def __init__(self, react='sidis'):
if react == 'sidis':
self.reaction = 'sidis'
self.tabs = conf['sidis tabs']
elif react == 'sidisEIC':
self.reaction = 'sidisEIC'
self.tabs = conf['sidisEIC tabs']
elif react == 'sidisSoLID':
self.reaction = 'sidisSoLID'
self.tabs = conf['sidisSoLID tabs']
self.dis_stfuncs = DIS_STFUNCS()
self.setup()
def __init__(self):
self.reaction = 'sidis'
self.tabs = conf['sidis tabs']
self.dis_stfuncs = DIS_STFUNCS()
self.setup()
class RESIDUALS(_RESIDUALS):
def __init__(self):
self.reaction = 'sidis'
self.tabs = conf['sidis tabs']
self.dis_stfuncs = DIS_STFUNCS()
self.setup()
def _get_theory(self, entry):
k, i = entry
x = self.tabs[k]['x'][i]
y = self.tabs[k]['y'][i]
z = self.tabs[k]['z'][i]
Q2 = self.tabs[k]['Q2'][i]
pT = self.tabs[k]['pT'][i]
exp = self.tabs[k]['value'][i]
tar = self.tabs[k]['target'][i]
had = self.tabs[k]['hadron'][i]
obs = self.tabs[k]['obs'][i].strip()
col = self.tabs[k]['col'][i].strip().upper()
if tar == 'proton': tar = 'p'
elif tar == 'neutron': tar = 'n'
elif tar == 'deuteron': tar = 'd'
if obs == 'FUU':
thy = upol.get_FUU(x, z, Q2, pT, tar, had)
elif obs == 'M':
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
F2 = self.dis_stfuncs.get_F2(x, Q2, tar)
thy = FUU / F2
if col == 'HERMES': thy *= 2 * np.pi * pT
elif obs == 'AUTcollins':
# convention factor
coeff = 1.
if col == 'HERMES': coeff = 1 # hermes is sin(phi_s+phi_h)
elif col == 'COMPASS': coeff = -1 # compass is sin(phi_s+phi_h+pi)
# add depolarization factor
if col == 'HERMES': coeff *= 2 * (1 - y) / (1 + (1 - y)**2)
FUT = collins.get_FUT(x, z, Q2, pT, tar, had)
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
thy = coeff * FUT / FUU
elif obs == 'AUTsivers':
# convention factor
coeff = 1.
FUT = sivers.get_FUT(x, z, Q2, pT, tar, had)
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
thy = coeff * FUT / FUU
elif obs == 'AUUcos2':
epsilon = (1 - y) / (1 - y + 0.5 * y**2)
coeff = 1.
if col == 'HERMES':
coeff = epsilon # add depolarization factor for HERMES
if col == 'COMPASS':
coeff = 1.
if col == 'JLAB':
coeff = epsilon # add depolarization factor for CLAS
FUUcos2 = boermulders.get_FUU(x, z, Q2, pT, tar, had)
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
thy = coeff * FUUcos2 / FUU
elif obs == 'AUTsinphiS': # This is for collinear!
if tar == 'p':
pT = None
FUTsinphiS = AUTsinphiS.get_FX(x, z, Q2, pT, 'p', had)
FUU = upol.get_FUU(x, z, Q2, pT, 'p', had)
coeff = 1.
if col == 'HERMES':
# add depolarization factor for HERMES
coeff = np.sqrt(1.0 - y) * (2 - y) / (1 - y + 0.5 * y**2)
if col == 'COMPASS':
coeff = 1.
