def _h1_cmp_pdf_(h1, h2, density=False, draw=True, silent=True):
"""Compare histograms by refit of the first with functions,
extracted from the second one
>>> h1 = ... ## the first histo
>>> h2 = ... ## the second histo
>>> r = h1.cmp_fit ( h2 )
>>> if r : print r.Prob()
"""
if density:
h1_ = h1.density()
h2_ = h2.density()
cmp = _h1_cmp_pdf_(h1_, h2_, density=False, draw=draw, silent=silent)
del h1_
del h2_
return cmp
from ostap.fitting.basic import H1D_pdf, Fit1D
from ostap.fitting.background import Bkg_pdf
pdf2 = H1D_pdf('_H1', h2, density=True, silent=True)
model2 = Fit1D(signal=pdf2,
background=Bkg_pdf('_B2', mass=pdf2.mass, power=0, tau=0))
model2.b.fix(0)
from ostap.utils.utils import invisibleCanvas
with invisibleCanvas():
r2, f2 = model2.chi2fitTo(h1,
silent=silent,
draw=draw,
density=False,
sumw2=True)
return r2, f2.chiSquare()
def test_pygauss():
logger.info('Test PyGauss: simple Gaussian signal')
gauss = PyGauss('PyGauss', xvar=mass)
## model
model = Fit1D(signal=gauss, background=None, name='M0')
## r1, f1 = gauss .fitTo ( dataset , draw = True , silent = True , ncpu=1 )
## print (r1)
r2, f2 = model.fitTo(dataset, draw=True, silent=True, ncpu=1)
print(r2)
def test_pygauss2():
## the function
def function(x, m, s):
dx = (x - m) / s
return math.exp(-0.5 * dx * dx) * NORM / s
## construct PDF
gauss = PyGauss2(name='G2', function=function, xvar=mass)
## model
model = Fit1D(signal=gauss, background='p0', name='Q2')
## r1, f1 = gauss .fitTo ( dataset , draw = True , silent = True , ncpu=1 )
## print (r1)
r2, f2 = model.fitTo(dataset, draw=True, silent=True, ncpu=1)
print(r2)
def test_PyPDF() :
"""Test pure python PDF: pyPDF
- For *OLD* PyROOT only!
- see Ostap.Models.PyPdf
"""
logger = getLogger("test_PyPDF")
if not old_PyROOT :
logger.warning("test enabled only for OLD PyROOT!")
return
logger.info ("Test pure python PDF: PyPDF ")
from ostap.fitting.pypdf import PyPDF
# =============================================================================
## @class PyGauss
# local ``pure-python'' PDF
class PyGauss(MASS,PyPDF) :
"""Local ``pure-python'' PDF
"""
def __init__ ( self ,
name ,
xvar ,
mean = ( 3.080 , 3.05 , 3.15 ) ,
sigma = ( 0.010 , 0.005 , 0.020 ) ,
pypdf = None ) :
MASS .__init__ ( self , name , xvar , mean , sigma )
PyPDF.__init__ ( self , self.name , ( self.xvar ,
self.mean ,
self.sigma ) , pypdf = pypdf )
self.pdf = self.pypdf
self.config = {
'name' : self.name ,
'xvar' : self.xvar ,
'mean' : self.mean ,
'sigma' : self.sigma ,
'pypdf' : None ## attention!
}
## the main method
def evaluate ( self ) :
vlist = self.varlist
x = self.variable ( 0 )
m = self.variable ( 1 )
s = self.variable ( 2 )
dx = ( x - m ) / s
return math.exp ( -0.5 * dx * dx ) * NORM / s
with timing ("Using-PyPDF", logger ) :
pdf.mean = random.gauss ( 3.100 , 0.010 )
pdf.sigma = random.gauss ( 0.012 , 0.001 )
## create PDF
gauss = PyGauss( 'PyGauss' , xvar = mass ,
mean = pdf.mean , sigma = pdf.sigma )
## build fit model
model = Fit1D ( signal = gauss , background = None , name = 'M0' )
## fit!
r, _ = model .fitTo ( dataset , draw = False , silent = True , ncpu=1 )
with wait ( 1 ) , use_canvas ( "test_PyPDF" ) :
r, f = model .fitTo ( dataset , draw = True , silent = True , ncpu=1 )
logger.info ("Fit result for``pure python'' PDF: PyPDF \n%s" % r.table ( prefix = "# " ) )
def test_PyPdf_AI() :
"""Test pure python PDF: pyPDF
- For *NEW* PyROOT only!
