def __init__(self, t, f, fe, texp, tol, mdep):
self.t = t - np.min(t)
self.period = 10 * np.max(self.t)
self.texp, self.tol, self.mdep = texp, tol, mdep
# Sample some parameters.
self.q1, self.q2 = np.random.rand(), np.random.rand()
self.t0 = np.max(self.t) * np.random.rand()
self.tau = 0.05 + (1-0.05) * np.random.rand()
self.ror = 0.008 + (0.2-0.008) * np.random.rand()
self.b = np.random.rand()
# Generate a synthetic transit.
u1, u2 = 2*self.q1*self.q2, self.q1*(1-2*self.q2)
lc = transit.ldlc_simple(self.t, u1, u2, self.period, self.t0,
self.tau, self.ror, self.b, texp, tol, mdep)
# Inject the transit and normalize the data.
mu = np.median(f * lc)
self.f = f * lc / mu
self.fe = fe / mu
# Set up the results directory.
bp = self.basepath
try:
os.makedirs(bp)
except os.error:
pass
pl.plot(self.t, self.f, ".k", ms=4, alpha=0.6)
pl.savefig(os.path.join(bp, "data.png"))
def lnlike(self, lna, lns, q1, q2, t0, tau, ror, b):
kernel = ExpSquaredKernel(np.exp(lna), np.exp(lns))
gp = george.GaussianProcess(kernel)
gp.compute(self.t, self.fe)
u1, u2 = 2*q1*q2, q1*(1-2*q2)
lc = transit.ldlc_simple(self.t, u1, u2, self.period, t0, tau, ror, b,
self.texp, self.tol, self.mdep)
return gp.lnlikelihood(self.f - lc)
def model(p):
q1, q2, per, t0, tau, ror, b = p
u1, u2 = 2*q1*q2, q1*(1-2*q2)
return transit.ldlc_simple(t, u1, u2, per, t0, tau, ror, b,
texp, 1e-1, 4)
def model(t, fstar, q1, q2, t0, tau, ror, b, texp=54.2 / 86400.):
u1, u2 = 2*q1*q2, q1*(1-2*q2)
lc = ldlc_simple(t, u1, u2, period, t0, tau, ror, b, texp, tol, maxdepth)
return fstar * lc
def model(fstar, q1, q2, t0, tau, ror, b):
u1, u2 = 2*q1*q2, q1*(1-2*q2)
lc = ldlc_simple(t, u1, u2, period, t0, tau, ror, b, texp, tol, maxdepth)
return fstar * lc