This repo contains a C implementation of the transit search algorithm
of Aigrain & Irwin (2004), and a Python interface to it.

MyBLS is very similar to the well-known box-least squares (BLS)
algorithm of Kovacs, Zucker & Mazeh (2002). The main differences are
a) that the out-of-transit level is assumed to be the light curve
median rather than being re-computed at every step, and b) that the
sums involved in the transit detection statistic are computed first
for single transits, and then combined for each trial period /
epoch. These modifications both speed up the code without altering the
performance signficantly. 

Additionally, the range of transit durations to search is specified in
units of the duration that would be expected for an equatorial transit
of a planet in front of a Sun-like stars, rather than in hours. This
means that different actual durations are explored for different trial
periods, if the period range is broad.

To compile and test the code:
 > make -f makefile.c
 > ./bls
for the C version, and 
 > make -f makefile.py
 > ./MyBLS_PY_example.py
for the Python version. The two versions should hopefully produce the same output.

The python wrapper is contained in MyBLS_PY.c and MyBLS_PY.h, which
creates the library (aka python module)
MyBLS_PY.so. MyBLS_PY_example.py is an example python script which
calls the MyBLS python function (which is mybls_py in the C file).

To get the code to compile, you may have to change the include
directories in the makefile which point to the relevant C header files
Python.h and numpy/arrayobject.h.