We simulate the impact of binary interactions on the timing and spatial distribution of core-collapse supernovae and consider how these change may affect galaxy evolution.

This paper is my first use of cogsworth for a full-scale project and I'm very excited to share it with you! I started this project while working as a pre-doctoral fellow at the CCA and learnt a lot during the project from my wonderful co-authors - particularly those working on galaxy evolution.

The goal of this paper was to assess just how reasonable it is for subgrid models of supernova feedback to assume that all stars undergo single star evolution in hydrodynamical galaxy simulations. We find that the timing and spatial distribution of supernovae is significantly altered by binary interactions, and that these changes exist even for extreme variations in binary physics, initial conditions and galaxy settings.

Given the robustness of our findings, we created an analytic model for supernova feedback that accounts for metallicity-dependent, binary-driven supernovae. This model reproduces our simulations closely and can be used directly in future galaxy simulations. The code to use this model is available in the GitHub repo that I link up above.

Overall, I'm hopeful that galaxy simulators will use our results to account for binaries in their feedback models and then we can find out how this changes the evolution of galaxies. I think this might be particularly interesting in high-redshift, low-metallicity simulations where galaxies are smaller and the spatial displacement of supernovae is increased.

Read the paper Explore interactive plots!

We compare the ejection velocities of secondary stars after their companion disrupts the binary after core collapse in three different population synthesis codes.

This paper started with finding a bug 🐛

When using COSMIC to evolve a population of binaries, I found a peculiar system where its pre-supernova orbital velocity was 12km/s, but somehow the companion was being ejected at more than 60km/s. This was clearly unphysical and, after a long and arduous investigation (with the help of Katie Breivik), we eventually realised that the original paper by Kiel & Hurley 2009 had a mistake, such that all companion ejection velocities are wrong. Motivated by this, we decided to investigate the ejection velocities of stars in the binary supernova scenario across three different population synthesis codes: COMPAS, COSMIC and binary_c.

Surprisingly, we found that all three codes produced erroneous ejection velocities for disrupted binaries due to a variety of subtle bugs. Gladly, after fixing these bugs, we found that the ejection velocities of the secondary stars relative to their pre-supernova orbital velocities were very similar across all three codes. This nicely cross-validates three independent algorithms.

My personal takeaway from this quick project is that comparison projections of this nature, and open science in general, is essential to catching these subtle bugs and improving our models!

Read the paper

We present a new Python package that combines population synthesis and galactic dynamics simulations to allow users to make predictions for the positions and kinematics of stars and compact objects.

I'm really excited to share cogsworth with the community! My co-authors and I first talked about making a tool like this back when a microlensed black hole was discovered (Sahu+2022, Lam+2022) and we wondered if we could make predictions for the rates of these objects and use them to constrain binary evolution. Soon after Andrews & Kalogera 2022 did exactly that in an excellent paper, so we wondered if we could instead cover an even larger scope and investigate a "Galaxy of Binaries".

Fast forward a year or so and now we have cogsworth, a code that can make predictions for not just microlensed black holes, but also X-ray binaries, runaway stars, $r$-process enrichment sites, supernovae, LISA gravitational-wave sources and much more!

With cogsworth you can define custom star formation histories, vary binary physics, use different galactic potentials, start a simulation from a post-processed hydrodynamical zoom-in and convert all of your results to observables. In the release paper we explain its full capabilities in detail and also demonstrate its usage in some example use cases. I hope cogsworth can be useful for you too!

Read the paper Explore cogsworth

A new website for streamlining and standardizing software citations in your papers

Floor Broekgaarden and I have designed The Software Citation Station to help people accurately and easily cite the software used in their papers. We chatting over coffee at the Flatiron Institute about how there's no central location for finding software citations and it can be confusing to work out how to properly cite everything that you use. This motivated us to create the Software Citation Station over the next few weeks as a resource for the whole community.

The Software Citation Station allows users to simply (1) select the software they've used in their work (2) choose the versions of each package that they used (3) copy a dynamically generated LaTeX acknowledgement and associated BibTeX. There's also functionality to search and filter the package list, as well as a form for adding new software to the page. I hope you find it useful!

Visit the site!

We demonstrate how a previous mass transfer episode can impact the asteroseismic signals of an SPB star and discuss the implications for inferring stellar properties as well as the potential for constrain mass transfer

Last summer (2023) I took a deep dive into asteroseismology at the Kavli Summer Program in Munich and got the chance to work on this project with some awesome people! We used MESA to a simulate an accretor that experiences mass transfer in a binary system, and then applied GYRE to calculate its pulsation periods and compare it to a single star.

We found that the asteroseismic signals of a star that's undergone accretion are significantly different from an equivalent single star and that these differences remain throughout the star's main sequence.

We show that, for a star that's undergone accretion, using asteroseismology to infer its age and mass based on its period spacing pattern can be very inaccurate when assuming single star evolution. We also discuss the potential for using the period spacing pattern to constrain aspects of binary mass transfer, in particular rejuvenation. For more details, read the paper!

Explore interactive plots!

We make predictions for the impact that the Rubin Observatory Legacy Survey of Space and Time (LSST) will have on the Near Earth Object (NEO) follow-up system and present strategies for mitigating this impact.

The sheer volume of objects that will be detected by LSST presents a problem for the NEO follow-up system, since so many potential NEOs will be reported each night that efficient follow-up will be very difficult. Even more than that, the submissions will be heavily polluted by main belt asteroids masquerading as NEOs.

