This is an outline of the calibration pipeline for processing ARGUS data taken with the GBT. The pipeline approach relies on two packages that I maintain:

In interferometry, deconvolving the image replaces the interferometer’s ugly point spread function (PSF) with a nice well-behaved Gaussian. The original PSF is formally the transform of where the samples are found in the Fourier ($uv$) plane. Since changes in frequency and the relative weights of antennas can affect the importance or even presences of certain samples in the $uv$ plane, the interferometer PSF varies across an image mosaic and with frequency. This means that, once deconvolved, the cleaned up beam is also a function of frequency and position. CASA does an excellent job of tracking all these effects in detail. However, it means that the resolution of the image varies through an image. This can be non-ideal. In theory, you can deconvolve the image with a restoringbeam='common' call, but current versions of CASA are buggy. My student Eric K. found the following code snippet. It builds a fake image with elliptical resolution elements. The beam position angle rotates and the order of the beams varies in the cube… but the cube has the same resolution elements.

Radio interferometry is amazing. Nobel prize winning amazing since it turns the weak floppy photons of the radio spectrum into powerhouses of resolution. However, there remains a sad fact of life that interferometers by themselves do not measure all the information in an image. Specifically, interferometers measure Fourier components of the image. But they are poorly suited to actually measuring Fourier components with small wave numbers ($\mathbf{k} \sim 0$). These spatial frequencies are typically measured by single dish telescopes and then… somehow… stitched together into single images. In theory, this is a well understood process, but there is a complication. In addition to only measuring large wave numbers, interferometers also only measure certain parts of the Fourier plane and so we need to… somehow… interpolate a complex function into places where it is not measured.

With a student, I’ve been working of developing structure function and correlation function approach to analyzing spectral line data cubes. First, we’ve been working with the structure function of order $p$:

In Winter 2017, I will be teaching a PHYS 495/595 course on Star Formation. The course is intended for senior undergraduates and graduate students.

When dealing with spectral line data cubes, we will often use moments as simple descriptions of the emission in the data set. This is an aid to understanding what is happening inside the three dimensional data volume of the cube. Let’s set up a little formalism. Consider the spectral line data cube to be a map of emission on the sky, with two spatial coordinates and one spectral coordinate. Since moments only make sense when dealing with spectral line data, let us further assume that the spectral coordinate is a line-of-sight velocity calculated from the Doppler formula. Thus our cube is $I(x,y,v)$. A single spectrum at position $(x_0, y_0)$ will be denoted $T(v)$, dropping the spatial coordinate for brevity.

A good way of keeping track of heterogeneous data is to use an astropy table. This allows you to track a variety of different types of data in a single, manipulable format. There are a bunch of other approaches to this from python dictionaries or fancy things like PANDAS data frames. Astropy tables are good because they interface neatly with FITS binary tables, a common format in our field. There is also some minimal LaTeX support should you need it for journal articles.

If I’ve told you to read this post, you’re at a point where you have some preliminary results and it’s time to start thinking about how to present them. There are two components to this process. First, is the nuts-and-bolts of how to write up text that can be used for a thesis or journal article. The second is the matter of writing and making figures in a scientific style, suitable for the consumption by nerds.

In the continuing quest for ammonia column densities, we return to the question of how to turn observables into the total optical depth for a line. We have several options in front of us, all of which are nominally equivalent.

Many populations of astronomical objects follow power-law distributions in their proprties, also known as Pareto distributions to pretty much everyone else in the statistical world. Of note the masses of stars produced by the star formation process follows an initial mass function (IMF). When first calculated, the functional form of the IMF appears to follow a power-law distribution, though subsequent work has started to argue about whether this is actually the case. A reason for this debate is that physical processes, notably turbulence, can easily produce a log-normal distribution and it is difficult to distinguish between these two cases in the data. Lately, it appears that turbulence produces the log-normal and then gravity amplifies the positive wing of the log-normal to create a power-law. This leads to the truly awesome power-law-log-normal distributions which are finally getting some attention in astronomy (see Basu et al., 2015, Brunt et al., 2015).

Translating from sky coorindates (angles) to a position in a galaxy’s disk requires doing some clever three-dimensional geometry. This geometry is easy to get wrong, especially for galaxies that are large on the sky so that the curvature of the celestial sphere starts to become problematic. To this end, I’ve started work on a python package that provides methods to make this more tractable. Ultimately the galaxies package will give lots of galatic properties, but for now it just calculates position in the galaxy’s disk given inputs.

