Project ThaiPASS: international outreach blending astronomy and Python

This portion of the summer school discussed the production of elements in stars—from low-mass stars like the Sun to massive core-collapse supernova explosions. Lectures covered all of the major burning regimes found in main sequence stars, and beyond to the products of stellar collapse and explosion. This material complements and extends the material covered in some A-level courses—for example, the optional AQA astrophysics module which covers the Hertzsprung–Russell (HR) diagram (which also featured at ThaiPASS—see section 3.2) and supernova. Figure 1 shows HR diagram tracks for various masses of stars, an example of some of the material that was covered during the two ThaiPASS summer schools.

Zoom In Zoom Out Reset image size

The students were then presented with a data-mining task, in which they were asked to investigate stellar models computed by the NuGrid collaboration [3]. They were asked to find various points in the evolution of a star, and show how the abundances of various isotopes changed as the star evolved using published tools [5]. The nugridpy Python package was used by the students, in the same way, that it is used to interface with real data-mining research. This parallels the work of IRIS (Institute for Research in Schools) which promotes cutting edge research in UK schools in collaboration with UK universities [7, 8].

Whilst not directly addressed in A-level courses, the natural progression after discussing how isotopes and elements are produced in stars is to ask how those products are distributed around the galaxy, and indeed how a galaxy forms. A series of lectures was delivered in the 2018 ThaiPASS school addressing this, and the effects of supernova explosions on the formations of galaxies were discussed. Students were then trained in using OMEGA [9] a one-dimensional GCE code, stellar data library and data handling package which is used in current GCE research. Students were tasked with fitting the evolution of different elements in the Milky Way through time, by adjusting the ratio of different stars and the products of burning from these. Students enjoyed this task and showed good progress in using the code to achieve the expected evolution of a number of key elements.

Two lectures were given regarding galaxy rotation curves (orbital velocity as a function of galactic radius) and how the discrepancy between theoretical predictions and observed rotation curves led to the theory of dark matter. Though these topics are not covered in A-level syllabuses, the level of the lectures was not beyond that of an A-level student.

The task consisted of three parts, which drew on Python skills learned in previous workshops during the week. Specifically, the tasks required data handling, graph plotting and defining functions. Students were provided with rotation curve data for the galaxy M31 (Andromeda) obtained from [10] and density profiles calculated from [11]. Task one required students to plot the rotation curve using the data from [10]; task two was to plot the theoretically predicted rotation curve from only baryonic matter contributions. The two graphs from tasks one and two would not match, which indicates baryonic matter is not sufficient to explain observations. Task three was then intended for students to fix the theoretical rotation curve by adding a dark matter component.

The purpose of the task was to consolidate the programming skills learned over the week by presenting a guided enquiry-based task. The physics of the task was signposted (and the data curated to avoid complexity that would distract from the overall aims) whereas the method of programming was left to the student.

Tasks one and two, which involved loading and plotting data along with a simple calculation, were completed by all students with little to no help required. Task three proved more difficult. Most students approached the task correctly from a programming perspective; however, the increased conceptual and mathematical complexity gave rise to erroneous results.

This topic covered the ends of the lives of massive stars, their explosions as supernova and the evolution of supernova remnants (SNRs). During the lecture, supernova were discussed in general including an explanation of the different types of supernova, the physics of the supernova event and examples of observations. A large part of the lecture focused on the details of SNRs, discussing the three phases of the expansion of the shell of ejected material and showing examples of observations of SNRs at each phase. The analytic equations that describe the evolution of the first two phases (the free expansion and Sedov–Taylor phases) were introduced as a way to estimate the radius and velocity of the SNR as a function of time. Modelling supernova and their remnants was discussed, both using the analytical equations and hydrodynamical simulations. The end of the lecture included a brief explanation of how the evolution of galaxies is affected by supernova feedback.

The first task involved modelling the dynamical evolution of an SNR starting from the Sedov–Taylor phase. Using Python, the students were required to use the given equations to calculate the initial kinetic energy and radius resulting from the end of the free expansion phase. Then, using this as a starting point for the Sedov–Taylor phase use the given equations to calculate the radius and velocity of the expanding shell of gas as a function of time and plot their results. The second task used a module called yt (https://yt-project.org) to analyse data that had been produced by a hydrodynamical simulation of a supernova blast wave. Using this they could visualise different properties (e.g. density, temperature) in 'slices' through the simulation box and compare the expansion over time to that estimated from the analytical equations. As an extra task for those who had completed the first two, the students could download the plots produced from analysing the simulation data and combine them into a movie.

One of the more common introductory topics for astrophysics is the HR diagram. Colour-magnitude diagrams are directly related to these HR diagrams, which show the luminosity (or brightness) of an object as a function of its colour (a proxy for temperature). These are fundamental tools in observational astronomy and allow classification of stars by their evolutionary stages. Therefore, we provided the students with a sample of real observational data obtained from the sloan digital sky survey (SDSS) [12] SEGUE value added to catalogue, which contains a mixture of well-studied open and globular stellar clusters [13]. Students were tasked with replicating the colour-magnitude diagrams for each stellar cluster from this sample in order to illustrate the observed differences between open and globular cluster systems. Figure 2 shows an example Python plot of the M13 cluster using the SDSS data supplied to the students.

Zoom In Zoom Out Reset image size

This task calls on a wide range of data-handling skills such as accessing data in a file, manipulating that data, and finally presenting it in a legible way, all of which are key skills for any fields involving data. Students took to this task well, with all participants being able to produce their own colour magnitude plot from the data provided, though varying degrees of help were required. By comparison with any of the widely available images of M13 and its accompanying HR diagram, Python plotting tools produce publication-quality figures. Students were able to produce plots similar to that shown in figure 2 after only a few hours tuition on the first day, and were instructed to verify their work by comparing with any published HR diagrams of M13 that they could find.