thy = coeff * FUTsinphiS / FUU
else:
print 'ERR: exp=%d obs=%s and target=%s not implemented' % (k, obs,
tar)
sys.exit()
return thy
def gen_report(self, verb=1, level=1):
"""
verb = 0: Do not print on screen. Only return list of strings
verv = 1: print on screen the report
level= 0: only the total chi2s
level= 1: include point by point
"""
L = []
L.append('reaction: %s' % self.reaction)
L.append('%7s %10s %10s %10s %10s %5s %10s %10s %10s %10s' %
('idx', 'tar', 'had', 'col', 'obs', 'npts', 'chi2',
'chi2/npts', 'rchi2', 'nchi2'))
for k in self.tabs:
#print k,len(self.tabs[k]['value'])
if self.tabs[k]['value'].size == 0:
continue
res = self._get_residuals(k)
rres = self._get_rres(k)
nres = self._get_nres(k)
chi2 = np.sum(res**2)
rchi2 = np.sum(rres**2)
nchi2 = nres**2
tar = self.tabs[k]['target'][0]
col = self.tabs[k]['col'][0].split()[0]
obs = self.tabs[k]['obs'][0].split()[0]
had = self.tabs[k]['hadron'][0].split()[0]
npts = res.size
if npts > 0:
L.append(
'%7d %10s %10s %10s %10s %5d %10.2f %10.2f %10.2f %10.2f'
% (k, tar, had, col, obs, npts, chi2, chi2 / npts, rchi2,
nchi2))
elif npts == 0:
L.append(
'%7d %10s %10s %10s %10s %5d %10.2f %10.2f %10.2f %10.2f' %
(k, tar, had, col, obs, npts, chi2, 0.0, rchi2, nchi2))
if level == 1:
L.append('-' * 100)
msg = 'idx=%7d, '
msg += 'col=%7s, '
msg += 'tar=%7s, '
msg += 'had=%7s, '
msg += 'obs=%7s, '
if 'dependence' in self.tabs[k]:
msg += 'dep=%7s, '
if 'Dependence' in self.tabs[k]:
msg += 'dep=%7s, '
msg += 'x=%10.3e, '
msg += 'z=%10.3e, '
msg += 'pT=%10.3e, '
msg += 'Q2=%10.3e, '
msg += 'yh=%10.3e, '
msg += 'yp=%10.3e, '
msg += 'dy=%10.3e, '
msg += 'exp=%10.3e, '
msg += 'alpha=%10.3e, '
msg += 'thy=%10.3e, '
if 'dthy' in self.tabs[k]:
msg += 'dthy=%10.3e, '
msg += 'shift=%10.3e, '
msg += 'chi2=%10.3f '
for k in self.tabs:
if len(self.tabs[k]['value']) == 0:
continue
for i in range(len(self.tabs[k]['value'])):
row = [k]
row.append(self.tabs[k]['col'][i])
row.append(self.tabs[k]['target'][i])
row.append(self.tabs[k]['hadron'][i])
row.append(self.tabs[k]['obs'][i])
if 'dependence' in self.tabs[k]:
row.append(self.tabs[k]['dependence'][i].strip())
if 'Dependence' in self.tabs[k]:
row.append(self.tabs[k]['Dependence'][i].strip())
row.append(self.tabs[k]['x'][i])
row.append(self.tabs[k]['z'][i])
row.append(self.tabs[k]['pT'][i])
row.append(self.tabs[k]['Q2'][i])
row.append(self.tabs[k]['yh'][i])
row.append(self.tabs[k]['yp'][i])
row.append(self.tabs[k]['dy'][i])
row.append(self.tabs[k]['value'][i])
row.append(self.tabs[k]['alpha'][i])
row.append(self.tabs[k]['thy'][i])
if 'dthy' in self.tabs[k]:
row.append(self.tabs[k]['dthy'][i])
row.append(self.tabs[k]['shift'][i])
# row.append(self.tabs[k]['residuals'][i])
# row.append(self.tabs[k]['r-residuals'][i])
res = self.tabs[k]['residuals'][i]
if res < 0:
chi2 = -res**2
else:
chi2 = res**2
row.append(chi2)
row = tuple(row)
L.append(msg % row)
if verb == 0:
return L
elif verb == 1:
for l in L:
print l
return L
class RESIDUALS(_RESIDUALS):
def __init__(self):
self.reaction = 'sidis'
self.tabs = conf['sidis tabs']
self.dis_stfuncs = DIS_STFUNCS()
self.setup()
def _get_theory(self, entry):
k, i = entry
x = self.tabs[k]['x'][i]
y = self.tabs[k]['y'][i]
z = self.tabs[k]['z'][i]
Q2 = self.tabs[k]['Q2'][i]
pT = self.tabs[k]['pT'][i]
exp = self.tabs[k]['value'][i]
tar = self.tabs[k]['target'][i]
had = self.tabs[k]['hadron'][i]
obs = self.tabs[k]['obs'][i].strip()
col = self.tabs[k]['col'][i].strip().upper()
if tar == 'proton': tar = 'p'
elif tar == 'neutron': tar = 'n'
elif tar == 'deuteron': tar = 'd'
if obs == 'FUU':
thy = upol.get_FUU(x, z, Q2, pT, tar, had)
elif obs == 'M':
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
F2 = self.dis_stfuncs.get_F2(x, Q2, tar)
thy = FUU / F2
if col == 'HERMES': thy *= 2 * np.pi * pT
elif obs == 'AUTcollins':
# convention factor
coeff = 1.
if col == 'HERMES': coeff = 1 # hermes is sin(phi_s+phi_h)
elif col == 'COMPASS': coeff = -1 # compass is sin(phi_s+phi_h+pi)
# add depolarization factor
if col == 'HERMES': coeff *= 2 * (1 - y) / (1 + (1 - y)**2)
FUT = collins.get_FUT(x, z, Q2, pT, tar, had)
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
thy = coeff * FUT / FUU
elif obs == 'AUTsivers':
# convention factor
coeff = 1.
