As we get ready for the start of a new academic year, we would like to take a moment to look back at the fantastic work done by our 8 exceptional summer interns! Learn more about the research projects and experiences of Mitchell Barrett, Audrey Bourdon, Margaret Bruna, Maigan Devries, William Frost, Numa Karolinski, Kim Morel and Jasmine Parsons in the testimonials below.
Moreover, René Doyon and Étienne Artigau collaborated with two Université de Montréal professors, Julie Hlavacek-Larrondo and Laurence Perreault-Levasseur, to supervise two additional interns. These students, Louis-Simon Guité and Guillaume Payeur, are the first (but not the last, we hope) interns to work on interdisciplinary projects with iREx members.
Trottier intern from the University of Toronto, working with Björn Benneke at the Université de Montréal
Intern from McGill University, working with Nicolas Cowan at McGill University
The topic of my internship was direct imaging, an exoplanet detection technique. This technique will work with visible light for future telescopes. From each image, we can only retrieve the projected separation from the planet to its star, as well as the flux coming from the reflected light on the planet. My goal was to develop an analytic method with Bayes theorem in order to obtain a probability for the Keplerian parameters, which describe the orbit of a planet.
What is interesting, and surprising, is to see how it is possible to obtain, with the help of our prior knowledge on the Keplerian parameters, a probability density function for those parameters based only on a single direct image. No numerical program can converge after just one image, but this analytic method can give us some information about the orbit with such little data. It really is impressive.
My most important result is the probability density function I was able to derive for the general case of eccentric orbits. It still needs polishing but getting an answer to this complicated analytic problem is encouraging.
I learned about Bayesian statistics working on my project, but also a lot about exoplanets. The online meetings and talks were important in obtaining this new knowledge; I was able to learn about so much more than what my internship was about.
The problem I was working on had already been solved for circular orbits, for which the eccentricity of the planet is zero, and I was trying to derive the more general case of eccentric orbits. I had to be very careful with all the steps of my derivations, and I had to start from the beginning multiple times, after I’d found a mistake in my work. It was hard sometimes to not get discouraged.
I really liked that this internship showed students a glimpse of what exoplanet research can be. As interns, we got to attend different talks and presentations given by other astrophysicists, as well as interact with graduate students and supervisors. Feeling treated as a researcher more than an intern, I really appreciated the welcoming feeling.
Intern from McGill University, working with Nicolas Cowan at McGill University
This summer I continued my Honours Research thesis with Prof. Nicolas Cowan. The project I worked on studied the retrieval of Keplerian orbital parameters via direct imaging with visible light. We are trying to determine if information from visible light reflected by the planet helps us to retrieve the orbit of an exoplanet more efficiently and accurately, with as few images as possible!
This project is geared towards exoplanets which will be studied when telescopes will have the precision to resolve dim planets compared to their super bright stars. This will probably be in about 10 years. For me, what I find to be interesting about this project is how forward thinking it is. I think it’s neat to be able to help further our knowledge on data which has not yet been collected.
It is a bit too early for me to be able to write about conclusive results. Many aspects of my project are close to determining information on whether light information helps us retrieve exoplanet orbits, and under what circumstances. I’m excited for the next couple of weeks, because I’ll have concrete results from my work!
I definitely learned a lot about the way in which telescopes like HabEx and LUVOIR are going to work. My project is aimed towards retrieving exoplanets that those telescopes will detect. I also became more comfortable and knowledgeable when it comes to coding in Python. I’ve spent a lot of time adding to the code I started during my Honours Thesis, and it’s helped me understand how to problem solve and write more efficiently.
I think it’s safe to say everyone’s summer did not go quite as planned this year. A major hurdle for me was adapting to working from home. I find it quite difficult to be productive when I haven’t commuted to a new environment. Also, navigating Zoom meetings was quite the bumpy learning experience!
The best part about my summer internship was learning about the community of researchers around me! I have more time to get to know the people in my research group, and I’ve been able to meet all the other summer interns at iREx. This summer especially, I’ve been appreciative of the scientists I’ve met and learning about their research (and life) experiences and perspectives. It was cool to hear the amazing work my peers have been up to over the summer!
Trottier intern from University of Northern British Columbia, working with Jason Rowe at Bishop’s University
The topic of my internship was to create a new hybrid MCMC method using three pre-existing ones. The methods used were Metropolis-Hastings with Gibbs sampler, Differential Evolution MCMC, and affine invariance with ensemble sampler. MCMC methods can be used to obtain estimates for model parameters. With this, the hope is to have a method with faster convergence and the ability to handle correlated parameters, which will be helpful when working with exoplanet systems with hundreds of parameters.
I was interested in how the MCMC chains and convergence rates changed with the addition of each method. Each method led to chains with better mixing and convergence was achieved much earlier on.
We tested our method with a Gaussian function with correlated parameters and found that it led to the expected results with much faster convergence than the three methods on their own. This let us know that our method works as expected and can be used on exoplanet systems in the future.
