gttrcr / correlator

correlator

Release and tag

Latest	Release	Tag	Win32	Win64	Linux	Link
*	Release	0.2	❌	✔️	✔️	releases/tag/0.2
	Pre	0.1	❌	✔️	✔️	releases/tag/0.1

Each release folder also contains 4 example datasets so you can immediately test the correct functioning of the program on real data: in case of errors, any issue is welcome 😉.

Introduction and motivation

Many datasets are composed of real scalar functions (scalar function for short) or are amenable to scalar functions. In the most general case, a scalar function is a function $f\colon\mathbb{R}^n\to\mathbb{R}$. Therefore an input csv file with $n+1$ columns is interpreted as a function $f\colon\mathbb{R}^n\to\mathbb{R}$.

Just add the csv files you want to analyze inside the executable folder and the program will start processing. Every execution of correlator do the following:

1. Look for every csv file in the execution folder.
2. Compute the best polynomial fit for every pair of input (on the y axis) and for every file (x with respect to y).
3. Compute fft for every input file.
4. Compute peaks in every fft.
5. Compute the best polynomial fit for the migration of peaks at point 4.

The results are placed inside the output folder. The folder contains many files (sometimes hundreds 😉). But the filename structure is pretty straightforward. Let $n$ be the number of input datasets, then consider the following table.

File name	File type	Category	Content
polyfit_filename	csv	polyfit	Coefficients of the polynomial that best fits the data contained within filename up to the polyfit_max_degree-th degree
polyfit	csv	polyfit	Coefficients of the polynomials that best fits the $y$-values (codomain) of every dataset up to the polyfit_max_degree-th degree
fft_filename	csv	FFT	FFT of filename
fft_peaks_filename	csv	FFT	First fft_peaks_number peaks of FFT of filename
peak_migration_$i$	csv	FFT and polyfit	Coefficients of the polynomial that best first the migration of the $i$-th peak. $i\in\left[1, n\right]$

upcoming features

1. fft of peaks at point 4.
2. multidimensional fft
3. Compute the best polynomial fit for the distribution of peaks at point 4.
4. correlator will listen to new data (by socket, by pipe, by stream, by api, ...) and compute real time statistics.

Dev section

setup

Launch ./setup.sh platform where platform is:

• win32 for windows 32 bits
• win64 for windows 64 bits
• linuxfor linux

It sets up the system, update and init all submodules, configure the platform and the compiler.

release

Launch ./release.sh. It releases every supported platform