FUT = sivers.get_FUT(x, z, Q2, pT, tar, had)
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
thy = coeff * FUT / FUU
elif obs == 'AUUcos2':
''' The data from JLab and HERMES do not require any integrations,
but require a y-based coefficient. The COMPASS data requires an
integration over y, and does not require the coefficient. '''
def yield_thy(accelerator, should_integrate):
# depending on input parameters, either selects the coefficient
# method or the integration method of producing a residual value.
# root_s and w2_min also fluctuate based on accelerator.
# set known values based on the source of AUUcos2 data
if accelerator == 'COMPASS':
ROOT_S = 17.3
W2_MIN = 25
RANGE_MIN = 0.2
RANGE_MAX = 0.9
#TODO: define accelerator constants for JLab and HERMES data
elif accelerator == 'JLAB':
ROOT_S = 7.25
W2_MIN = 4
RANGE_MIN = 0.2
RANGE_MAX = 0.85
elif accelerator == 'HERMES':
ROOT_S = 3.42
W2_MIN = 10
RANGE_MIN = 0.2
RANGE_MAX = 0.85
if should_integrate:
def fast_integrate(f, low, hi):
f = np.vectorize(f)
return integrate.fixed_quad(f, low, hi, n=10)
M2 = conf['aux'].M**2
Q2_MIN = 1
Q2_MAX = 1000
yA = max( # lower bound of integration
RANGE_MIN, Q2_MAX / (x * ((ROOT_S**2) - M2)),
(W2_MIN - M2) / ((1 - x) * ((ROOT_S**2) - M2)))
yB = min( # upper bound of integration
Q2_MAX / (x * ((ROOT_S**2) - M2)), RANGE_MAX)
# any value dependent on y must be a function so that its value
# may be recalculated during the integration process
Q_compass = lambda y: np.sqrt(((ROOT_S**2) - M2) * x * y)
FUU = lambda y: upol.get_FUU(x, z,
Q_compass(y)**2, pT, tar, had)
FUUcos2 = lambda y: (boermulders.get_FUU(
x, z,
Q_compass(y)**2, pT, tar, had) + cahn.get_cahn(
x, z,
Q_compass(y)**2, pT, tar, had))
# integrate over y for the numerator and denominator of AUUcos2
FUUcos2_integral = fast_integrate(
lambda y: (1 / (Q_compass(y)**4)) * FUUcos2(y), yA,
yB)[0]
FUU_integral = fast_integrate(
lambda y: (1 / (Q_compass(y)**4)) * FUU(y), yA, yB)[0]
theory = FUUcos2_integral / FUU_integral
else: # use coefficient method
coeff = (1 - y) / (1 - y + 0.5 * y**2)
FUUcos2 = (boermulders.get_FUU(x, z, Q2, pT, tar, had) +
cahn.get_cahn(x, z, Q2, pT, tar, had))
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
theory = coeff * FUUcos2 / FUU_integral
return theory
thy = yield_thy(col, should_integrate=True)
elif obs == 'AUTsinphiS': # This is for collinear!
if tar == 'p':
pT = None
FUTsinphiS = AUTsinphiS.get_FX(x, z, Q2, pT, 'p', had)
FUU = upol.get_FUU(x, z, Q2, pT, 'p', had)
coeff = 1.
if col == 'HERMES':
# add depolarization factor for HERMES
coeff = np.sqrt(1.0 - y) * (2 - y) / (1 - y + 0.5 * y**2)
if col == 'COMPASS':
coeff = 1.