- see Ostap.Models.PyPdf
"""
logger = getLogger("test_PyPdf_AI")
if old_PyROOT :
logger.warning("test enabled only for NEW PyROOT!")
return
logger.info ("Test pure python PDF: PyPdf with analytical integral")
# =============================================================================
## @class PyGauss
# local ``pure-python'' PDF
class MyGauss(Ostap.Models.PyPdf) :
"""Local ``pure-python'' PDF
"""
def __init__ ( self , name , xvar , mean , sigma ) :
vars = ROOT.RooArgList()
vars.add ( xvar )
vars.add ( mean )
vars.add ( sigma )
super(MyGauss,self).__init__ ( name , 'title' , vars )
## the main method
def evaluate ( self ) :
x = self.variable ( 0 )
m = self.variable ( 1 )
s = self.variable ( 2 )
dx = ( x - m ) / s
return math.exp ( -0.5 * dx * dx ) * NORM / s
def clone ( self , newname ) :
name = newname if newname else self.name
vlist = self.variables()
xvar = vlist[0]
mean = vlist[1]
sigma = vlist[2]
cl = MyGauss( name , xvar , mean , sigma )
ROOT.SetOwnership ( cl , False )
return cl
## declare analytical integral
def get_analytical_integral ( self ) :
"""Declare the analytical integral"""
x = self.variables()[0]
if self.matchArgs ( x ) : return 1 ## get the integration code
return 0
## calculate the analytical integral
def analytical_integral ( self ) :
"""Calculate the analytical integral"""
assert 1 == self.intCode () , 'Invalid integration code!'
vlist = self.variables()
rn = self.rangeName()
xv = vlist [ 0 ]
xmax = xv.getMax ( rn )
xmin = xv.getMin ( rn )
m = float ( vlist [ 1 ] )
s = float ( vlist [ 2 ] )
return CDF ( xmax , m , s ) - CDF ( xmin , m , s )
with timing ("Using-PyPdf+AI", logger ) :
pdf.mean = random.gauss ( 3.100 , 0.010 )
pdf.sigma = random.gauss ( 0.012 , 0.001 )
pdf_ = MyGauss ( 'MyGauss' , pdf.xvar , pdf.mean , pdf.sigma )
gauss = Generic1D_pdf ( pdf_ , xvar = pdf.xvar )
## build fit model
model = Fit1D ( signal = gauss , background = None , name = 'M3' )
## fit!
r, _ = model .fitTo ( dataset , draw = False , silent = True , ncpu=1 )
with wait ( 1 ) , use_canvas ( "test_PyPdf_AI" ) :
r, f = model .fitTo ( dataset , draw = True , silent = True , ncpu=1 )
logger.info ("Fit result for``pure python'' PDF: PyPdf with analytical integral\n%s" % r.table ( prefix = "# " ) )
def test_PyPdf() :
"""Test pure python PDF: pyPDF
- For *NEW* PyROOT only!
- see Ostap.Models.PyPdf
"""
logger = getLogger("test_PyPdf")
if old_PyROOT :
logger.warning("test enabled only for NEW PyROOT!")
return
logger.info ("Test pure python PDF: PyPdf ")
# =============================================================================
## @class PyGauss
# local ``pure-python'' PDF
class MyGauss(Ostap.Models.PyPdf) :
"""Local ``pure-python'' PDF
"""
def __init__ ( self , name , xvar , mean , sigma ) :
vars = ROOT.RooArgList()
vars.add ( xvar )
vars.add ( mean )
vars.add ( sigma )
super(MyGauss,self).__init__ ( name , 'title' , vars )
## the main method
def evaluate ( self ) :
vlist = self.varlist
x = self.variable ( 0 )
m = self.variable ( 1 )
s = self.variable ( 2 )
dx = ( x - m ) / s
return math.exp ( -0.5 * dx * dx ) * NORM / s
def clone ( self , newname ) :
name = newname if newname else self.name
vlist = self.variables()
xvar = vlist[0]
mean = vlist[1]
sigma = vlist[2]
cl = MyGauss( name , xvar , mean , sigma )
ROOT.SetOwnership ( cl , False )
return cl
with timing ("Using-PyPdf", logger ) :
pdf.mean = random.gauss ( 3.100 , 0.010 )
pdf.sigma = random.gauss ( 0.012 , 0.001 )
pdf_ = MyGauss ( 'MyGauss' , pdf.xvar , pdf.mean , pdf.sigma )
gauss = Generic1D_pdf ( pdf_ , xvar = pdf.xvar )
## build fit model
model = Fit1D ( signal = gauss , background = None , name = 'M3' )
