koi-fpp

False positive probabilities for all KOIs in the Q1-Q17 (DR24) Kepler table. The one-stop-shop table for the summarized FPP results is fpp_final_table.csv.

If you're curious about individual KOIs and want to browse the diagnostic/result plots, please go here.

If you're really curious, all the data files produced by these calculations are currently hosted at Princeton. To browse the results from a given KOI, visit tigress-web.princeton.edu/~tmorton/koi-fpp/K?????.??, with the appropriate KOI identifier. These directories contain the entire vespa and isochrones outputs for every KOI (including the stellar posteriors for the single-, binary- and triple-star fits). If you have specific questions, please feel free to contact me.

Paper

The manuscript PDF can be generated by entering the document directory and running make.

The fpp-results-analysis.ipynb and starprop-analysis.ipynb notebooks contain all the commands to make the figures & tables that live in the document directory, with the exception of the Santerne+ (2015) comparison table, which is created by santerne_compare.ipynb.

Reproducing results

You can run your own FPP calculations using all of the same data and constraints that I used for this work. If you wish to do this, first test out the calculation for the KOI for which I provide example data in this repository (KOI-7016.01/Kepler-452b).

First, clone this repository and hop on in:

git clone https://github.com/timothydmorton/koi-fpp
cd koi-fpp

Then define some temporary enivronment variables (assuming you're calling this from inside this repository):

export KOI_FPPDIR="$PWD/example_data"
export JROWE_DIR="$PWD/example_data/photometry"

Then, set up the python environment using the environment.yml file in the top level of this repository, as follows. [If you do not have conda available, install miniconda first.]

conda env create -f environment.yml
source activate koifpp

This should install all the required packages. Now from within this environment you can run the following:

koifpp-config K07016.01

This will set up the vespa and isochrones config files in a K07016.01 directory under KOI_FPPDIR (if you haven't yet used isochrones, then be prepared for a few minutes' worth of stellar model downloads). It will also do the trapezoid shape fitting to the photometry. Take a look at the fpp.ini and star.ini files to see what the inputs look like. You can then run the calculation as follows:

cd $KOI_FPPDIR
calcfpp K07016.01

Note that the stellar parameter inference (and thus the FPP calculation) will be much more reliable if you have multinest/pymultinest installed. The calculations will run just using the default emcee sampler, but I stronly recommend using multinest becaues of the inherently multimodal nature of the problem. Follow these instructions to install.

In order to get the data necessary for all the KOI FPP calculations beyond this single example, you will need the following:

• TRILEGAL starfield simulations. Define a KOI_FPPDIR environment variable to be somewhere sensible to you, and unpack this starfield directory within that.
• Kepler photometry. This is in two separate tarballs (Zenodo max file size is 2Gb), so unpack them both and combine their contents (all the koi*.n sub-directories) into a single directory. Define a JROWE_DIR environment variable referring to this location.

Then you can run the koifpp-config and calcfpp commands for your favorite KOI(s).

Other Data

The data directory contains the following files:

• fpp_all.txt: concatenation of all the results.txt files from all the successful vespa calculations. [TDM note: created by summarize_fpp.py script in /tigress/tmorton/kepler]
• fpp_err.txt: concatenation of all the exceptions raised by all the failed vespa calculations. [TDM note: created by summarize_fpp.py script]
• ttvdata.txt: table of whether TTV information was used to create the folded Kepler photometry used for the trapezoid fits. [TDM note: created by compile_ttv.py]
• starprops_all.txt: summarized information from all the single-star isochrones fits to all the KOI stars. [TDM note: created by summarize_starprops.py after running starfit-summarize on all KOIs]
• positional_probability.csv: table of positional probability data from Steve Bryson. This should correspond to the "positional probability" table at the NASA Exoplanet Archive.

Running the make_finale_fpptable.py script uses the data from these files as well as the DR24 table to produce the fpp_final_table.csv file in the top level of this repository.