Changelog

Changelog#

11-Mar-2026: version 3.3.13#

  • numba acceleration for the occultquad transit model (significant speed-up)

  • improved binning function

  • allow resuming emcee sampling at any stage of a previous sampling run

  • included shared_params in rerun_result function

  • allow to decontaminate TESS and Kepler data based on CROWSAP keyword

  • corrected indices of customRV parameters

  • fixed spleaf GP omission in get_decorr; fixed bug when decorr parameter list is empty; fixed step size for f1ev with emcee sampler

  • sorting option for RVs

  • get_TESS_data: improved flux selection and added mask_points method for masking data based on time and flux ranges

  • fix load_lightcurves to use zeros for missing columns when column is absent in the input file

  • fixed ValueError when calling np.where on empty GPindex/rvGPindex (0d array) when no GPs are configured

  • CI: updated GitHub Actions to Node.js 24 (setup-python@v5, FORCE_JAVASCRIPT_ACTIONS_TO_NODE24); added macOS and Windows wheel builds

27-Aug-2025: version 3.3.12#

  • renamed Transit_Model and RadialVelocity_Model to Planet_LC_Model and Planet_RV_Model respectively

  • added exclude+pars argument in get_decorr() to exclude specific parameters from the decorrelation model

  • allow loading lightcurves/rvs from a dictionary instead of from files

  • included kwargs for getting cheops data taking into account the SAA, Earth, Moon, Sun and CR limits

  • can now get K2 data from MAST using CONAN.get_files.get_K2_data

  • implemented multiseries (multidimensional) GP fitting for RVs using spleaf

  • expanded george multidimensional GP fitting which was not working for more than 2 dimensions

  • more robust and scalable alternative for .dat config file. config file can now be saved and loaded as yaml file.

27-May-2025: version 3.3.11#

  • including GPkernels from with more than 2 parameters: quasiperiodic, expsine2, SHO,…

  • allow to pick between logUniform or uniform prior for GP params and also rho_star and Duration. default is logUniform

  • added option to chose planet phase variation model: cosine or lambert

  • elliposidal variation model now include fractional amplitude of the other cosine harmonics

  • added draft of Joss paper

  • imporved documentation with full API reference

  • notebook updates

28-Feb-2025: version 3.3.8#

  • implemented multiprocessing for plotting figures

29-jan-2025: version 3.3.3#

  • improved generation of smooth time array (now at 1min sampling) for plotting best fit model

  • result object now allows to get baseline model and the stdev using result.lc.get_baseline() and result.rv.get_baseline()

  • fixed binning and offset issues in ttv lc plot

20-Jan-2025: version 3.3.2#

  • added capability for different parameters to share the same value.

e.g. shared_params = {‘lc1_logjitter’: [‘lc2_logjitter’, ‘lc3_logjitter’, ‘lc4_logjitter’]} * Quality factor Q of the SHO kernel can be fixed to a value using lc_obj._Qsho or rv_obj._Qsho * when multiplying Gp kernels, amplitude of the second kernel should be set to -1 to deactivate it * fixed bug in post-sampling GP prediction when “same” GP is used for lcs * added reflective boundary condition for impact parameter when sampling with dynesty

18-Jan-2025: version 3.3.1#

  • changed pause statement in occultquad.f to print warning so as not to pause program execution

  • added sesinw-secosw parameterization for planet model

  • added filter dependent GP fit

6-Dec-2024: version 3.3#

  • conversion functions (e.g., rho_to_aR) can take uncertainties.ufloat objects as input

  • custom func can be given as a class

  • automatic linking of lc_obj and rv_obj with seamless parameter sharing

  • fix for installation on python>=3.10

  • added .detrend_data() method to remove best fit decorr model from data.

  • added option to specify whether to fit jitter and offset for each light curve

  • fixed bug in transit model for eccentric orbits

  • added spleaf for GP fitting

  • reorganized result folder: meidan and max model plots are now saved in med/ and max/ folders respectively. *out.dat files in out_data/ folder

  • setup_phasecurve() method of the lc object now changed to phasecurve()

  • corrected ecc anomaly calculation to be accurate for large eccentricities

  • added tests

12-July-2024: version 3.2.2dev#

  • added light travel time correction to the lightcurve model, user only required to set the stellar radius

  • phase curve model properly account for orbital eccentricity.

  • users can define a custom function which can be used to modify or replace the native CONAN lightcurve model. lc_obj.add_custom_LC_function()

  • changed import pickle to import dill as pickle which allows to pickle more complex objects like functions.