## fit!
r, _ = model .fitTo ( dataset , draw = False , silent = True , ncpu=1 )
with wait ( 1 ) , use_canvas ( "test_PyPdf" ) :
r, f = model .fitTo ( dataset , draw = True , silent = True , ncpu=1 )
logger.info ("Fit result for``pure python'' PDF: PyPdf \n%s" % r.table ( prefix = "# " ) )
def test_PyPDF2 () :
"""Test pure python PDF: PyPDF2
- see Ostap.Models.PyPdf2
"""
logger = getLogger("test_PyPDF2")
logger.info ("Test pure python PDF: PyPDF2 with python function")
from ostap.fitting.pypdf import PyPDF2
# =============================================================================
## @class PyGauss2
# local ``pure-python'' PDF
class PyGauss2(MASS,PyPDF2) :
"""Local ``pure-python'' PDF """
def __init__ ( self ,
name ,
function ,
xvar ,
mean = ( 3.080 , 3.05 , 3.15 ) ,
sigma = ( 0.010 , 0.005 , 0.020 ) ,
title = ''
) :
MASS .__init__ ( self , name , xvar , mean , sigma )
PyPDF2.__init__ ( self ,
name = self.name ,
function = function ,
vars = ( self.xvar , self.mean , self.sigma ) )
self.config = {
'name' : self.name ,
'function' : self.function ,
'xvar' : self.xvar ,
'mean' : self.mean ,
'sigma' : self.sigma ,
}
## the function
def function ( x , m , s ) :
dx = ( x - m ) / s
return math.exp ( -0.5 * dx * dx ) * NORM / s
with timing ("Using-PyPDF2", logger ) :
pdf.mean = random.gauss ( 3.100 , 0.010 )
pdf.sigma = random.gauss ( 0.012 , 0.001 )
## construct PDF
gauss = PyGauss2 ( name ='G2' , function = function ,
xvar = mass , mean = pdf.mean , sigma = pdf.sigma )
## model
model = Fit1D ( signal = gauss , background = 'p0' , name = 'M2' )
## fit!
r, _ = model .fitTo ( dataset , draw = False , silent = True , ncpu=1 )
with wait ( 1 ) , use_canvas ( "test_PyPDF2" ) :
r, f = model .fitTo ( dataset , draw = True , silent = True , ncpu=1 )
logger.info ("Fit result for``pure python'' PDF: PyPDF2 with python function \n%s" % r.table ( prefix = "# " ) )
def test_PyPDF_AI() :
"""Test pure python PDF: PyPDF with analytical integral
- For *OLD* PyROOT only!
- see Ostap.Models.PyPdf
"""
logger = getLogger("test_PyPDF_AI")
if not old_PyROOT :
logger.warning("test enabled only for OLD PyROOT!")
return
logger.info ("Test pure python PDF: PyPDF with analytical integral")
from ostap.fitting.pypdf import PyPDF
# =========================================================================
## @class PyGaussAI
# local ``pure-python'' PDF with analytical integrals
class PyGaussAI(MASS,PyPDF) :
"""Local ``pure-python'' PDF with analytical integrals """
def __init__ ( self ,
name ,
xvar ,
mean = ( 3.080 , 3.05 , 3.15 ) ,
sigma = ( 0.010 , 0.005 , 0.020 ) ,
pypdf = None ) :
MASS .__init__ ( self , name , xvar , mean , sigma )
PyPDF.__init__ ( self , self.name , ( self.xvar ,
self.mean ,
self.sigma ) , pypdf = pypdf )
self.pdf = self.pypdf
self.config = {
'name' : self.name ,
'xvar' : self.xvar ,
'mean' : self.mean ,
'sigma' : self.sigma ,
'pypdf' : None ## attention!
}
## the main method
def evaluate ( self ) :
vlist = self.varlist
x = self.variable ( 0 )
m = self.variable ( 1 )
s = self.variable ( 2 )
dx = ( x - m ) / s
return math.exp ( -0.5 * dx * dx ) * NORM / s
## declare analytical integral
def get_analytical_integral ( self ) :
"""Declare the analytical integral"""
x = self.varlist[0]
if self.matchArgs ( x ) : return 1 ## get the integration code
return 0
## calculate the analytical integral
def analytical_integral ( self ) :
"""Calculate the analytical integral"""
assert 1 == self.intCode , 'Invalid integration code!'
vlist = self.varlist
rn = self.rangeName
xv = vlist [ 0 ]
xmax = xv.getMax ( rn )
xmin = xv.getMin ( rn )
m = float ( vlist [ 1 ] )
s = float ( vlist [ 2 ] )
return CDF ( xmax , m , s ) - CDF ( xmin , m , s )
with timing ("Using-PyPDF+AI", logger ) :
pdf.mean = random.gauss ( 3.100 , 0.010 )
pdf.sigma = random.gauss ( 0.012 , 0.001 )
## create PDF
gauss = PyGaussAI( 'PyGaussAI' , xvar = mass ,
mean = pdf.mean , sigma = pdf.sigma )
## build fit model
model = Fit1D ( signal = gauss , background = None , name = 'M1' )
## fit!
r, _ = model .fitTo ( dataset , draw = False , silent = True , ncpu=1 )
with wait ( 1 ) , use_canvas ( "test_PyPDF_AI" ) :
r, f = model .fitTo ( dataset , draw = True , silent = True , ncpu=1 )
logger.info ("Fit result for``pure python'' PDF: PyPDF with analytical integral \n%s" % r.table ( prefix = "# " ) )