For me, this paper marks the first fruit of an idea that was planted in early 2018. Applying Bayesian inference to radio pulsar data.
I'd recently moved to Monash University to take up a position as an assistance lecturer with Paul Lasky. The group was young, exciting, and had a plan: to write a user-friendly Bayesian inference package, Bilby, for the analysis of gravitational waves. During the initial planning stage, where most of the ideas floated around on whiteboards, we realised the idea had legs beyond gravitational waves. If we could write the software in a sufficiently general way, we could apply the cutting edge methods to lots of problems!
Around the same time, Jim Palfreyman and his collaborators published their ground-breaker article "Alteration of the magnetosphere of the Vela pulsar during a glitch" which detailed the data and new insights they had made from the first pulse-to-pulse data from a radio pulsar glitch. Usually, radio astronomers observe a pulsar intermittently, say once a week or so, and only see the effect of the glitch, Jim had been continuously monitoring the Vela pulsar and managed to catch it in the act.
Vanessa Graber and I both did our doctoral studies at the University of Southampton and often joked about writing a paper together, but so far our paths had diverged. She was working at McGill in Montreal and becoming more entrenched as as expert in theoretical modelling of the neutron star interior while I was predominately working in the data analysis of gravitational waves.
After Jim's paper was published, I got in contact with Vanessa to get her thoughts on it as I knew she would have a lot of insight. She told me that she had been thinking a lot about the glitch as a test for a model she had been developing. Typically, neutron star glitches are modelled as the interactions of two part of the star, the crust (which we observe from the electromagnetic emission) and a superfluid core. Her modelling suggested something more complicated, there might instead be two parts to the core giving a total of three components interacting. If true, this predicted an overshoot in the rotation frequency of the star. Along with her collaborators, Vanessa had already written an article "Glitch Rises as a Test for Rapid Superfluid Coupling in Neutron Stars", which detailed the model and had a preliminary look at the Vela data. We finished our conversation by asking, could we use Bilby to perform a direct model comparison directly testing Vanessa's model?
Paul's office is right down the corridor from mine. I recall that it took all of a minute for Paul to get the idea and understand the significance. If this worked, it would be a direct way to prove the interior of the neutron star with data. Paul emailed Jim to find out if he would be interested in collaborating. Within minutes we had an affirmative email and the four of us where assembled.
The idea itself was straight forward: use Bilby to analyse Jim's data and understand what was happening to the rotation of the star during the glitch, testing Vanessa's model. However, we were doing something no one had done before, direct inference based on the data. This introduced a lot of unexpected problems in handling the data which plagued us early on. Fortunately, Jim knows his stuff and was able to help fix the bugs.
The analysis itself was plain sailing. Helped, in late 2018 by a visit to Melbourne from Vanessa supported by the OzGrav international visitors program. We wrapped up in early 2019 and started discussing the results with colleagues whilst preparing our publication. This was extremely exciting. As soon as we explained the idea and the results, peoples eye lit up. They knew the importance.
Working with a team of international experts is always thrilling. That this project yielding insight into neutron stars is directly attributable to the quality and diversity of the researchers. I'd like to take this opportunity to thank them for there hard work. I'd also like to thank the excellent Carl Knox from OzGrav for creating the stunning images of neutron stars (see the post artwork and the picture below).
This analysis was the first of its kind, but we don't expect it to be the last. There are other glitches from the Vela pulsar, albeit not pulse-to-pulse, which we want to look at and with Tushar Nagar, a Monash student, we are applying model-agnostic inference methods which might help further untangle the story. We hope this work helps drive forward our understanding of neutron stars, one the Universe's great enigmas.