Another computational topic covered at ThaiPASS was producing a Forward-Euler solver—a numerical method for solving ordinary differential equations (for a full discussion see, for example, [14, 15]). Because of its complexity, this task lends itself well to discussing the idea of building an algorithm, and the idea of planning out what needs to be coded up. Future ThaiPASS events may extend this task further to allow students the opportunity to experience algorithm development and gain proficiency with planning code methodology as it is well suited to this application.

Numerical modelling of ordinary differential equations is the backbone of many codes used in computational astrophysics and a key tool in all of the computational work undertaken in STEM subjects. In both ThaiPASS'18 and 19, students were tasked with producing their own numerical solver for integrating ordinary differential equations.

This task is particularly difficult, and was set with the expectation that students would work collaboratively. Most students succeeded, with assistance either from peers or the teaching staff, in the programming of the forward Euler solver, a particularly impressive feat considering the difficulty. The full task of modelling the orbits of an example planet was completed only by a few students; however, the difficulty of this task is more appropriate for a university level course with undergraduates at the University of Hull performing similar computational tasks in years 2 and 3. This task was largely included as a stretch and challenge goal for the most capable. Based on the experience built in ThaiPASS 2018, the 2019 ThaiPASS summer school included a brief introduction to basic calculus (as finding the area under a curve) which was not known by all students at the start of the summer school. We noted a significant increase in attainment and engagement from students with the addition of this material. Future introductory material should include either a basic description of the principals of calculus or a thorough description of the Euler method if this is to be retained. Another option to increase confidence in participating students would be to make available a model answer with solutions., which they may then compare to in order to verify their results.

Results for each of the questions are presented below:

As can be seen from figures 3 and 4, the feedback from the week long school is largely positive. Figure 3 shows student enjoyment of the various activities presented during the 2018 summer school on a scale of one to 5, with 5 being the most positive response. Overwhelmingly students enjoyed the activities. The Euler method received the worst response however the majority of students still rated this activity from average to good. The GCE lectures are clearly the most well received. The material for this session was accessible, students were engaged, and progress was identifiable in the tasks that they were set. Future ThaiPASS summer school lecture material should also be accessible, engaging and relevant, with measurable outcomes for student progression which they can appreciate at the end of the course. Further development of teaching materials is underway to facilitate this.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 4 asks students to rate their experience according to the various questions outlined. As is shown in (a) and (b), almost all students report a large increase in their technical skills in Python and the applications of computer science more generally. Almost all students also rate the enjoyment of lectures highly too, showing that the material is engaging to the target audience. The response to the increase in programming and plotting skills is more varied; however, this is likely due to the difficult nature of the material, and the response still shows that students believe their skills have improved significantly. The likelihood of students studying a STEM subject further is also encouragingly high.

For ThaiPASS'19, some changes to the questionnaire were made, and a pre- and post-school survey was undertaken with the intention of trying to better measure the impact of the school. The full questionnaire is provided in the appendix.

We see from figure 5 that students have engaged with the lecture material and found it useful and informative, feedback from NARIT and teaching staff is also very positive. Students were generally well disposed to the subject material before the school, with most giving a score of 4 or 5. The post-school surveys showed a higher proportion of 5 s, suggesting we have helped to enhance students' perceptions. Particularly gratifying is the response to the question 'Are you interested in a STEM career?': after the school, the student response was overwhelming of the highest response. While students said they enjoyed both the astronomy lectures and the coding sessions, they expressed a clear preference for the hands-on activities. This suggests it may be beneficial to either change the balance of lectures to activities, or to intersperse the two so as not to lose students' interest.

Zoom In Zoom Out Reset image size

Although English is a compulsory subject in all Thai schools, participants came to us with various levels of skill in English. Most students had little trouble comprehending written instructions in English, although oral communication was occasionally challenging. It was helpful to have Thai instructors on hand in the practical sessions to help lessen the language barrier. Perhaps the most important mitigating measure was The Starter Pack which ensured that participants were familiar with key Python and astronomical terms in English before arrival. A good number of students was able to enthusiastically communicate with the UK team and asked interesting questions (both orally and through Mentimeter an anonymous online forum).

ThaiPASS participation is based on nominations from schools by teachers. From the very beginning, we have always strongly encouraged each school to nominate at least one female student to attend ThaiPASS (unless the school is single gender). Tables 1–3 break down the participating students, teachers and instructors respectively by gender, for the 2018 and 2019 workshops.

Although ThaiPASS'19 performed worse in terms of overall F:M ratio, we were able to award 'outstanding participant' prizes to three male and three female students. In terms of speakers, we have had a slight increase over the past two years, though more work needs to be done to achieve an even balance. Feedback from female participants clearly praised the positive impact of having female instructors, who were regarded as positive role models for women in STEM. A female participant at ThaiPASS'19 expressed this as follows:

I got a chance to see my dream self. I have a dream to be a woman who graduated with a doctoral degree. After seeing Dr Claire's lectures, I have a feeling that this is possible. She looks special and even more extraordinary when she is talking about something educational; which inspired me to try harder to become like that.

Funding received for ThaiPASS'20 and '21 allows us to plan for a far bigger event. We, therefore, hope for a much larger pool of applicants from which we hope to be able to draw a better gender balance. We will explicitly ask schools to nominate female students, as we have done with the past schools. We will also try to recruit more female instructors, particularly when recruiting PhD assistants, with the aim of achieving a 50:50 balance.

Finally, because teachers were not the primary target audience for the 2018–19 workshops, we did not monitor the gender balance nor actively encouraged participation from female teachers. However, the next ThaiPASS will feature a separate 5-day Python training for teachers (around 20 expected), and we will work to improve the gender balance within this group.