Skip to content

Commit

Permalink
updated fit tests; added info matrix tracking
Browse files Browse the repository at this point in the history
  • Loading branch information
@rahil-makadia
rahil-makadia committed Oct 3, 2024
1 parent bb282ef commit fc5ab28
Show file tree
Hide file tree
Showing 5 changed files with 161 additions and 158 deletions.
101 changes: 55 additions & 46 deletions grss/fit/fit_simulation.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,7 @@
import sys
import datetime
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from astropy.time import Time

Expand Down Expand Up @@ -562,7 +563,7 @@ def __init__(self, x_init, obs_df, cov_init=None, n_iter_max=10,
self.fixed_propsim_params[key] = nongrav_info[key]
self.fixed_propsim_params['events'] = events if events is not None else []
self.prior_est = None
self.prior_sigs = None
self.prior_sig = None
self._priors_given = False
self._xbar0 = None
self._info0 = None
Expand All @@ -574,6 +575,7 @@ def __init__(self, x_init, obs_df, cov_init=None, n_iter_max=10,
num_auto_rejected = np.sum(sel_ast == 'D')
num_force_rejected = np.sum(sel_ast == 'd')
self.num_rejected = num_auto_rejected + num_force_rejected
self.info_mats = []
self.converged = False
return None

Expand Down Expand Up @@ -642,10 +644,10 @@ def _check_priors(self):
ValueError
If the prior estimates and sigmas do not have the same length.
"""
# check that the keys in self.x_init are the same as in self.prior_est and self.prior_sigs
if self.prior_est is not None and self.prior_sigs is not None:
# check that the keys in self.x_init are the same as in self.prior_est and self.prior_sig
if self.prior_est is not None and self.prior_sig is not None:
self._priors_given = True
if len(self.prior_est) != len(self.prior_sigs):
if len(self.prior_est) != len(self.prior_sig):
raise ValueError("Prior estimates and sigmas must have the same length.")
for key in self.prior_est.keys():
if key not in self.x_init:
Expand All @@ -654,7 +656,7 @@ def _check_priors(self):
self._xbar0 = np.zeros(self.n_fit)
for key in self.prior_est.keys():
idx = list(self.x_init.keys()).index(key)
self._info0[idx, idx] = 1/self.prior_sigs[key]**2
self._info0[idx, idx] = 1/self.prior_sig[key]**2
self._xbar0[idx] = self.prior_est[key] - self.x_init[key]
self._prior_constant = list(self.x_nom.values())+self._xbar0
return None
Expand Down Expand Up @@ -693,56 +695,60 @@ def _add_simulated_obs(self):
raise ValueError("Simulated observation data must have the same length.")
perm_id = self.obs['permID'][0]
prov_id = self.obs['provID'][0]
new_sim_obs = []
for i, time in enumerate(times):
idx = len(self.obs)
this_obs = {}
mode = modes[i]
if mode not in {'SIM_CCD', 'SIM_OCC', 'SIM_RAD_DEL', 'SIM_RAD_DOP'}:
raise ValueError(f"Unknown simulated observation mode {mode}.")
weight = weights[i]
if isinstance(perm_id, str):
self.obs.loc[idx, 'permID'] = f'SIM_{perm_id}'
this_obs['permID'] = f'SIM_{perm_id}'
if isinstance(prov_id, str):
self.obs.loc[idx, 'provID'] = f'SIM_{prov_id}'
self.obs.loc[idx, 'obsTime'] = f'{time.utc.isot}Z'
self.obs.loc[idx, 'obsTimeMJD'] = time.utc.mjd
self.obs.loc[idx, 'obsTimeMJDTDB'] = time.tdb.mjd
self.obs.loc[idx, 'mode'] = mode
this_obs['provID'] = f'SIM_{prov_id}'
this_obs['obsTime'] = f'{time.utc.isot}Z'
this_obs['obsTimeMJD'] = time.utc.mjd
this_obs['obsTimeMJDTDB'] = time.tdb.mjd
this_obs['mode'] = mode
if mode in {'SIM_CCD', 'SIM_OCC'}:
self.obs.loc[idx, 'stn'] = 'SIM'
self.obs.loc[idx, 'ra'] = np.nan
self.obs.loc[idx, 'rmsRA'] = weight[0]
self.obs.loc[idx, 'sigRA'] = weight[0]
self.obs.loc[idx, 'dec'] = np.nan
self.obs.loc[idx, 'rmsDec'] = weight[1]
self.obs.loc[idx, 'sigDec'] = weight[1]
self.obs.loc[idx, 'rmsCorr'] = weight[2]
self.obs.loc[idx, 'sigCorr'] = weight[2]
self.obs.loc[idx, 'cosDec'] = 1.0
this_obs['stn'] = 'SIM'
this_obs['ra'] = np.nan
this_obs['rmsRA'] = weight[0]
this_obs['sigRA'] = weight[0]
this_obs['dec'] = np.nan
this_obs['rmsDec'] = weight[1]
this_obs['sigDec'] = weight[1]
this_obs['rmsCorr'] = weight[2]
this_obs['sigCorr'] = weight[2]
this_obs['cosDec'] = 1.0
elif mode in {'SIM_RAD_DEL', 'SIM_RAD_DOP'}:
self.obs.loc[idx, 'trx'] = 'SIM'
self.obs.loc[idx, 'rcv'] = 'SIM'
self.obs.loc[idx, 'com'] = 1
this_obs['trx'] = 'SIM'
this_obs['rcv'] = 'SIM'
this_obs['com'] = 1
if mode == 'SIM_RAD_DEL':
self.obs.loc[idx, 'delay'] = np.nan
self.obs.loc[idx, 'rmsDelay'] = weight[0]
self.obs.loc[idx, 'sigDelay'] = weight[0]
self.obs.loc[idx, 'doppler'] = np.nan
self.obs.loc[idx, 'rmsDoppler'] = np.nan
self.obs.loc[idx, 'sigDoppler'] = np.nan
this_obs['delay'] = np.nan
this_obs['rmsDelay'] = weight[0]
this_obs['sigDelay'] = weight[0]
this_obs['doppler'] = np.nan
this_obs['rmsDoppler'] = np.nan
this_obs['sigDoppler'] = np.nan
else:
self.obs.loc[idx, 'delay'] = np.nan
self.obs.loc[idx, 'rmsDelay'] = np.nan
self.obs.loc[idx, 'sigDelay'] = np.nan
self.obs.loc[idx, 'doppler'] = np.nan
self.obs.loc[idx, 'rmsDoppler'] = weight[0]
self.obs.loc[idx, 'sigDoppler'] = weight[0]
self.obs.loc[idx, 'frq'] = 8560.0
self.obs.loc[idx, 'ctr'] = 399
self.obs.loc[idx, 'sys'] = 'ITRF'
self.obs.loc[idx, 'pos1'] = 0.0
self.obs.loc[idx, 'pos2'] = 0.0
self.obs.loc[idx, 'pos3'] = 0.0
self.obs.loc[idx, 'selAst'] = 'a'
this_obs['delay'] = np.nan
this_obs['rmsDelay'] = np.nan
this_obs['sigDelay'] = np.nan
this_obs['doppler'] = np.nan
this_obs['rmsDoppler'] = weight[0]
this_obs['sigDoppler'] = weight[0]
this_obs['frq'] = 8560.0
this_obs['ctr'] = 399
this_obs['sys'] = 'ITRF'
this_obs['pos1'] = 0.0
this_obs['pos2'] = 0.0
this_obs['pos3'] = 0.0
this_obs['selAst'] = 'a'
new_sim_obs.append(this_obs)
sim_obs_df = pd.DataFrame(new_sim_obs)
self.obs = pd.concat([self.obs, sim_obs_df], ignore_index=True)
return None