thy = coeff * FUTsinphiS / FUU
else:
print 'ERR: exp=%d obs=%s and target=%s not implemented' % (k, obs,
tar)
sys.exit()
return thy
def gen_report(self, verb=1, level=1):
"""
verb = 0: Do not print on screen. Only return list of strings
verv = 1: print on screen the report
level= 0: only the total chi2s
level= 1: include point by point
"""
L = []
L.append('reaction: %s' % self.reaction)
L.append('%7s %10s %10s %10s %10s %5s %10s %10s %10s %10s' %
('idx', 'tar', 'had', 'col', 'obs', 'npts', 'chi2',
'chi2/npts', 'rchi2', 'nchi2'))
for k in self.tabs:
#print k,len(self.tabs[k]['value'])
if self.tabs[k]['value'].size == 0:
continue
res = self._get_residuals(k)
rres = self._get_rres(k)
nres = self._get_nres(k)
chi2 = np.sum(res**2)
rchi2 = np.sum(rres**2)
nchi2 = nres**2
tar = self.tabs[k]['target'][0]
col = self.tabs[k]['col'][0].split()[0]
obs = self.tabs[k]['obs'][0].split()[0]
had = self.tabs[k]['hadron'][0].split()[0]
npts = res.size
if npts > 0:
L.append(
'%7d %10s %10s %10s %10s %5d %10.2f %10.2f %10.2f %10.2f'
% (k, tar, had, col, obs, npts, chi2, chi2 / npts, rchi2,
nchi2))
elif npts == 0:
L.append(
'%7d %10s %10s %10s %10s %5d %10.2f %10.2f %10.2f %10.2f' %
(k, tar, had, col, obs, npts, chi2, 0.0, rchi2, nchi2))
if level == 1:
L.append('-' * 100)
msg = 'idx=%7d, '
msg += 'col=%7s, '
msg += 'tar=%7s, '
msg += 'had=%7s, '
msg += 'obs=%7s, '
if 'dependence' in self.tabs[k]:
msg += 'dep=%7s, '
if 'Dependence' in self.tabs[k]:
msg += 'dep=%7s, '
msg += 'x=%10.3e, '
msg += 'z=%10.3e, '
msg += 'pT=%10.3e, '
msg += 'Q2=%10.3e, '
msg += 'yh=%10.3e, '
msg += 'yp=%10.3e, '
msg += 'dy=%10.3e, '
msg += 'exp=%10.3e, '
msg += 'alpha=%10.3e, '
msg += 'thy=%10.3e, '
if 'dthy' in self.tabs[k]:
msg += 'dthy=%10.3e, '
msg += 'shift=%10.3e, '
msg += 'chi2=%10.3f '
for k in self.tabs:
if len(self.tabs[k]['value']) == 0:
continue
for i in range(len(self.tabs[k]['value'])):
row = [k]
row.append(self.tabs[k]['col'][i])
row.append(self.tabs[k]['target'][i])
row.append(self.tabs[k]['hadron'][i])
row.append(self.tabs[k]['obs'][i])
if 'dependence' in self.tabs[k]:
row.append(self.tabs[k]['dependence'][i].strip())
if 'Dependence' in self.tabs[k]:
row.append(self.tabs[k]['Dependence'][i].strip())
row.append(self.tabs[k]['x'][i])
row.append(self.tabs[k]['z'][i])
row.append(self.tabs[k]['pT'][i])
row.append(self.tabs[k]['Q2'][i])
row.append(self.tabs[k]['yh'][i])
row.append(self.tabs[k]['yp'][i])
row.append(self.tabs[k]['dy'][i])
row.append(self.tabs[k]['value'][i])
row.append(self.tabs[k]['alpha'][i])
row.append(self.tabs[k]['thy'][i])
if 'dthy' in self.tabs[k]:
row.append(self.tabs[k]['dthy'][i])
row.append(self.tabs[k]['shift'][i])
# row.append(self.tabs[k]['residuals'][i])
# row.append(self.tabs[k]['r-residuals'][i])
res = self.tabs[k]['residuals'][i]
if res < 0:
chi2 = -res**2
else:
chi2 = res**2
row.append(chi2)
row = tuple(row)
L.append(msg % row)
if verb == 0:
return L
elif verb == 1:
for l in L:
print l
return L
class RESIDUALS(_RESIDUALS):
def __init__(self, react='sidis'):
if react == 'sidis':
self.reaction = 'sidis'
self.tabs = conf['sidis tabs']
elif react == 'sidisEIC':
self.reaction = 'sidisEIC'
self.tabs = conf['sidisEIC tabs']
elif react == 'sidisSoLID':
self.reaction = 'sidisSoLID'
self.tabs = conf['sidisSoLID tabs']
self.dis_stfuncs = DIS_STFUNCS()
self.setup()
def _get_theory(self, entry):
k, i = entry
x = self.tabs[k]['x'][i]
try:
y = self.tabs[k]['y'][i]
except (ValueError, KeyError):
y = None
z = self.tabs[k]['z'][i]
try:
Q2 = self.tabs[k]['Q2'][i]
except ValueError:
Q2 = None
pT = self.tabs[k]['pT'][i]
exp = self.tabs[k]['value'][i]
tar = self.tabs[k]['target'][i]
had = self.tabs[k]['hadron'][i]
obs = self.tabs[k]['obs'][i].strip()
col = self.tabs[k]['col'][i].strip().upper()
try:
F2 = self.tabs[k]['F2'][i]
except KeyError:
F2 = None
M = conf['aux'].M
M2 = conf['aux'].M**2
Mpi2 = conf['aux'].Mpi**2
if tar == 'proton': tar = 'p'
elif tar == 'neutron': tar = 'n'
elif tar == 'deuteron': tar = 'd'
if obs == 'FUU':
thy = upol.get_FUU(x, z, Q2, pT, tar, had)
elif obs == 'M':
if had == 'h+': had = 'pi+'
if had == 'h-': had = 'pi-'
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
if F2 == None: F2 = self.dis_stfuncs.get_F2(x, Q2, tar)
thy = FUU / F2
if col == 'HERMES' or col == 'hermes': thy = 2 * np.pi * pT * thy
if col == 'COMPASS' or col == 'compass': thy = np.pi * thy
elif obs == 'AUTcollins':
# convention factor
coeff = 1.
if col == 'HERMES': coeff = 1 # hermes is sin(phi_s+phi_h)
elif col == 'COMPASS': coeff = -1 # compass is sin(phi_s+phi_h+pi)
# add depolarization factor
if col == 'HERMES': coeff *= 2 * (1 - y) / (1 + (1 - y)**2)
FUT = collins.get_FUT(x, z, Q2, pT, tar, had)
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
thy = coeff * FUT / FUU
elif obs == 'AUTsivers':
# convention factor
coeff = 1.
