def fbDiffDist(xi):
x, y = np.meshgrid(np.linspace(-1., 1., 101), np.linspace(-1., 1., 101))
z = csec2DRaw(x, y, xi)
z = z / z.max()
plt.figure(num=1, figsize=(5, 4))
cnt = plt.contourf(x,
y,
z,
levels=100,
cmap=plt.cm.viridis,
vmin=0,
vmax=1)
plt.colorbar(cnt)
plt.axis('equal')
plt.xlabel(r'$\cos{\theta}$', fontsize=16)
plt.ylabel(r'$\cos{\theta_1}$', fontsize=16)
plt.tight_layout()
plt.savefig('/'.join([plotpath, 'fbdist_xi{}.pdf'.format(strxi(xi))]))
plt.savefig('/'.join([plotpath, 'fbdist_xi{}.png'.format(strxi(xi))]))
plt.show()
def drawCorrMtx(lbl, mtx, xi=None, show=True):
""" """
labels = [varttl[key] for key in list(varttl.keys())[:mtx.shape[0]]]
print(labels)
fig, ax = plt.subplots()
im, cbar = heatmap(mtx, labels, labels, ax=ax, cmap="coolwarm_r", cbarlabel=None)
texts = annotate_heatmap(im, valfmt="{x:.2f}")
fig.tight_layout()
fname = 'corr_mtx_{}'.format(lbl)
if xi is not None:
fname = fname + '_xi_{}'.format(strxi(xi))
print('Correlation matrix saved in file {}'.format(fname))
plt.savefig('/'.join([plotpath, '{}.pdf'.format(fname)]))
plt.savefig('/'.join([plotpath, '{}.png'.format(fname)]))
if show:
plt.show()
def drawFitBiasXi(lbl, fresxi, nevt):
""" """
xil = [floxi(xi) for xi in fresxi.keys()]
keys = list(fresxi[strxi(xil[0])].keys())
vals = {key: [] for key in keys}
for xi, fres in fresxi.items():
for key, [mean, err] in fres.items():
tval = floxi(xi) if key == 'xi' else pars[key]
err = err / np.sqrt(nevt)
bias = (mean - tval) / err
vals[key].append(bias)
for key, val in vals.items():
val = np.array(val)
plt.figure(figsize=(6, 5))
plt.errorbar(xil,
val,
yerr=1.,
marker='.',
markersize=12,
linestyle='none',
label=varttl[key])
print('chi2 = {:.3}'.format(np.sum(val**2) / val.shape[0]))
ylabel = r'pull({})'.format(varttl[key])
plt.xlabel(r'$\xi$', fontsize=16)
plt.ylabel(ylabel, fontsize=16)
plt.ylim([-3., 3.])
plt.grid()
plt.tight_layout()
fname = 'bias_{}_{}'.format(lbl, key)
print('Save in {}'.format(fname))
plt.savefig('/'.join([plotpath, fname + '.png']))
plt.savefig('/'.join([plotpath, fname + '.pdf']))
plt.show()
def drawPrecisionNevt(lbl, var, xil, fresxi):
""" """
plt.rc('font', size=14)
n = np.logspace(6, 9)
for xi in xil:
err = fresxi[strxi(xi)][var][1]
print(xi, err)
if err is not None:
plt.plot(n,
err / np.sqrt(n) / xi,
label=r'$\xi=${:.1f}'.format(xi))
plt.grid(which='both')
plt.semilogx()
plt.semilogy()
plt.ylabel(r'$d\xi/\xi$')
plt.xlabel(r'$N$')
plt.xlim((10**6, 10**9))
plt.ylim((1e-4, 2e-2))
plt.legend(loc='best')
plt.tight_layout()
plt.savefig('/'.join([plotpath, '{}_{}_prec.pdf'.format(lbl, var)]))
plt.savefig('/'.join([plotpath, '{}_{}_prec.png'.format(lbl, var)]))
plt.show()
def corxi(dataxi):
return OrderedDict([[strxi(xi), d['fitcor']] for xi, d in dataxi.items()])
def frexi(dataxi):
return OrderedDict([[strxi(xi), d['fitres']] for xi, d in dataxi.items()])