  • allow masking data points with user-defined condition. e.g. lc_obj.mask_data(lc_list=”all”, condition=”lc[‘col0’]<lc[‘col0’][10]”) to mask first 10 data points.

  • allow saving of modified injested data. e.g. lc_obj.save_LCs(save_path=”data_preproc/”)

  • calculation of AIC,BIC and chisqr now take into account the jitter if added to the errorbars

  • allow installation of CONAN when fortran compiler is not available, in which case python implementation of the transit model is used.

  • extra fixed arguments can be passed to custom function

  • accounted better for eccentricity in different parameter conversions (rho_to_aR, aR_to_rho, Tdur_to_aR, aR_to_Tdur, etc)

  • fixed bug in lc_out.dat file where the gp baseline was not properly subtracted to create the detrended LC

  • new lc_obj.add_sinusoid() function to fit a sinusoidal model to the lc baseline

  • added gp component to get_decorr function

  • contamination factors can now be setup as fitting parameters.

  • minor correction for TTV model of multiplanet sysyem when transits overlap

  • new CONAN.compare_results() class with methods that allows to compare results from different fits.

  • implemented cosine kernel in celerite

3-Jun-2024: version 3.2.1#

  • added function to read the parameters values and errors from the result_**.dat file –> result.get_all_params_dict()

  • automatically save fit config file in the output folder as config_save.dat

  • input files no longer overwritten when modified within CONAN. CONAN injests the file and works with a copy of it.

  • fixed bug in splitting filenames with multiple dots

  • selection of system parameter reference when using ‘ps’ table of the NASA exoplanet archive

  • changed reported stellar density unit from rho_sun to g/cm3

  • outlier rejection can be performed on selected columns of the data. e.g. lc_obj.clip_outliers(clip=4,width=11,selected_column=[“col1”,”col3”,”col4”])

  • when Downloading TESS data, user can now split the sector into orbits to create different file that can be detrended differently

  • fixed bug in writing result files when duration is a fixed parameter

7-May-2024: version 3.2.0#

  • added detrend argument to .plot() function to plot the detrended data and model. also adjusted space betweem data and residuals

  • fixed bug in mutiprocessing which was not closing the pool after fitting

  • added run_kwargs and dyn_kwargs arguments to the .run_fit() function to pass optional arguments to the emcee/dynesty sampler

  • remove nan rows when loading data.

  • corrected box-shaped occultation model by properly scaling transit model for occultations.

  • modified binning function to account for large data gaps and also modified other plotting outlooks.

  • fixed bug in obtaining upper sigma of the best-fit model.

  • spline fitting can now take ‘r’ for knot_spacing argument in order to fit a single spline to the range of the data array.

  • celerite gp fit of non-sorted array (e.g roll-angle). Now the array is sorted before fitting and the result is sorted back to the original order

  • Added doppler beaming signal with amplitude A_db to phase curve model.

  • D_occ, A_atm, A_ev and A_db now given in ppm.

  • modified phase curve section of config file to print priors for D_occ, A_atm, ph_off, and A_ev side-by-side instead of in separate lines.

  • choose between duration or rho_star for transit model.

  • corrected formating issues in *out.dat files when significant digits are too many.

  • automatically determine limits on jitter terms and baseline parameters from the data

  • added dynamic nested sampling with dynesty

  • sampling can be resumed by passing resume_sampling=True to the .run_fit() function

  • changed lamdas argument in load_lightcurves to wl. lamdas is still accepted but will be deprecated in future versions.

  • support for nested output folder path.

  • allowed float clip values in clip_outliers() function

  • out.dat files now contain arrays of the different baseline components [base_para, base_spl, base_gp, base_total] and the residuals.

  • check that user actually set planet parameters before attempting fit.

23-Feb-2024: version 3.1.5#

  • specified version of ldtk to install (1.7.0)

  • added information about flux arrays names available in tess data

  • fixed bug in making dynesty traceplot

  • changed mean of gp from 1 to 0 to reduce correlation with the offset parameter. full baseline thus changes from gp*base to gp+base

  • modified GP parameters labels (Amp{n} and len{n}) to count from 1 instead of 0 where n is the kernel number

19-Feb-2024: version 3.1.4#

  • fixed 2D spline fit for lc and rv data

  • implemented TTV for multiplanetary systems, added test notebook

  • added column 8 for decorrelation of lcs

  • estimate of rms and jitter for lc_obj and rv_obj upon ingestion of lc and rv data

  • “auto” option for the limits of the lc and rv paramteric baseline parameters.