FUT = sivers.get_FUT(x, z, Q2, pT, tar, had)
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
thy = coeff * FUT / FUU
elif obs == 'AUUcos2':
if had == 'h+': had = 'pi+'
if had == 'h-': had = 'pi-'
''' The data from JLab and HERMES do not require any integrations,
but require a y-based coefficient. The COMPASS data requires an
integration over y, and does not require the coefficient. '''
def yield_thy(accelerator, should_integrate, ny=10):
# depending on input parameters, either selects the coefficient
# method or the integration method of producing a residual value.
# root_s and W2_min also fluctuate based on accelerator.
# set known values based on the source of AUUcos2 data
if accelerator == 'COMPASS':
ROOT_S = 17.3
W2_MIN = 25.0
RANGE_MIN = 0.2
RANGE_MAX = 0.9
elif accelerator == 'CLAS': #JLab col in the data files
ROOT_S = 3.42
W2_MIN = 4.0
RANGE_MIN = 0.2
RANGE_MAX = 0.85
elif accelerator == 'HERMES':
ROOT_S = 7.25
W2_MIN = 10
RANGE_MIN = 0.2
RANGE_MAX = 0.85
if should_integrate:
def fast_integrate(f, low, hi, ny):
f = np.vectorize(f)
return integrate.fixed_quad(f, low, hi, n=ny)
Q2_MIN = 1
Q2_MAX = 1000
yA = max( # lower bound of integration
RANGE_MIN, Q2_MIN / (x * ((ROOT_S**2) - M2)),
(W2_MIN - M2) / ((1 - x) * ((ROOT_S**2) - M2)))
yB = min( # upper bound of integration
Q2_MAX / (x * ((ROOT_S**2) - M2)), RANGE_MAX)
# any value dependent on y must be a function so that its value
# may be recalculated during the integration process
Q = lambda y: np.sqrt(((ROOT_S**2) - M2) * x * y)
if accelerator == 'HERMES':
FUU = lambda y: (1. - y + 0.5 * y**2) * upol.get_FUU(
x, z,
Q(y)**2, pT, tar, had)
FUUcos2 = lambda y: (1. - y) * (boermulders.get_FUU(
x, z,
Q(y)**2, pT, tar, had) + cahn.get_cahn(
x, z,
Q(y)**2, pT, tar, had))
elif accelerator == 'COMPASS':
FUU = lambda y: upol.get_FUU(x, z,
Q(y)**2, pT, tar, had)
FUUcos2 = lambda y: (boermulders.get_FUU(
x, z,
Q(y)**2, pT, tar, had) + cahn.get_cahn(
x, z,
Q(y)**2, pT, tar, had))
elif accelerator == 'CLAS':
# CLAS accelerators measure H4 / (H2 + H1), which can be related
# to the AUUcos2 asymmetry.
gamma = lambda y: (2 * M * x) / Q(y)
kappa = lambda y: 1 / (1 + gamma(x, y)**2)
zeta = lambda y: 1 - y - (0.25 * (gamma(x, y)**2) *
(y**2))
epsilon = lambda y: 1 / (1 + (
(y**2) / (2 * kappa(x, y) * zeta(x, y))))
ppa_over_Eh = lambda y: np.sqrt(1 - (pT**2 + Mpi**2) *
(2. * M * x /
(z * Q(y)**2))**2)
FUU = lambda y: ppa_over_Eh * (kappa / epsilon) * (
1 + gamma**2 /
(2. * x)) * upol.get_FUU(x, z,
Q(y)**2, pT, tar, had)
FUUcos2 = lambda y: (ppa_over_Eh / 2.) * (
1 + gamma**2 / (2. * x)) * (boermulders.get_FUU(
x, z,
Q(y)**2, pT, tar, had) + cahn.get_cahn(
x, z,
Q(y)**2, pT, tar, had))
# integrate over y for the numerator and denominator of AUUcos2
FUUcos2_integral = fast_integrate(
lambda y: (1. / (Q(y)**4)) * FUUcos2(y), yA, yB, ny)[0]
FUU_integral = fast_integrate(
lambda y: (1. / (Q(y)**4)) * FUU(y), yA, yB, ny)[0]
theory = FUUcos2_integral / FUU_integral
else: # no integration
if accelerator == 'HERMES':
coeff = (1 - y) / (1 - y + 0.5 * y**2)
elif accelerator == 'COMPASS':
coeff = 1.
