Congratulations to Edvig Selimaj, who today successfully defended his MSc thesis, by the title of “Non-parametric and symbolic reconstruction of the late-time expansion history” (with the opponent being Prof. Albino Perego), receiving the final grade of 110/110! In his thesis which I supervised together with Miguel Sabogal, Edvig reconstructed the late-time expansion history using Gaussian Processes applied to a range of late-time observations, and then applied symbolic regression to these reconstructions to derive simple, interpretable expressions for the expansion rate and dark energy properties, finding very interesting results, including a physically motivated parametrization which performs substantially better than CPL in fitting DESI data.
Nonparametric spatial curvature from DESI DR2
Happy to see my last preprint of the year (since I’m on paternity leave) out on arXiv: with Hannah Turner (kudos for doing all the heavy lifting), Suhail Dhawan, Alice Townsend, and Peter Massey, we inferred spatial curvature nonparametrically, and with minimal assumptions, using DESI DR2 BAO data in combination with other late-time expansion rate probes. This is essentially an update of my 2021 MNRAS Letters paper with Suhail and Justin Alsing, applied for the first time to Stage IV cosmological data. We find up to a factor of 3 improvement compared to the earlier application, driven by both new BAO and SNeIa data. Most importantly, we provide support for the hints of a spatially open Universe found in other works, although our results do not rely on any specific model. You can read our results in the preprint we just posted on arXiv: 2609.22470.