  • modified fit plots to only phasefold lcs of same filter

  • uniform prior on rho_star changed to loguniform following literature convention

  • added ellipsoidal variation amplitude, A_ev, to phase curve model

  • renamed planet atmospheric variation in phasecurve from A_pc to A_atm

  • new configfile version to ingest new inputs [ttv,A_ev,A_atm]

  • added dynesty trace plot to view exploration of parameter space

12-Feb-2024: version 3.1.3#

  • added estimate of rms and jitter for lc_obj and rv_obj

  • “auto” option for the limits of the lc and rv paramteric baseline parameters.

8-Feb-2024: version 3.1.2#

  • fit configfile from command line: >>> conanfit config_file.dat

  • added notebook for occultation fitting (KELT20b)

2-Feb-2024: version 3.1.0#

  • added function get_decorr() now available for lc and rv objects

  • get_decorr() can now use spline. useful when analysing CHEOPS data or to remove other long-term trend in the data

  • added function clip_outliers() and rescale_column_data() to lc object

  • download data directly from TESS archive using function get_tess_data()

  • changed transit_rv_pars() function of the lc object to planet_parameters()

  • get system parameters from NASA exoplanet archive

  • limb darkening model now uses kipping parameterization.

  • get LD priors from phoenix

  • automatically fit .FITS and .dat files from EULER pipeline

  • supersampling of long cadence data

  • improved results object that allows to re-evaluate the model at user-defined times and make plots

  • phase curve fitting

  • sampling also with dynesty, now the default sampler

  • reimplementation of GP, for lc and rv, including more kernels and combinations of kernels

  • support for multiplanet system in lc and rv. no defined limit on the number of planets

  • see example implementations in the Notebooks folder

20-11-2023: version 3.0.5(dev)#

  • added functions clip_outliers() and split_transits to the lc object.

  • get_decorr() now uses same planet parameter names [“Period”, “Duration”,”Impact_para”,”RpRs”,…]

  • get_decorr() also allows to exclude some columns from the decorrelation analysis while enforcing others

  • load_chain() has been renamed as load_result() which returns an object that contains all the results from the MCMC run.

  • improved results object allows to reevaluate the bestfit model at user-given times

  • use can now specify output directory for the results using the “out_folder” argument of CONAN.fit_data().

  • columns of the input data are now described by col0, col3, col4, etc. instead of the legacy xshift, yshift which dont necessarily describe the input columns

  • added shoTerm gp for celerite

  • added function get_decorr() to the rv object to find best baseline model for each rv data.

  • added 2D spline fit for decorrelation of lc and rv data

  • support for multiplanet system in lc and rv

28-12-2022: version 3.0.5#

  • fixed bug in fit_data.py when dt is set for RV

  • from Andreas: removed function call to grweights in fit_data.py, caused errors for TDVs

  • from Andreas: added some attributes (TO,P,dur) to the result object

25-07-2022: version 3.0.4#

  • added phases to the output files

  • fixed problem with RV jitter and gamma indexes when jit_apply = “n”

  • fixed RV filepath

  • added spline for roll-angle decorrelation (added roll and spline_fit columns to output file)

  • can obtain priors for limb darkening using ldtk

  • smooth sampled transit model in decorr plot

  • allow setting up lc object without any lc file.

  • create bin_data function in plots.py

22-07-2022: version 3.0.3#

  • Modified automatic steps assignment in planet_parameterss() function

  • allow fitting single LDC while keeping the other fixed

  • correction in celerite fitting

  • corrected setup_occultation() which erroneously showed fit=”y” when not fitting a lc for occultation

  • specify delta_BIC threshold for selecting parameters in get_decorr() function

  • reduced bounds on the offset parameter -> [0.9, 1.2]

  • white noise in celerite uses bounds: [-5,-12]

  • burn-in chains now saved as .png before running the production chain

  • increased max number of parameters for making cornerplot from 12 to 14

  • increase maximum of impact parameter to 1.5

  • pointing input errors back to the concerned functions/methods

  • fixed issue with RV jitter not jumping

  • corrected error for celerite when WN=”n”

10-07-2022: version 3.0.2#

  • changed format of quadratic limb darekening: now allows either gaussian or uniform priors. Gaussian still recommended

  • included 1D GP fitting using Celerite (~5X faster than with George)

    to use celerite gp for a lightcurve, use “ce” instead of “y” in function lc_baseline().

  • added function get_decorr() for light curves object to find best baseline model for each lc.

  • added function plot_burnin_chains() to the results object to see how the chains evolved during burn-in.

  • added function load_result_array() to load result array for customized plotting.

  • some notebooks in example folder

Contents