Skip to content

Added calc_chisq method to fix issue with calculated chi2 #63

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
wbalmer merged 5 commits into from
May 21, 2025
Merged

Added calc_chisq method to fix issue with calculated chi2 #63

Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
06672c3
Added calc_chisq method to fix issue with calculated chi2
tomasstolker May 21, 2025
d8a7114
Fixed typo
tomasstolker May 21, 2025
bd76575
Another minor fix
tomasstolker May 21, 2025
838372a
Another fix
tomasstolker May 21, 2025
7606b73
More fixing
tomasstolker May 21, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
41 changes: 39 additions & 2 deletions backtracks/backtracks.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -475,6 +475,43 @@ def fmodel(self, param):
"""
return radecdists(self, self.epochs_tt, param)

def calc_chisq(self, param):
"""
Function to calculate the chi2 for correlated (e.g., GRAVITY) and uncorrelated datapoints given certain model parameters.

Args:
param (np.array of float): set of parameters describing 5D position of background star and 6D position of host star.

Returns:
Total chi2 given the observed datapoints and a set of model parameters.
"""

if not np.isfinite(np.sum(param)):
# if nan or inf, avoid calling fmodel
return -np.inf

xs, ys = self.fmodel(param)

# separate terms where there is a correlation
corr_terms = ~np.isnan(self.rho)

chi_sq = 0.0

if np.sum(~corr_terms) > 0:
# calculate chi2 for uncorrelated terms
chi_sq += np.sum((self.ras[~corr_terms] - xs[~corr_terms])**2 / self.raserr[~corr_terms]**2)
chi_sq += np.sum((self.decs[~corr_terms] - ys[~corr_terms])**2 / self.decserr[~corr_terms]**2)

if np.sum(corr_terms) > 0:
# calculate the chi2 for correlated terms
# following equations with correlation terms taken from orbitize! routine
# _chi2_2x2cov (https://orbitize.readthedocs.io/en/latest/modules/orbitize/lnlike.html)
det_C = (self.raserr[corr_terms]**2)*(self.decserr[corr_terms]**2)*(1-self.rho[corr_terms]**2)
covs = self.rho[corr_terms]*self.raserr[corr_terms]*self.decserr[corr_terms]
chi_sq += np.sum(((self.ras[corr_terms]-xs[corr_terms])**2*self.decserr[corr_terms]**2+(self.decs[corr_terms]-ys[corr_terms])**2*self.raserr[corr_terms]**2-2*(self.ras[corr_terms]-xs[corr_terms])*(self.decs[corr_terms]-ys[corr_terms])*covs)/det_C)

return chi_sq

def loglike(self, param):
"""
Function to calculate Log likelihood for correlated (e.g., GRAVITY) and uncorrelated datapoints given certain model parameters.
Expand Down Expand Up @@ -725,7 +762,7 @@ def fit(self, dlogz=0.5, npool=4, dynamic=False, nlive=200, mpi_pool=False, resu

# Compute useful chi2 value
self.median_loglike = self.loglike(self.run_median)
self.median_chi2_red = -2.*self.median_loglike/((2*len(self.epochs))-self.ndim)
self.median_chi2_red = self.calc_chisq(self.run_median)/((2*len(self.epochs))-self.ndim)

return self.results

Expand Down Expand Up @@ -762,7 +799,7 @@ def load_results(self, fileprefix: str = './'):

# recompute useful chi2 value
self.median_loglike = self.loglike(self.run_median)
self.median_chi2_red = -2.*self.median_loglike/((2*len(self.epochs))-self.ndim)
self.median_chi2_red = self.calc_chisq(self.run_median)/((2*len(self.epochs))-self.ndim)

def generate_plots(
self,
Expand Down