Philip Clarke

About

I'm currently finishing a PhD in computational cosmology, at the University of Cambridge. I enjoy physics, coding and communication, sometimes all at the same time. Here is part of my CV (full CV upon request).

Writing

If you're curious about what our universe has been up to for the past 14 billion years, check out this timeline of the universe. For what it's up to today, take a look at how fast is the universe expanding?

Here's some more technical stuff. My master's essay from 2017 on two-field inflation (how can we probe the earliest moments of the universe?), my undergraduate project report from 2016 on the conformal bootstrap (what can symmetries tell us about physics?), and a poster and slides for the same project.

During my PhD I developed methods to connect a particular early-universe signature to today's sky, making its precise calculation feasible for the first time. These methods are described in my paper Probing Inflation with Precision Bispectra (published in the Journal of Cosmology and Astroparticle Physics) and have been implemented in the Primodal code, soon to be released.

Some Primodal helper functions can be found here. There is also an example notebook here showing some of the functions in use. Primodal itself is here.

Code

I've been working on some small projects recently to sharpen my knowledge of common data science tools. First, I asked the question which days have fewer births? Pandas was great for making the many plots that helped find the answers. It was pretty interesting! Another write-up is a little exploration of PCA and clustering. I wanted to practise using Sklearn, especially writing things in terms of pipelines, and to gain some intuition for how clustering and PCA interact. Finally, here's my write-up of one of the classic machine learning examples, digit recognition using the MNIST dataset. I compared a few different models, and visualised how a couple of different methods of dimensional reduction compress the pixel images.

Here's some fun code for demonstrating gravitational lensing. It warps your webcam feed in the same way gravity warps passing light. If you'd like to use this code for a physics demo, let me know if you have any trouble getting it working! For something more technical, here is some code for generating weights for numerical integration. In my PhD I needed to integrate a lot of functions over the same sample points extremely efficiently - this is the code that let me do that.