Time Series Analysis

Software for performing Time Series Analysis (targeted at astrophysical applications) written in C. The package contains modules written in Cython and Python as well.

Author: Jakob Rørsted Mosumgaard (jakob@phys.au.dk).

Features

Core software written in C:

• Program to make a power spectrum (the fourier transform of a time series) with and without statistical weights. Option to calculate the spectral window.
• Program to run the CLEAN algorithm on a time series.
• Program to filter data using band-, high- and low-pass filters.
• The software is using OpenMP for a performance boost using multithreading.

Extra features:

• Stand-alone Cython-module, which is providing a Python interface to the fast C function for calculation of the power spectrum. The interface has a very low overhead and almost as fast runtimes as the pure C.
• Pure Python/NumPy implementation of the algorithm for comparison (and for easy-to-read overview of the algorithm).
• Python script to generate artificial test data for easy verification.

Installation

Requirements:

• GCC (tested with v5.3.0)
• Python 3 (tested with v3.5.1)
• Cython (tested with v0.24)
• NumPy (tested with v1.11.0)
• Gnuplot (tested with v5.0; not a strict requirement, only for running tests)

Setup: The whole project is controlled by several Makefile. Note that the Makefiles assumes working in a virtual environment, where python is calling python3. The targets available from the root-dir are:

• make will build the C-executable.
• make data will create artificial data for testing.
• make test will run the abovementioned targets and make a test-run and a plot.
• make cython will compile the stand-alone Cython module.

Usage

The usage of the software is documented in the different files; it should be straigtforward for both C and Python/Cython.

File structure

The project contains the following directories:

• cython: Cython module and a Python wrapper.
• python: Pure Python implementation.
• source: C source code.
• test: Default test-target.
• testdata: Generation and storage of artificial data.

Furthermore, the following directories should be created:

• output: Default place for output from all programs. Automatically created by make test.