def test_get_order_plot_multi(make_data_path, clean_astro_ui):
"""Rather than fake data, use a known dataset.
Here we pretend we have three orders but with the same
response (except that the ARF is 0.5, 0.4, 0.25 of the
normal ARF).
"""
setup_order_plot(make_data_path)
fplot = ui.get_fit_plot()
oplot = ui.get_order_plot()
# The idea is to compare the X range of plot_fit to plot_order
# (but accessed just using the plot objects rather than creating
# an actual plot).
#
# First some safety checks
assert fplot.dataplot.xlo == pytest.approx(fplot.modelplot.xlo)
assert fplot.dataplot.xhi == pytest.approx(fplot.modelplot.xhi)
assert len(oplot.xlo) == 3
assert len(oplot.xhi) == 3
assert len(oplot.y) == 3
assert oplot.xlo[1] == pytest.approx(oplot.xlo[0])
assert oplot.xlo[2] == pytest.approx(oplot.xlo[0])
assert oplot.xhi[1] == pytest.approx(oplot.xhi[0])
assert oplot.xhi[2] == pytest.approx(oplot.xhi[0])
# We know the y values are 0.5, 0.4, 0.25 times the original arf
# so we can compare them.
#
assert oplot.y[1] == pytest.approx(oplot.y[0] * 0.4 / 0.5)
assert oplot.y[2] == pytest.approx(oplot.y[0] * 0.25 / 0.5)
xlo = oplot.xlo[0]
xhi = oplot.xhi[0]
assert len(xlo) == 564
assert xlo[0] == pytest.approx(0.46720001101493835)
assert xhi[-1] == pytest.approx(9.869600296020508)
# The model plot is technically drawn the same way as the order plot
# (ungrouped) but it uses different code (sherpa.astro.plot.ModelHistogram)
# so let's compare.
#
mplot = ui.get_model_plot()
assert mplot.xlo[0] == pytest.approx(0.46720001101493835)
assert mplot.xhi[-1] == pytest.approx(9.869600296020508)
# Also compare to the fit plot (which is grouped)
#
assert fplot.modelplot.xlo[0] == pytest.approx(0.46720001101493835)
assert fplot.modelplot.xhi[-1] == pytest.approx(9.869600296020508)
def plot(self):
'''Plot model with current parameters'''
import sherpa.astro.ui as sherpa
if self.xpa is None:
import matplotlib.pylab as plt
sherpa.plot_fit_delchi()
plt.show()
return
plots = self.xpaget(self.xpa, "plot") # Get a list of plots.
plots.split(" ")
newplot = ("dax_model_editor" not in plots)
_f = sherpa.get_fit_plot()
_d = _f.dataplot
_m = _f.modelplot
if _d.xerr is None:
_d.xerr = (_d.x-_d.x) # zeros
if self.x_label is None:
xlab = _f.dataplot.xlabel
else:
xlab = self.x_label
if self.y_label is None:
ylab = _f.dataplot.ylabel
else:
ylab = self.y_label
import dax.dax_plot_utils as dax_plot
if hasattr(_m, "xlo"):
mx = list(_m.xlo)
mx.append(_m.xhi[-1])
my = list(_m.y)
my.append(_m.y[-1])
step = True
else:
mx = _m.x
my = _m.y
step = False
dax_plot.blt_plot_model(self.xpa, mx, my,
"Dax Model Editor Plot",
xlab, ylab, step=step,
new=newplot, winname="dax_model_editor")
dax_plot.blt_plot_data(self.xpa, _d.x, _d.xerr/2.0, _d.y, _d.yerr)
delta = (_d.y-_m.y)/_d.yerr
ones = _d.yerr*0.0+1.0
dax_plot.blt_plot_delchisqr( self.xpa, _d.x, _d.xerr/2.0, delta, ones, "")
def test_plot_order_multi(make_data_path, clean_astro_ui):
"""Rather than fake data, use a known dataset.
Here we pretend we have three orders but with the same
response (except that the ARF is 0.5, 0.4, 0.25 of the
normal ARF).
"""
pha = make_data_path('3c273.pi')
ui.load_pha(pha)
# It has already loaded in one response
arf = ui.get_arf(resp_id=1)
arf.specresp *= 0.5
for order, scale in enumerate([0.4, 0.25], 2):
ui.load_arf(make_data_path('3c273.arf'), resp_id=order)
ui.load_rmf(make_data_path('3c273.rmf'), resp_id=order)
arf = ui.get_arf(resp_id=order)
arf.specresp *= scale
ui.set_source(ui.powlaw1d.pl)
ui.notice(0.5, 7)
ui.ignore(3, 4)
fplot = ui.get_fit_plot()
oplot = ui.get_order_plot()
# The idea is to compare the X range of plot_fit to plot_order
# (but accessed just using the plot objects rather than creating
# an actual plot).
#
# First some safety checks
assert fplot.dataplot.xlo == pytest.approx(fplot.modelplot.xlo)
assert fplot.dataplot.xhi == pytest.approx(fplot.modelplot.xhi)
assert len(oplot.xlo) == 3
assert len(oplot.xhi) == 3
assert len(oplot.y) == 3
assert oplot.xlo[1] == pytest.approx(oplot.xlo[0])
assert oplot.xlo[2] == pytest.approx(oplot.xlo[0])
assert oplot.xhi[1] == pytest.approx(oplot.xhi[0])
assert oplot.xhi[2] == pytest.approx(oplot.xhi[0])
# We know the y values are 0.5, 0.4, 0.25 times the original arf
# so we can compare them.
#
assert oplot.y[1] == pytest.approx(oplot.y[0] * 0.4 / 0.5)
assert oplot.y[2] == pytest.approx(oplot.y[0] * 0.25 / 0.5)
xlo = oplot.xlo[0]
xhi = oplot.xhi[0]
assert len(xlo) == 564
assert xlo[0] == pytest.approx(0.46720001101493835)
assert xhi[-1] == pytest.approx(9.869600296020508)
# The model plot is technically drawn the same way as the order plot
# (ungrouped) but it uses different code (sherpa.astro.plot.ModelHistogram)
# so let's compare.
#
mplot = ui.get_model_plot()
assert mplot.xlo[0] == pytest.approx(0.46720001101493835)
assert mplot.xhi[-1] == pytest.approx(9.869600296020508)
# Also compare to the fit plot (which is grouped)
#
assert fplot.modelplot.xlo[0] == pytest.approx(0.46720001101493835)
assert fplot.modelplot.xhi[-1] == pytest.approx(9.869600296020508)
solver = BXASolver(id=id, prior=priorfunction, parameters=parameters)
solver.run(resume=True,
verbose=True,
n_live_points=args.num_live_points,
outputfiles_basename=prefix,
frac_remain=0.01,
min_ess=1000)
if not os.path.exists('%sfit.json' % prefix):
print('collecting fit plot data')
set_analysis(id, 'ener', 'counts')
group_adapt(id, 30)
set_stat('chi2gehrels')
pl = get_fit_plot(id)
myplot = dict(counts=int(get_data(id).counts.sum()),
background_counts=int(get_bkg(id).counts.sum()),
data=pl.dataplot.y.tolist(),
dataerr=pl.dataplot.yerr.tolist(),
x=pl.dataplot.x.tolist(),
xerr=pl.dataplot.xerr.tolist(),
instances=[])
for row in tqdm.tqdm(solver.results['samples']):
for p, v in zip(parameters, row):
p.val = v
pl = get_fit_plot(id)
myplot['instances'].append(pl.modelplot.y.tolist())
srcnh.val = 20
pl = get_fit_plot(id)