I am not a sysadmin. In fact, I’m every sysadmin’s worst nightmare: I’m a power user that has access to sudo. I run a few machines and when there are problems, I follow the xkcd approach. This leads to me copy-pasting in commands that I found over google, usually on stackexchange. This is how science advances. This post details my experience trying to run a multi-user OpenStack project here. The basic issue is that we have one (1) public IP address for our VMs and many users and instances. There are two possible approaches. The nice route would be to just configure the virtual router on the openstack cloud to do this forwarding automagically. Alternatively, we could make a routing VM where we can address this directly.

The main output of CPROPS and cpropstoo are tables of cloud parameters. In astronomy, a common table file format is that of the FITS table where FITS refers to the Flexible Image Transport System. The original FITS format was designed for images, but it has been extended in hundreds of different ways including for tables and catalogs. We can explore the table from one of two approaches. The first is in an exploratory mode, and I recommend using the programs TOPCAT or glue. The second mode is more making proper plots and doing good analysis using python. If you are given a binary table (of GMC properties), here’s how to start exploring.

This post presents a quick introduction to the relevant parts of the literature to give a crash course on molecular cloud properties, a mainstay of my research.

This is a summary of how to get started in my research group. At present, we use a few different technologies for carrying out our science and for documenting what we do. Of note, we will be using python and several packages in a standard python distribution. This allows us to analyze numerical data using a suite of Extremely Useful packages. We will also use a version control system called git. Git allows us to develop software collaboratively and keep a good record of how the projects are proceeding. This document is an introduction to these two packages. All this will usually operate on a linux server, and I’ve put an introduction to linux servers here.

We are going to use the Linux computing environment for the analysis of our data. Linux is extremely useful because a lot of scientific software is built to run on Linux. You’ll be connecting to a linux computer that I manage. If you have issues with the computer, the default action is to send me an email and I will try to sort it out.

In ASTRO 320, we motivated a cartoon view of degeneracy pressure based on filling states of phase space. We argued that a one-dimensional phase space can be separated into cells of phase volume equal to Planck’s constant $h$. Into each cell, we can pack at most 2 fermions (usually electrons), where the 2 comes from the two spin states of the electron. This means that in a three dimensional phase space, the cell volumes are $h^3 = \Delta x \Delta y \Delta z \Delta p_x \Delta p_y \Delta p_z$, and each cell can hold the two electrons.

In teaching stellar evolution for third year students, there are a few things that link tightly to the physics covered earlier in the class, and are worth highlighting.

The question came up as to whether you can do beam nodding on a telescope that doesn’t explicitly support beam nodding. Under some assumptions, it seems possible to do it with the calibrated scans alone. If we define $N$ to be the ON scan, $F$ to be the OFF scan then the usual calibration gives

In third-year astronomy, I’ve steered the homework straight into integrals of the form

Here are a few questions about sky motions to give practice to students in first-year astronomy. Recall that the latitude of Edmonton is $53.5^{\circ}$.

The effective temperatures of planets can be derived under the assumptions of their energy input coming solely from the Sun. The relations in Essential Astrophysics are derived from the assumptions of energy balance. First, we assume that the planets are in thermal equilibrium so that the energy into the planet equals the energy out of the planet $E_{\mathrm{out}}=E_{\mathrm{in}}$.

The documentation to CLASS is opaque. Thus, trying to find how to convert LMV to FITS files is tricky. Adam Ginsburg solved this, of course, and I simply consulted the Codex Ginsburgium.

I’ve adopted Kenneth Lang’s Essential Astrophysics for my ASTRO 320 class at U. Alberta this year. It has the wonderful feature of being relatively low cost owing to the University Library subscribing to the Springer eBooks. Thanks, Library. Unfortunately, the book is a First Edition. As such there are typos. Here is a list of the ones we’ve found so far.

Eric asked for a list of galaxies with position angles and inclinations. While I have a nice list for internal use, Adam and Karin have put together an extensive set of information here. This does beg the question of how to do a good table join and get out the information from a huge set of tables. After much munging around with astropy.table, I managed to extract things using this.

Despite good attempts to get the UNWISE data working, the data turn out to be poorly suited to our application. We don’t need high resolution and the lack of the zero-spacing information causes a suite of problems. Thus, we should probably use ALLWISE and a cursory look at the data suggests that they are better calibrated and don’t suffer from spatial filtering.

This is a small fragment developed by Adam Leroy for the data of UNWISE.

To finish up extragalactic star formation rates, we need H$\alpha$ images of the Galactic plane. Here, the IPHAS survey does the best job of providing coverage thanks to the VTSS going quietly into the night. The latest release is DR2. The data are photometric, but they are in optical people units. Specifically, every image has a well defined PHOTZP such that, in the usual way,

Does math work here: \(\int x^2\, dx\)