def _parse_observation_arrays(self, obs_df):
Expand Down Expand Up @@ -1759,6 +1765,7 @@ def _get_lsq_state_correction(self, partials, residuals):
atwb = np.zeros(self.n_fit)
j = 0
sel_ast = self.obs['selAst'].values
self.info_mats = []
for i, obs_info_len in enumerate(self.observer_info_lengths):
if obs_info_len in {4, 7}:
size = 2
Expand All @@ -1768,10 +1775,12 @@ def _get_lsq_state_correction(self, partials, residuals):
raise ValueError("Observer info length not recognized.")
if sel_ast[i] in {'D', 'd'}:
j += size
self.info_mats.append(atwa.copy())
continue
atwa += partials[j:j+size, :].T @ self.obs_weight[i] @ partials[j:j+size, :]
atwb += partials[j:j+size, :].T @ self.obs_weight[i] @ residuals[i]
j += size
self.info_mats.append(atwa.copy())
# use pseudo-inverse if the data arc is less than 7 days
if self.obs.obsTimeMJD.max() - self.obs.obsTimeMJD.min() < 7.0:
self.covariance = np.linalg.pinv(atwa, rcond=1e-20, hermitian=True)
Expand Down Expand Up @@ -2074,7 +2083,7 @@ def save(self, filename):
f.write(f" {units_dict[key[:2]]}")
if self._priors_given and key in self.prior_est:
p_val = self.prior_est[key]
p_sig = self.prior_sigs[key]
p_sig = self.prior_sig[key]
if key in ['om', 'w', 'i']:
p_val *= 180/np.pi
p_sig *= 180/np.pi
Expand Down Expand Up @@ -2114,7 +2123,7 @@ def save(self, filename):
f.write(f" {units_dict[key[:2]]}")
if self._priors_given and key in self.prior_est:
p_val = self.prior_est[key]
p_sig = self.prior_sigs[key]
p_sig = self.prior_sig[key]
if key in ['om', 'w', 'i']:
p_val *= 180/np.pi
p_sig *= 180/np.pi
Expand Down
60 changes: 30 additions & 30 deletions tests/python/fit/chesley.ipynb
View file

Large diffs are not rendered by default.

34 changes: 17 additions & 17 deletions tests/python/fit/eggl.ipynb
View file

Large diffs are not rendered by default.

66 changes: 30 additions & 36 deletions tests/python/fit/farnocchia.ipynb
View file

Large diffs are not rendered by default.

58 changes: 29 additions & 29 deletions tests/python/fit/shantanunaidu.ipynb
View file

Large diffs are not rendered by default.

0 comments on commit fc5ab28

Please sign in to comment.