Having an up-to-date machine and not having to reinstall software at each OS upgrade is considered a chimera. If one wants an up-to date system with the shiny new version of the presentation software, or the latest upgrade of that linux graphical environment, the tradeoff is often having to reinstall all the data analysis software, because the system libraries, because permissions, because whatever. So, shiny new system with periodic full upgrade of the analysis software, or stable, long-term-support boring version of the OS, maintaining software compatibility? And what about incompatibilities with new software? A trick might be using a virtual machine. You fire up VirtualBox, you get an Ubuntu ISO, you install it on a 20-GB virtual disk, and you install all your software there. Point is, the VM will take up half of your RAM if you want some performance, mostly because of duplicate and useless things such as the desktop environment and tons of additional programs (I guess you are not using firefox and thunderbird on your virtual machine, when you can run them from the host OS). In the last few years, I've used Travis CI for unit testing, and I heard about *Containers*. Containers are much lighter than virtual machines, they contain only the minimum pre-installed software needed, and you choose what to install in them. Here, I will not give a full account of the possibilities that Docker offers. Rather, I will throw down a few notes on how to obtain a working Linux container with your software installed inside, that you can fire up whenever you need it. Add a comment

MaLTPyNT logo

MaLTPyNT, Matteo's Library and Tools in Python for NuSTAR Timing [ascl:1502.021], is a set of Python (2.7, 3.3, 3.4) scripts designed to perform correctly and fairly easily a quick-look timing analysis of NuSTAR (and partially XMM-Newton/EPIC and RXTE/PCA) data.

It employs fairly standard procedures to perform the Fourier analysis of data and calculate periodograms. Data are automatically extracted from Good Time Intervals if specified, in order to avoid the spurious contributions to the periodograms arising from bad intervals.

Also, when data from two detectors are used (like for the two focal planes of NuSTAR) it gives the possibility to calculate the cospectrum (the real part of the cross-spectrum). This is necessary for NuSTAR observations of bright sources: in these observations, dead time effects produce a distortion in the power spectrum (in particular in the Poisson noise baseline) that is not easy to model and makes it very tricky to calculate the real source periodogram. The cospectrum, by eliminating all correlations between the two detectors, permits to have a noise-subtracted periodogram.

The software is released under the BSD license. It can be downloaded from the two repositories below:

Bitbucket: https://bitbucket.org/mbachett/maltpynt

GitHub: https://github.com/matteobachetti/MaLTPyNT/

The documentation is available here: http://maltpynt.readthedocs.org/

More information

Indexed in NASA ADS: http://adsabs.harvard.edu/abs/2015ascl.soft02021B

ASCL entry: http://ascl.net/1502.021

Continuous Integration with Travis CI: https://travis-ci.org/matteobachetti/MaLTPyNT/

Add a comment