This was my first time learning about exoplanets, so I learned a lot about exoplanet detection methods and exoplanet terminology. I also learned how to code in python, and how to implement multi-processing pools. As well, I learned about MCMC, some of the MCMC methods, and how to implement them.
MCMC was a new concept to me so the biggest challenge was familiarizing myself with the different methods and trying to make sure I had a solid understanding of each. As well, the code took quite a while to run so there were times when I would spend a day running code only to realize I had made a mistake and it didn’t work.
I really enjoyed the opportunity to learn more about exoplanets and MCMC through working on my project and attending two conferences this summer. As well, I enjoyed working with Dr. Rowe and his team of students and getting to know more about other projects that people are working on.
Trottier intern from McGill University, working with René Doyon and Loïc Albert at the Université de Montréal
My internship was spent working on algorithms related to the NIRISS-SOSS extraction pipeline. In short, NIRISS is a near-infrared imaging instrument aboard the James Webb Space Telescope. It has a Single Object Slitless Spectroscopy mode (SOSS) used to obtain spectra from transiting exoplanets. The goal of the internship was to implement a spectral extraction technique which accounts for contamination from unwanted sources in the NIRISS field of view. It is referred to as a “differential spectral extraction”. Therefore, during the internship I played with simulating data and extracted results from those simulations using the differential technique and others.
What I found most interesting was being involved in the development process for something applied to the JWST! Being able to observe the team and contribute to their work was the highlight of the internship for me. Additionally, the SOSS extraction technique I was tasked to implement could be of good use for future JWST users. Being able to work on something that could have a direct impact on data obtained from a JWST instrument was quite cool.
Most of what was discovered was quite niche, but the general takeaway was that NIRISS-SOSS differential extraction works! Applied correctly, it can account for contamination from unwanted sources that appear on the NIRISS detector. Since the likelihood of this background contamination appearing is much higher than with usual spectrographic instruments, having a way to deal with it is very important.
Plenty! I learned about the whole transit spectroscopy extraction process and how to implement it in code. I learned about what has to be considered when simulating scientific data and about the many ways a noisy signal can be processed to obtain better precision. I also learned about how the NIRISS-SOSS team operates and about the complexity of such projects.
My biggest challenge was having to write increasingly complex code in order to make my spectral extraction processes as correct as possible. Whether it was adding new functionalities or improving an existing one, I often found myself tweaking the same code functions as I learned more about the correct methods to use.
I really enjoyed interacting with the NIRISS-SOSS team and the people at iREx in general. I especially liked the conversations I had with my supervisors Loïc Albert and René Doyon.
Université de Montréal intern, working with Julie Hlavacek-Larrondo in collaboration with René Doyon and Étienne Artigau at the Université de Montréal
My project was interested in getting the first high contrast images of exoplanets orbiting powerful high-mass X-ray binaries using data from the Keck Observatory. My work was focused on one system, namely RX J1744.7-2713.
The most interesting thing about my internship was the interdisciplinary aspect of it. Indeed, I was fortunate enough to work on both exoplanets and X-ray binaries, two relatively distinct areas of research. However, my internship brought these fields of study together, which allowed us to make a very interesting discovery.
We observed several candidate sub-stellar companions orbiting at various distances from RX J1744.7-2713. One companion orbits exceedingly close compared to the other candidates, at around 300 AU from the central system. This is very important because it was the first time such companions were directly detected around high-mass X-ray binaries.
First, I had to learn the theoretical foundations regarding high-mass X-ray binaries and direct imaging techniques. Secondly, I had to improve my programming skills in Python in order to correctly reduce the raw data coming from the Keck Observatory and reveal the exoplanets in the final high contrast image.
In the middle of the summer, my project changed quite a lot. Originally, I was supposed to work on a disk discovered around RX J1744.7-2713 back in 2017. However, more precise observations made this summer suggested that this disk did not exist and that it was in fact a candidate sub-stellar companion. Therefore, I had to adapt quickly in order to produce the high contrast images before the end of the summer.
The thing I enjoyed the most during my internship was the making of an image where we could see the exoplanets. The fact that I started with raw data where no exoplanets were visible to end up with an image with multiple detections was very rewarding and exciting.
Intern from McGill University, working with Nicolas Cowan at McGill University
Trottier intern from the Université de Montréal, working with David Lafrenière at the Université de Montréal
The main goal of my project was to find a new cleaning method that would be able to eliminate systematic noise from SPiRou’s high dispersion spectroscopy data while minimizing the loss of planetary signal from the exoplanet we observe. The method currently used works well, but there is a relatively important loss of planetary signal on the corrected spectra, so this is the element we want to improve
One of the most difficult methods to try out was based on advanced mathematical concepts. I had to do some research to better understand how some already implemented Python algorithms work and then try to imitate them by myself. The result was fairly conclusive, but the method still needs to be refined.
