def _test_plots(self, search_keys):
for key in search_keys:
self.search.plot_1D(xkey=key, savefig=True)
if len(search_keys) == 2:
self.search.plot_2D(xkey=search_keys[0],
ykey=search_keys[1],
colorbar=True)
vals = [
np.unique(self.search.data[key]) - getattr(self.Writer, key)
for key in search_keys
]
twoF = self.search.data["twoF"].reshape([len(kval) for kval in vals])
corner_labels = [f"${key} - {key}_0$" for key in search_keys]
corner_labels.append("2F")
gridcorner_fig, gridcorner_axes = pyfstat.gridcorner(
twoF,
vals,
projection="log_mean",
labels=corner_labels,
whspace=0.1,
factor=1.8,
)
gridcorner_fig.savefig(
os.path.join(self.search.outdir,
self.search.label + "_corner.png"))
tglitch_vals_days = (tglitch_vals - tstart) / 86400.0 - dtglitch / 86400.0
print("Making gridcorner plot...")
twoF = search.data["twoF"].reshape(
(len(F0_vals), len(F1_vals), len(delta_F0s_vals), len(tglitch_vals)))
xyz = [F0_vals * 1e6, F1_vals * 1e12, delta_F0s_vals * 1e6, tglitch_vals_days]
labels = [
"$f - f_\\mathrm{s}$\n[$\\mu$Hz]",
"$\\dot{f} - \\dot{f}_\\mathrm{s}$\n[$p$Hz/s]",
"$\\delta f-\\delta f_\\mathrm{s}$\n[$\\mu$Hz]",
"$t^\\mathrm{g} - t^\\mathrm{g}_\\mathrm{s}$\n[d]",
"$t^\\mathrm{g} - t^\\mathrm{g}_\\mathrm{s}$\n[d]",
"$\\widehat{2\\mathcal{F}}$",
]
fig, axes = pyfstat.gridcorner(
twoF,
xyz,
projection="log_mean",
labels=labels,
showDvals=False,
lines=[0, 0, 0, 0],
label_offset=0.25,
max_n_ticks=4,
)
fig.savefig("{}/{}_projection_matrix.png".format(outdir, label),
bbox_inches="tight")
print(("Prior widths =", F0_width, F1_width))
print(("Actual run time = {}".format(dT)))
print(("Actual number of grid points = {}".format(search.data.shape[0])))
for key in search_keys:
gridsearch.plot_1D(xkey=key, xlabel=labels[key], ylabel=labels["2F"])
print("Making GridSearch {:s} corner plot...".format("-".join(search_keys)))
vals = [np.unique(gridsearch.data[key]) - inj[key] for key in search_keys]
twoF = gridsearch.data["twoF"].reshape([len(kval) for kval in vals])
corner_labels = [
"$f - f_0$ [Hz]",
"$\\dot{f} - \\dot{f}_0$ [Hz/s]",
]
if sky:
corner_labels.append("$\\alpha - \\alpha_0$")
corner_labels.append("$\\delta - \\delta_0$")
corner_labels.append(labels["2F"])
gridcorner_fig, gridcorner_axes = pyfstat.gridcorner(
twoF, vals, projection="log_mean", labels=corner_labels, whspace=0.1, factor=1.8
)
gridcorner_fig.savefig(os.path.join(outdir, gridsearch.label + "_corner.png"))
plt.close(gridcorner_fig)
print("")
print("Performing MCMCSearch...")
# set up priors in F0 and F1 (over)covering the grid ranges
if sky: # MCMC will still be fast in 4D with wider range than grid
DeltaF0 *= 50
DeltaF1 *= 50
theta_prior = {
"F0": {
"type": "unif",
"lower": inj["F0"] - DeltaF0 / 2.0,
"upper": inj["F0"] + DeltaF0 / 2.0,
tref,
tstart,
tend,
)
search.run()
print("Plotting 2F(F0)...")
search.plot_1D(xkey="F0", xlabel="freq [Hz]", ylabel="$2\\mathcal{F}$")
print("Plotting 2F(F1)...")
search.plot_1D(xkey="F1")
print("Plotting 2F(F0,F1)...")
search.plot_2D(xkey="F0", ykey="F1", colorbar=True)
print("Making gridcorner plot...")
F0_vals = np.unique(search.data["F0"]) - F0
F1_vals = np.unique(search.data["F1"]) - F1
twoF = search.data["twoF"].reshape((len(F0_vals), len(F1_vals)))
xyz = [F0_vals, F1_vals]
labels = [
"$f - f_0$",
"$\\dot{f} - \\dot{f}_0$",
"$\\widetilde{2\\mathcal{F}}$",
]
fig, axes = pyfstat.gridcorner(twoF,
xyz,
projection="log_mean",
labels=labels,
whspace=0.1,
factor=1.8)
fig.savefig(os.path.join(outdir, label + "_projection_matrix.png"))