elif accelerator == 'CLAS':
gamma = (2. * M * x) / np.sqrt(Q2)
kappa = 1. / (1. + gamma**2.)
zeta = 1. - y - (0.25 * (gamma**2.) * (y**2.))
epsilon = 1. / (1. + ((y**2.) / (2. * kappa * zeta)))
coeff = epsilon / (2. * kappa)
FUUcos2 = (boermulders.get_FUU(x, z, Q2, pT, tar, had) +
cahn.get_cahn(x, z, Q2, pT, tar, had))
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
theory = coeff * FUUcos2 / FUU
return theory
if col == 'COMPASS':
thy = yield_thy(col, should_integrate=True, ny=10)
elif col == 'HERMES':
thy = yield_thy(col, should_integrate=False, ny=10)
elif col == 'CLAS':
thy = yield_thy(col, should_integrate=False, ny=10)
elif obs == 'AUTsinphiS': # This is for collinear!
if tar == 'p':
pT = None
FUTsinphiS = AUTsinphiS.get_FX(x, z, Q2, pT, 'p', had)
FUU = upol.get_FUU(x, z, Q2, pT, 'p', had)
coeff = 1.
if col == 'HERMES':
# add depolarization factor for HERMES
coeff = np.sqrt(1.0 - y) * (2 - y) / (1 - y + 0.5 * y**2)
if col == 'COMPASS':
coeff = 1.
thy = coeff * FUTsinphiS / FUU
else:
print('ERR: exp=%d obs=%s and target=%s not implemented' %
(k, obs, tar))
sys.exit()
return thy
def gen_report(self, verb=1, level=1):
"""
verb = 0: Do not print on screen. Only return list of strings
verv = 1: print on screen the report
level= 0: only the total chi2s
level= 1: include point by point
"""
L = []
L.append('reaction: %s' % self.reaction)
L.append('%7s %10s %10s %10s %10s %5s %10s %10s %10s %10s' %
('idx', 'tar', 'had', 'col', 'obs', 'npts', 'chi2',
'chi2/npts', 'rchi2', 'nchi2'))
for k in self.tabs:
#print k,len(self.tabs[k]['value'])
if self.tabs[k]['value'].size == 0:
continue
res = self._get_residuals(k)
rres = self._get_rres(k)
nres = self._get_nres(k)
chi2 = np.sum(res**2)
rchi2 = np.sum(rres**2)
nchi2 = nres**2
tar = self.tabs[k]['target'][0]
col = self.tabs[k]['col'][0].split()[0]
obs = self.tabs[k]['obs'][0].split()[0]
had = self.tabs[k]['hadron'][0].split()[0]
npts = res.size
if npts > 0:
L.append(
'%7d %10s %10s %10s %10s %5d %10.2f %10.2f %10.2f %10.2f'
% (k, tar, had, col, obs, npts, chi2, chi2 / npts, rchi2,
nchi2))
elif npts == 0:
L.append(
'%7d %10s %10s %10s %10s %5d %10.2f %10.2f %10.2f %10.2f' %
(k, tar, had, col, obs, npts, chi2, 0.0, rchi2, nchi2))
if level == 1:
L.append('-' * 100)
msg = 'idx=%7d, '
msg += 'col=%7s, '
msg += 'tar=%7s, '
msg += 'had=%7s, '
msg += 'obs=%7s, '
if 'dependence' in self.tabs[k]:
msg += 'dep=%7s, '
if 'Dependence' in self.tabs[k]:
msg += 'dep=%7s, '
msg += 'x=%10.3e, '
msg += 'z=%10.3e, '
msg += 'pT=%10.3e, '
msg += 'Q2=%10.3e, '
msg += 'yh=%10.3e, '
msg += 'yp=%10.3e, '
msg += 'dy=%10.3e, '
msg += 'exp=%10.3e, '
msg += 'alpha=%10.3e, '
msg += 'thy=%10.3e, '
if 'dthy' in self.tabs[k]:
msg += 'dthy=%10.3e, '
msg += 'shift=%10.3e, '
msg += 'chi2=%10.3f '
for k in self.tabs:
if len(self.tabs[k]['value']) == 0:
continue
for i in range(len(self.tabs[k]['value'])):
row = [k]
row.append(self.tabs[k]['col'][i])
row.append(self.tabs[k]['target'][i])
row.append(self.tabs[k]['hadron'][i])
row.append(self.tabs[k]['obs'][i])
if 'dependence' in self.tabs[k]:
row.append(self.tabs[k]['dependence'][i].strip())
if 'Dependence' in self.tabs[k]:
row.append(self.tabs[k]['Dependence'][i].strip())
row.append(self.tabs[k]['x'][i])
row.append(self.tabs[k]['z'][i])
