(IMG.shape)
img = IMG * (mask + .1)
plt.imshow(img)
if (clims is None):
plt.clim(np.min(img * mask), np.max(img * mask))
plt.draw()
plt.pause(.001)
pixellist = np.where(mask.ravel() == 1)
print(
"\n\nfitAgBH: An attempt to fit the first silver behenate ring on a 2D area det"
)
print("Note if plotting, the masked region is brightened w.r.t. the rest")
print("Some tips: \n \
1. Try to set mask to only select region near where the ring is\n\
for a better fit\n\
2. Try a fit without the parasitic bg or const scattering first.\n\
")
if (pars is None):
pars = Parameters()
print(
"You did not specify parameters, so I will ask you a few questions"
)
print("Specify the variables in one of two ways, either:")
print(": value, min, max")
print("or")
print(": value")
#it is "raw_input" in python 2x
xcenvals = np.array(
input("beam center x position guess: ").split(",")).astype(float)
pars.add('x0', xcenvals[0], min=xcenvals[0] - 40, max=xcenvals[0] + 40)
if (len(xcenvals) == 3):
pars['x0'].min = xcenvals[1]
pars['x0'].max = xcenvals[2]
ycenvals = np.array(
input("beam center y position guess: ").split(",")).astype(float)
pars.add('y0', ycenvals[0], min=ycenvals[0] - 40, max=ycenvals[0] + 40)
if (len(ycenvals) == 3):
pars['y0'].min = ycenvals[1]
pars['y0'].max = ycenvals[2]
if (plotwin is not None):
print("Plotted estimate of S(q) from given xcen, ycen")
sqx, sqy = circavg(IMG, x0=xcenvals[0], y0=ycenvals[0], mask=mask)
plt.figure(plotwin + 2)
plt.cla()
w = np.where(sqy > 0)
plt.loglog(sqx[w], sqy[w])
ampvals = np.array(
input("amplitude of ring: ").split(",")).astype(float)
pars.add('amp', ampvals[0], min=0, max=1e9)
if (len(ampvals) == 3):
pars['amp'].min = ampvals[1]
pars['amp'].max = ampvals[1]
ringvals = np.array(
input("Ring location (in pixels from center of beam, approx):").
split(",")).astype(float)
pars.add('r0', ringvals[0], min=0, max=4 * ringvals[0])
if (len(ringvals) == 3):
pars['r0'].min = ringvals[1]
pars['r0'].max = ringvals[2]
sigmavals = np.array(
input("sigma of ring (FWHM for Lorentzian) :").split(",")).astype(
float)
pars.add('sigma', sigmavals[0], min=.1, max=100) #,vary=False)
if (len(sigmavals) == 3):
pars['sigma'].min = sigmavals[1]
pars['sigma'].max = sigmavals[2]
bgvals = input("1/q^4 background (enter nothing to not vary): ")
if (len(bgvals) == 0):
pars.add('bg', 0, vary=False)
pars.add('bgexp', 4., vary=False)
else:
bgvals = np.array(bgvals.split(",")).astype(float)
pars.add('bg', bgvals[0], vary=False)
pars.add('bgexp', 4., vary=True, min=1, max=4)
if (len(bgvals) == 3):
pars['bg'].min = bgvals[1]
pars['bgexp'].max = bgvals[2]
constvals = input("Constant background (enter nothing to not vary):")
if (len(constvals) == 0):
pars.add('const', 0, vary=False)
else:
constvals = np.array(constvals.split(",")).astype(float)
pars.add("const", constvals[0])
if (len(constvals) == 3):
pars['const'].min = constvals[1]
pars['const'].max = constvals[2]
data = IMG.ravel()[pixellist]
fitobj = FitObject(fn=ring2Dlorentzfitfunc, pars=pars)
x = np.arange(IMG.shape[1])
y = np.arange(IMG.shape[0])
X, Y = np.meshgrid(x, y)
xy = np.array([X.ravel()[pixellist], Y.ravel()[pixellist]])
erbs = np.ones(len(pixellist[0]))
datatest = np.zeros(IMG.shape)
fitobj.fit(xy, data, erbs=erbs)
datatest.ravel()[pixellist] = fitobj.fn(xy, fitobj.pars)
XCENFIT = fitobj.pars['x0'].value
YCENFIT = fitobj.pars['y0'].value
RFIT = fitobj.pars['r0'].value
SIGMAFIT = fitobj.pars['sigma'].value
CONSTFIT = fitobj.pars['const'].value
BGFIT = fitobj.pars['bg'].value
BGEXPFIT = fitobj.pars['bgexp'].value
print("x center fit: {:3.2f}; y center fit: {:3.2f}; radius fit: {:3.2f}".
format(XCENFIT, YCENFIT, RFIT))
print(
"ring FWHM : {:3.2f}; const background: {:3.2f}; parasitic bg: {:3.2f}/q^{:3.2f}"
.format(SIGMAFIT, CONSTFIT, BGFIT, BGEXPFIT))
sqd_x, sqd_y = circavg(IMG, x0=XCENFIT, y0=YCENFIT, mask=mask)
sqf_x, sqf_y = circavg(datatest, x0=XCENFIT, y0=YCENFIT, mask=mask)
if (plotwin is not None):
plt.figure(plotwin)
plt.cla()
plt.imshow(IMG * mask)
if (clims is None):
plt.clim(np.min(img * mask), np.max(img * mask))
plcirc([XCENFIT, YCENFIT], RFIT, 'b')
#plt.clim(np.min(IMG*mask),np.max(IMG*mask))
plt.figure(plotwin + 1)
plt.cla()
plt.imshow(datatest * mask)
if (clims is None):
plt.clim(np.min(img * mask), np.max(img * mask))
#plt.clim(np.min(IMG*mask),np.max(IMG*mask))
plcirc([XCENFIT, YCENFIT], RFIT, 'b')
plt.figure(plotwin + 2)
plt.cla()
plt.loglog(sqd_x, sqd_y)
plt.loglog(sqf_x, sqf_y, 'r')
plt.gca().autoscale_view('tight')
plt.draw()
plt.pause(0.001)
resp = input("Get det distance?(y for yes, anything else for no): ")
if (resp == 'y'):
wv = float(input("wavelength (in angs): "))
dpix = float(input("pix size (in m): "))
L, err = getdistfrompeak(qAGBH, wv, dpix, RFIT)
print(
"Your length is approximately {:4.2f} m, and error in calc (from Ewald curvature) approx {:4.2f}%"
.format(L, err))
print("done")
def __init__(self, pars=None, xdata=None, ydata=None):
#default is to plot in logxy 1,1
self.logxy = [1, 1]
self.xdata = xdata
self.ydata = ydata
self.wdata = None
if (pars is None):
pars = Parameters()
pars.add("amp", value=1., vary=True)
pars.add("rbar", value=500., vary=True)
pars.add("z", value=100., vary=True)
pars.add("parasitic", value=0., vary=False)
pars.add("bg", value=0., vary=False)
else:
#If it's not in the Parameters object format, convert it
if (not hasattr(pars["amp"], "value")):
pars1 = pars
pars = Parameters()
for key in pars1:
pars.add(key, value=pars1[key])
super(SchultzFitObject, self).__init__(fn=sqsphereschultz, pars=pars)