Research

My research explores how observations of large-scale structure can reveal physics beyond the standard ΛCDM model. I work across cosmological theory, numerical simulations and statistical inference, developing both physical models and the methods needed to test them with data.

A central theme of my current research is whether modified gravity can provide a physical explanation for dynamical dark energy. More broadly, I investigate how new physics affects the expansion and structure of the Universe, how these effects can be simulated efficiently, and which observables are best able to distinguish them from ΛCDM.

Gravity and dynamical dark energy

The accelerated expansion of the Universe is usually attributed to a cosmological constant, but its physical origin remains unknown. Recent measurements have renewed interest in the possibility that dark energy evolves with time, raising the question of whether this behaviour could arise from a more fundamental theory of gravity.

I investigate scalar-tensor theories in which cosmic acceleration emerges from modifications to General Relativity. In particular, I developed the Asymptotic Cubic Galileon class of Horndeski models, which can produce an evolving dark-energy equation of state that crosses the phantom divide while remaining minimally coupled to matter. These models can reproduce the expansion history inferred from the cosmic microwave background, DESI baryon acoustic oscillations and supernova measurements.

Matching the expansion history is only part of the problem. Modified-gravity models also affect the growth of structure, gravitational lensing and the evolution of the metric potentials. My work therefore uses observables such as the integrated Sachs–Wolfe effect and forces around cosmic voids to identify which regions of theory space remain viable and observationally distinguishable.

I have also studied phenomenological dark-matter models that are degenerate with interactions between dark matter and dark energy. This work showed that such models can alleviate the H0 and S8 tensions and account for anomalous integrated Sachs–Wolfe signals associated with cosmic voids.

Selected work

Nonlinear structure formation beyond General Relativity

Many of the most distinctive signatures of modified gravity emerge on nonlinear scales, where analytic calculations are no longer sufficient. Simulations are therefore essential for connecting fundamental theories to the clustering, lensing and cosmic web measurements made by modern surveys.

I am a lead developer of Hi-COLA, a fast N-body simulation code for broad classes of Horndeski gravity theories. My work includes developing methods to calculate the cosmological background and scalar-field evolution, modelling modified gravitational forces and screening mechanisms, and connecting user-defined theories to nonlinear predictions for large-scale structure.

I am particularly interested in how screening depends on cosmic environment. Modified forces may behave differently in voids, walls, filaments and clusters, creating signatures that can be diluted in conventional statistics based only on scale or density.

I have also developed MIMIC for constructing constrained simulations of the local Universe beyond ΛCDM. Using this framework, I generated simulations with f(R) and nDGP gravity, allowing the effects of modified gravity to be studied while preserving the observed large-scale environment.

Selected work

  • Hi-COLA: fast N-body simulations in Horndeski gravity.
  • MIMIC: model universe constrained simulations.

The cosmic web and non-Gaussian statistics

The distribution of matter is not fully characterised by its two-point correlation function or power spectrum. Galaxies form a connected cosmic web of voids, walls, filaments and clusters whose geometry contains additional, non-Gaussian cosmological information.

I introduced the use of the minimum spanning tree as a practical statistic for cosmological point distributions and developed MiSTree, a public package for constructing and analysing minimum spanning trees. This approach characterises the lengths, shapes and connectivity of structures within galaxy catalogues without requiring galaxies to be assigned to a grid.

I have used minimum spanning tree statistics to study cosmological parameter sensitivity, massive neutrinos and observational systematics. This work showed that the geometry of the galaxy distribution can provide information complementary to conventional two-point measurements and can substantially strengthen constraints on neutrino mass.

I also develop methods for identifying and analysing the environments of the cosmic web. CACTUS provides implementations of several web-classification techniques, while my wider work examines how field-level, graph-based and geometric statistics can be made sufficiently robust for application to real survey data.

Selected work

Survey analysis and robust inference

Extracting cosmological information from survey data requires more than an informative statistic. Measurements must also account for survey geometry, observational systematics and uncertainty in estimated covariance matrices.

Within Euclid, I contribute to the weak-lensing angular power-spectrum analysis, including the development and validation of methods for estimating two-point statistics and their covariance directly from survey data. My work combines debiased jackknife estimation with shrinkage techniques to produce covariance matrices that are more accurate and stable when only a limited number of simulations or data subdivisions are available.

Within DESI, I work on large-scale structure analyses and alternative clustering methods. I lead the minimum spanning tree analysis within the Alternative Clustering Methods Key Project.

Within Rubin LSST DESC, I co-lead the Beyond-wCDM Models Topical Team. Across these collaborations, my aim is to connect theoretical models and simulation predictions to statistically robust measurements from Stage-IV surveys.

Selected work