row.append(self.tabs[k]['pT'][i])
row.append(self.tabs[k]['Q2'][i])
row.append(self.tabs[k]['yh'][i])
row.append(self.tabs[k]['yp'][i])
row.append(self.tabs[k]['dy'][i])
row.append(self.tabs[k]['value'][i])
row.append(self.tabs[k]['alpha'][i])
row.append(self.tabs[k]['thy'][i])
if 'dthy' in self.tabs[k]:
row.append(self.tabs[k]['dthy'][i])
row.append(self.tabs[k]['shift'][i])
# row.append(self.tabs[k]['residuals'][i])
# row.append(self.tabs[k]['r-residuals'][i])
res = self.tabs[k]['residuals'][i]
if res < 0:
chi2 = -res**2
else:
chi2 = res**2
row.append(chi2)
row = tuple(row)
L.append(msg % row)
if verb == 0:
return L
elif verb == 1:
for l in L:
print(l)
return L
class RESIDUALS(_RESIDUALS):
def __init__(self):
self.reaction = 'sidis'
self.tabs = conf['sidis tabs']
self.dis_stfuncs = DIS_STFUNCS()
self.setup()
def _get_theory(self, entry):
k, i = entry
x = self.tabs[k]['x'][i]
y = self.tabs[k]['y'][i]
z = self.tabs[k]['z'][i]
Q2 = self.tabs[k]['Q2'][i]
pT = self.tabs[k]['pT'][i]
exp = self.tabs[k]['value'][i]
tar = self.tabs[k]['target'][i]
had = self.tabs[k]['hadron'][i]
obs = self.tabs[k]['obs'][i].strip()
col = self.tabs[k]['col'][i].strip().upper()
M = conf['aux'].M
S = 299.29
W2_min = 25
Q2_min = 1
Q2_max = 1000
if tar == 'proton': tar = 'p'
elif tar == 'neutron': tar = 'n'
elif tar == 'deuteron': tar = 'd'
if obs == 'FUU':
thy = upol.get_FUU(x, z, Q2, pT, tar, had)
elif obs == 'M':
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
F2 = self.dis_stfuncs.get_F2(x, Q2, tar)
thy = FUU / F2
if col == 'HERMES': thy *= 2 * np.pi * pT
elif obs == 'AUTcollins':
# convention factor
coeff = 1.
if col == 'HERMES': coeff = 1 # hermes is sin(phi_s+phi_h)
elif col == 'COMPASS': coeff = -1 # compass is sin(phi_s+phi_h+pi)
# add depolarization factor
if col == 'HERMES': coeff *= 2 * (1 - y) / (1 + (1 - y)**2)
FUT = collins.get_FUT(x, z, Q2, pT, tar, had)
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
thy = coeff * FUT / FUU
elif obs == 'AUTsivers':
# convention factor
coeff = 1.
FUT = sivers.get_FUT(x, z, Q2, pT, tar, had)
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
thy = coeff * FUT / FUU
elif obs == 'AUUcos2':
# data from Hermes and Jlab do not need integration
if col == ['COMPASS', 'JLAB']:
coeff = (1 - y) / (1 - y + 0.5 * y**2)
FUUcos2 = boermulders.get_FUU(x, z, Q2, pT, tar,
had) + cahn.get_cahn(
x, z, Q2, pT, tar, had)
FUU = upol.get_FUU(x, z, Q2, pT, tar, had)
thy = coeff * FUUcos2 / FUU
# data from compass will be integrated over y
elif col == 'HERMES':
def integrand(f, low, hi):
f = np.vectorize(f)
return fixed_quad(f, low, hi, n=10)
yLOW = max(0.2, Q2_max / (x * (S - M**2)),
(W2_min - M**2) / ((1 - x) * (S - M**2)))
yUP = min(Q2_max / (x * (S - M**2)), 0.9)
Q_XY = lambda y: np.sqrt((S - M**2) * x * y)
FUUcos2 = lambda y: boermulders.get_FUU(
x, z,
Q_XY(y)**2, pT, tar, had) + cahn.get_cahn(
x, z,
Q_XY(y)**2, pT, tar, had)
FUU = lambda y: upol.get_FUU(x, z, Q_XY(y)**2, pT, tar, had)
#inegrate over y for top and bottom of AUUcos2 compass
FUUcos2_int = integrand(
lambda y: (1 / (Q_XY(y)**4)) * FUUcos2(y), yLOW, yUP)[0]
FUU_int = integrand(lambda y: (1 / (Q_XY(y)**4)) * FUU(y),
yLOW, yUP)[0]
thy = FUUcos2_int / FUU_int
elif obs == 'AUTsinphiS': # This is for collinear!