We quantify this problem and examine exactly how large of an impact LSST will have. We created a new algorithm for predicting which LSST observations will likely not require follow-up and show that the application of this algorithm can mitigate the impact significantly.

LEGWORK is a python package that does the LEGWORK for you by evolving binaries, calculating gravitational-wave strains, computing signal-to-noise ratios for binary systems potentially observable with LISA and visualising the results

My co-authors (Katie Breivik and Selma de Mink) and I have officially released LEGWORK and you can now read the paper on the ArXiv! This package was dreamt up almost entirely independently by both me and Katie, and when we realised that we had the same idea we decided to work together to create this package!

Our motivation behind LEGWORK was to create a package that was completely open-source, since this would (1) allow seasoned experts to collaborate and suggest new features or improvements and (2) make it easier for newcomers to quickly understand how to make their own predictions. Another motivation was that, by introducing this package to the community, we could create a collaborative code to avoid individual mistakes. We found that many papers have slightly different pre-factors in their equations, which could lead to very different predictions, so we aim for LEGWORK to provide a central tool and reference for the community to use in order to avoid this confusion.

The installation of the package should be very simple. Just run

pip install legwork

and you're all set! If you have any installation issues you can check out the full instructions here.

LEGWORK can be used to evolve the orbits of binaries, calculating their gravitational-wave strains, compute their signal-to-noise ratios and visualise the results! I encourage you to check out all of the tutorials, demos and documentation here. Below I've included some plots that demonstrate the sorts of things you could make with LEGWORK. The plot on the left illustrates how eccentricity can affect the detectability of a binary - some eccentricity can be good for increasing your SNR, but too much can move the GW emission into an area of the sensitivity curve to which LISA is less sensitive! The plot on the right shows the horizon distance for circular binaries over different frequencies and eccentricities. For more information about these plots I encourage you to read the "Use Cases" section of the release paper, in which we discuss them in detail!

We examine the population of LISA detectable BHBHs, BHNSs and NSNSs in the Milky Way for 20 different binary physics variations, using a new model for the Milky Way and large grid of population synthesis simulations

What happens when a grid of over 2 billion massive binaries, a new empirically-informed model for the Milky Way, a Python package for stellar-origin LISA sources and a group of astrophysicists walk into a bar? This paper!

Long long ago this paper started off as my senior thesis for college, supervised by Floor Broekgaarden and Selma de Mink. At the time I was quite pleased with it...but if you want to amuse yourself, go see how simple the thesis was by comparison (though I will say that I managed to accomplish all of the "future work" section, so there is that!). Then, just in the final days of college, the world erupted into chaos with a pandemic and so, given my previous experience with online classes, I decided to take a gap year before starting my PhD. Instead, I spent the last year working on this (and LEGWORK) and I'm so very excited to finally share it with everyone!

In this project we investigated the population of LISA detectable binary black holes (BHBHs), black hole neutron stars (BHNSs) and binary neutron stars (NSNSs) that reside in the Milky Way. Although similar studies have been done in the past, we made several improvements. We use an improved model for the Milky Way that accounts for the chemical enrichment history of the galaxy and the size of our simulations is far larger than those used by others, which reduces our sampling uncertainties. We also approached the problem with stellar evolution in mind and thus examined our results for 20 different model variations in order to see how things changed when we alter the underlying binary physics assumptions.

Read my paper summary post here!

We built a framework for creating hybrid catalogues of solar system objects that combine real objects with MPCORB with simulated catalogues to allow for LSST predictions that account for prior observations

Current predictions for LSST solar system observations are often done using synthetic solar system catalogues. These catalogues contain every expected solar system object, not accounting for the fact that we have already found many of the larger or brighter ones. I think the implications of this can be best expressed by Gru:

Predicting the re-discovery of Ceres, or any of the known solar system bodies for that matter, would of course be unfortunate. Therefore we've built a framework that can create a hybrid catalogue that combines the latest data from the Minor Planet Center with the Simulated Solar System Catalogue (S3M)

You can read more about it in the summary post (linked below) or the Appendix of Wagg+2024

Read the summary post here!

COMPAS combines tools for statistical analysis and model selection with rapid population synthesis, allowing inferences to be made about the details of stellar and binary evolution with ease

COMPAS is an open-source, rapid binary population synthesis code that I help to develop. I was introduced to the COMPAS team by Selma de Mink whilst working on my Galactic DCO paper and took a deep-dive into the codebase in order to help improve COMPAS' handling of low-mass stars with Selma. I've also taken a detailed look at COMPAS' wind mass loss prescriptions and helped to clean them up as well as change the way COMPAS handles luminous blue variable wind mass loss by default.

We recently released a comprehensive methods paper that details how the code works and the science behind it. I really enjoyed helping to write this paper as it gave me a great opportunity to get a clear understanding of all of the underlying physics that goes into the code!

I focussed on the single star evolution and, in particular, made Figures 5-8 for the paper. For example, below I include Figure 5 from the paper, which shows a Hertzsprung-Russell diagram at two different metallicities, which I produced used COMPAS. You can see the approximate solutions that COMPAS uses in the Hertzsprung Gap for example (where the lines are completely straight), yet the general shape of the evolution matches what we expect from stellar evolution. In case you're interested in checking them out, the other figures show: the maximum radial extension of each stellar type, the main sequence lifetime for different metallicities and the initial-core-remnant mass relations for different metallicities and prescriptions.

If you're interested in learning more about how COMPAS works then I recommend that you read the methods paper, it is very detailed and has a convenient table of contents for you to hop around. And if you're less interested in how the code was designed, but instead want to know how to use it then check out the documentation!