After having tried four cleaning methods in total, two of them stood out more. When we apply them on simulated data similar to SPiRou’s, these methods are able to retrieve the right parameters that were used to build the initial spectra sequence with a low error margin. Therefore, they could soon be tested on actual data and might potentially be great solutions to our issue.
I developed my programming skills with Python a lot. I also learned many things about the data analysis process when we want to identify the components of some exoplanet’s atmosphere. I also realized that the collected data from a ground-based telescope can reveal a lot about an object only once an arduous cleaning process has been done to subtract noise.
Considering it was my first experience in astrophysics, the learning process of basic concepts took a lot of time. Also, it could often happen that I faced some issues while programming with Python because I did not have so much experience in programming either. It could also be difficult sometimes to ask questions to my colleagues since we were working remotely because of the pandemic, so Google became my best friend!
I learned a lot this summer, both about exoplanets observation concepts and programming. I got the chance to gain experience in the field I am passionate about and I got to know many researchers who all work on amazing projects. This internship also allowed me to realise I was on the right direction and that I really hope to become an astrophysics researcher in the future.
Trottier intern from McGill University, working with René Doyon and Étienne Artigau at the Université de Montréal
My internship focused on exploring a new method to improve the accuracy of stellar radial velocity measurements. Essentially, when deriving a radial velocity measurement from a star’s spectrum, researchers traditionally make use of only negative features (i.e. the spectrum’s troughs). My summer was spent investigating whether the positive features (i.e. peaks) in the star’s spectrum might also be interesting to look at, and whether taking these features into account would improve the accuracy of the radial velocity measurement.
All the data I worked with this summer was from an instrument called SPIRou, on the Canada-France-Hawaii telescope. It was amazing working with data from observations of an actual star in the real world. This was my first experience with research, so I was used to working on problems in a classroom setting, problems that tend to be completely disconnected from actual data and research. A few times over the summer, I would suddenly be struck that the files and data I was dealing with could be traced back to a red dwarf six lightyears away. Pretty mind-blowing!
We discovered that including these aforementioned positive features does indeed improve the accuracy of radial velocity measurements. There is also evidence that these positive features, which are traditionally ignored, hold a wealth of information concerning the activity of a star. There definitely seems to be a lot of potential for even more research moving forward.
Put simply, I’ve learnt more programming skills in the last four months than in an entire year of taking classes. When learning how to code in class at university, every question has a single correct answer, which I’ve learnt is not reflective of the research experience. In reality, there are usually multiple ways to approach a problem, and the decision of which path to take is rarely straightforward.
The circumstances surrounding the internship were often more challenging than the work itself. It was difficult staying motivated, as everything was remote. I also contracted COVID-19 in the beginning of my internship, and had to take time off to recover. In spite of all this, I’m amazed at how much I’ve learnt this summer and what an eye-opening experience it was; I’m also grateful to the coordinators of iREx and to my supervisors for making it all possible!
To be honest, I was so relieved that the internship went forward remotely instead of being cancelled, since it gave me something to focus on and allowed me to keep moving forward and learning. The frequent meetings I had with my supervisors Étienne Artigau and Prof. René Doyon kept me on a structured, productive schedule. As classes are starting now and self-discipline is more important than ever, this means I feel well-prepared for the semester.
Université de Montréal intern, working with Laurence Perreault-Levasseur, in collaboration with René Doyon and Étienne Artigau at the Université de Montréal
My internship consisted of developing a new technique based on machine learning for reducing the quantity of noise present in data collected by the infrared spectrographs NIRPS and SPIRou. The objective was to improve on the conventional procedure, which does not involve machine learning. As such, I had to design a machine learning algorithm from scratch.
As this was the first time I got involved in research, I found it really interesting to discover how it feels to work on a research project. It is unlike anything I’ve done before, but I love it. I also found it interesting to collaborate on a project with successful researchers, because it meant I could learn from their expertise.
We’ve been able to show that machine learning really has its place in this field. Restricting ourselves to simulated data, we have succeeded in creating a machine learning algorithm that improves significantly on the conventional method used. The next step is to show that it can work well on real data too.
By the start of the summer, I had never done any machine learning in my life, and as such the most significant thing I’ve learned over the summer is to create machine learning algorithms. It was a very enjoyable journey. I’ve also had the chance to learn about some of the inner workings of NIRPS and SPIRou, which is wonderful as well.
There are times where I desperately wanted my machine learning algorithm to achieve some level of performance, but I couldn’t make it happen. Sometimes I’d run out of ideas for solutions to try, and in these cases it was difficult to stay positive. Eventually, though, I made a big breakthrough and things went much better there onwards.
I loved having a direct impact on science! It has been my dream for some time to participate in a research project, and it felt great to get to that point. I also loved being part of our small research group! I bonded fairly well with the people in it, so it was a real pleasure to interact with them. It felt particularly rewarding when I had some encouraging results to share with them because I could tell that they were excited about it too.