if tar == 'p':
pT = None
FUTsinphiS = AUTsinphiS.get_FX(x, z, Q2, pT, 'p', had)
FUU = upol.get_FUU(x, z, Q2, pT, 'p', had)
coeff = 1.
if col == 'HERMES':
# add depolarization factor for HERMES
coeff = np.sqrt(1.0 - y) * (2 - y) / (1 - y + 0.5 * y**2)
if col == 'COMPASS':
coeff = 1.
thy = coeff * FUTsinphiS / FUU
else:
print 'ERR: exp=%d obs=%s and target=%s not implemented' % (k, obs,
tar)
sys.exit()
return thy
def gen_report(self, verb=1, level=1):
"""
verb = 0: Do not print on screen. Only return list of strings
verb = 1: print on screen the report
level= 0: only the total chi2s
level= 1: include point by point
"""
L = []
L.append('reaction: %s' % self.reaction)
L.append('%7s %10s %10s %10s %10s %5s %10s %10s %10s %10s' %
('idx', 'tar', 'had', 'col', 'obs', 'npts', 'chi2',
'chi2/npts', 'rchi2', 'nchi2'))
for k in self.tabs:
#print k,len(self.tabs[k]['value'])
if self.tabs[k]['value'].size == 0:
continue
res = self._get_residuals(k)
rres = self._get_rres(k)
nres = self._get_nres(k)
chi2 = np.sum(res**2)
rchi2 = np.sum(rres**2)
nchi2 = nres**2
tar = self.tabs[k]['target'][0]
col = self.tabs[k]['col'][0].split()[0]
obs = self.tabs[k]['obs'][0].split()[0]
had = self.tabs[k]['hadron'][0].split()[0]
npts = res.size
if npts > 0:
L.append(
'%7d %10s %10s %10s %10s %5d %10.2f %10.2f %10.2f %10.2f'
% (k, tar, had, col, obs, npts, chi2, chi2 / npts, rchi2,
nchi2))
elif npts == 0:
L.append(
'%7d %10s %10s %10s %10s %5d %10.2f %10.2f %10.2f %10.2f' %
(k, tar, had, col, obs, npts, chi2, 0.0, rchi2, nchi2))
if level == 1:
L.append('-' * 100)
msg = 'idx=%7d, '
msg += 'col=%7s, '
msg += 'tar=%7s, '
msg += 'had=%7s, '
msg += 'obs=%7s, '
if 'dependence' in self.tabs[k]:
msg += 'dep=%7s, '
if 'Dependence' in self.tabs[k]:
msg += 'dep=%7s, '
msg += 'x=%10.3e, '
msg += 'z=%10.3e, '
msg += 'pT=%10.3e, '
msg += 'Q2=%10.3e, '
msg += 'yh=%10.3e, '
msg += 'yp=%10.3e, '
msg += 'dy=%10.3e, '
msg += 'exp=%10.3e, '
msg += 'alpha=%10.3e, '
msg += 'thy=%10.3e, '
if 'dthy' in self.tabs[k]:
msg += 'dthy=%10.3e, '
msg += 'shift=%10.3e, '
msg += 'chi2=%10.3f '
for k in self.tabs:
if len(self.tabs[k]['value']) == 0:
continue
for i in range(len(self.tabs[k]['value'])):
row = [k]
row.append(self.tabs[k]['col'][i])
row.append(self.tabs[k]['target'][i])
row.append(self.tabs[k]['hadron'][i])
row.append(self.tabs[k]['obs'][i])
if 'dependence' in self.tabs[k]:
row.append(self.tabs[k]['dependence'][i].strip())
if 'Dependence' in self.tabs[k]:
row.append(self.tabs[k]['Dependence'][i].strip())
row.append(self.tabs[k]['x'][i])
row.append(self.tabs[k]['z'][i])
row.append(self.tabs[k]['pT'][i])
row.append(self.tabs[k]['Q2'][i])
row.append(self.tabs[k]['yh'][i])
row.append(self.tabs[k]['yp'][i])
row.append(self.tabs[k]['dy'][i])
row.append(self.tabs[k]['value'][i])
row.append(self.tabs[k]['alpha'][i])
row.append(self.tabs[k]['thy'][i])
if 'dthy' in self.tabs[k]:
row.append(self.tabs[k]['dthy'][i])
row.append(self.tabs[k]['shift'][i])
# row.append(self.tabs[k]['residuals'][i])
# row.append(self.tabs[k]['r-residuals'][i])
res = self.tabs[k]['residuals'][i]
if res < 0:
chi2 = -res**2
else:
chi2 = res**2
row.append(chi2)
row = tuple(row)
L.append(msg % row)
if verb == 0:
return L
elif verb == 1:
for l in L:
print l
return L