PSD Module

This module provides easy-to-use tools for quick data visualization and spectral analysis.

Data must be stored on text, Numpy or HDF5 files, and all formats compatible with the standard numpy.loadtxt and numpy.load are accepted. First dimension, or rows, is used for time and second dimension, or columns, for series. The first column is always assumed to represent the times associated with each row.

Installation

Make sure that Python 3 is available on your machine, and run

pip3 install psd

The package is also available at https://pypi.org/project/psd/.

Getting Started

Time-Series Visualization

You can visualize time-series from Numpy or text files using

psd --time-series my_file.npy another_file.txt ...

You can read HDF5 files as well by specifying the path to the dataset inside your file using

psd --time-series my_hdf5_file.hdf5:mygroup/mydataset ...

Power Spectrum Estimation

To compute Power Spectral Density (PSD) estimates for each series using the Welch method, simply use

psd my_file.npy

You can specify the number of rows at the top of the files you want to skip using -s SKIPROWS option, the number of points per segment you want to use with -n NPERSEF option, or the windowing function using --window WINDOW.

For time-series visualization and spectral analysis, you can hide the legend with the --no-legend option, specify a title with --title TITLE, or save the output as a text file, a Numpy file or an image using -o OUTPUT. You can specify line and marker aspect using matplotlib notation with --aspect.

psd -s 500 -n 10000 --window nuttall my_file.npy --title "This is an example"

Format Conversion

You can easily convert from text files to Numpy binary files using the quick convert command-line tool included in this package, i.e.

convert my_file1.txt my_file2.text

To reverse the conversion and get a text file from a Numpy file, use the -r option. You can specify the output file name using -o OUTPUT.

convert -r my_numpy.npy -o my_text_file.txt

The tool can also remove original files as soon as they are converted if you specify the --delete or -d option.

convert -d file*.txt

Documentation

Other options are available, use psd --help or convert --help to show documentation.

Developped by Jean-Baptiste Bayle (APC/CNES/CNRS), bayle@apc.in2p3.fr.