Thesis & internship opportunities

Looking for a thesis or internship in cosmology?

Here are some projects currently open in our group — ranging from analytical theory to numerical simulations and observational data analysis. If any of these spark your interest, or if you have your own ideas that connect to our research themes, get in touch.

Beyond ΛCDM · Perturbation theory Defended

Hubble tension & evolving dark matter

The universe's expansion rate is one of the most precisely measured quantities in cosmology — and yet two independent ways of measuring it disagree at a level too large to ignore. Something is missing from our model.

Can a dark matter component with a slowly evolving equation of state resolve the H₀ tension? This master's thesis derived the perturbation equations for such a model and studied how small-scale density fluctuations grow differently compared to standard ΛCDM.

Methods: perturbation theory · ODE solving (Python/Mathematica) · cosmological background evolution
Preliminary steps
  1. Get familiar with background cosmology: Friedmann equation, continuity equation, Hubble law. Recommended: Dodelson & Schmidt "Modern Cosmology"; Peebles "Large Scale Structure" §9–13.
  2. Derive the relations connecting d/da, d/dt, and d/dz — these are the main independent variables throughout.
  3. Express energy contributions as fractional densities Ωᵢ(a). Find matter–radiation equality and dark-energy–matter equality epochs.
  4. Plot ρᵢ(a) and Ωᵢ(a) for radiation, matter, and Λ. Enforce flatness. Add the last-scattering epoch.
  5. Re-derive the standard growth equation (e.g. eq. 9.76 of Kolb & Turner). Recover known solutions for EdS, radiation-dominated, and mixed universes.
  6. Solve the perturbation equations numerically for the evolving dark matter equation of state from Naidoo et al. (2022). How many coupled equations? What value for c²ₛ?
Preliminary reading
  • Dodelson & Schmidt, "Modern Cosmology" (2nd ed.)
  • Amendola & Tsujikawa, "Dark Energy" §2, §6
  • Kolb & Turner, "The Early Universe" §9
  • Naidoo, Jaber et al. (2022), arXiv:2209.08102
  • Verde, Treu & Riess (2019), Nature Astronomy 3, 891 — H₀ tension review
Expected outcomes

Derivation and numerical solution of the perturbation equations for evolving dark matter; comparison with the ΛCDM growth function; assessment of whether the model's perturbation behaviour is viable.

Local universe · Dark energy Defended

Dark energy & the Local Group

Our nearest galactic neighbour, Andromeda, is falling toward us. From the details of that fall — how fast, from how far, over how long — we can weigh the Local Group. But does a cosmological constant change the answer?

This bachelor's thesis solved the Timing Argument equation — the two-body problem in an expanding universe — with and without a cosmological constant Λ, assessing whether dark energy alters our inference of the Local Group mass. This project gave rise to ongoing research.

Methods: classical mechanics · numerical ODE solving · cosmological dynamics
Preliminary steps
  1. Read Peebles "Large Scale Structure" §1–5 and §19 (Spherical Collapse). Understand comoving vs. physical coordinates.
  2. Review the Timing Argument literature: Kahn & Woltjer (1959); Kroeker & Carlberg (1991); Li & White (2008); van der Marel et al. (2012).
  3. Understand the role of Λ in driving background expansion: Amendola & Tsujikawa §6.1.
  4. Follow Partridge et al. (2013) eqs. 2–5 to solve the two-body equation for Λ = 0. Compare with their Figure 1.
  5. Understand the units of the Λ/3 · r term. What is the fractional density associated with a cosmological constant?
  6. Solve the full equation (Λ ≠ 0) numerically for a range of input parameters. Can you find a parametric form?
  7. Assess whether adding Λ changes the inferred Local Group mass, and by how much.
Preliminary reading
  • Peebles, "The Large Scale Structure of the Universe" §1–5, §19
  • Partridge, Lahav & Hoffman (2013), MNRAS 436, L45
  • Kahn & Woltjer (1959), ApJ 130, 705
  • van der Marel et al. (2012), ApJ 753, 8
  • Amendola & Tsujikawa, "Dark Energy" §6.1
Expected outcomes

Numerical solution of the Timing Argument with and without Λ; quantification of the effect of dark energy on the inferred Local Group mass.

Cosmic web · Galaxy formation In progress

Cosmic web & galaxy merger multiplicity

The majority of galaxy mergers are binary — two galaxies colliding. But a non-negligible fraction involve multiple galaxies merging at once. How does that fraction depend on where you live in the cosmic web?

Using IllustrisTNG and the public pipeline by Mack & Genel (2025), this master's thesis quantifies how the large-scale cosmic web environment — voids, walls, filaments, nodes — modulates the frequency and nature of multiple-galaxy mergers.

Methods: IllustrisTNG · merger trees · cosmic web segmentation · HPC (PLGrid)
Peculiar velocities · Machine learning Open

Pairwise velocities & the local universe

Cosmology describes the universe as a whole — but we observe it from one particular corner of it. Our local environment is not average, and if we ignore that, our measurements of how fast structure grows may be quietly biased.

The mean pairwise velocity v₁₂ has already been measured from the Quijote simulations and used to constrain fσ₈ via a full likelihood with Cobaya. The same statistic has been measured from constrained simulations of our local universe. The project asks: does our local v₁₂ profile produce a biased fσ₈?

Methods: N-body simulations · Gaussian process emulator · Cobaya · statistical inference
Local universe · Dynamics Open

A mass profile of the Local Group

The Timing Argument is a powerful but blunt tool: it gives you one number — the total mass of the Local Group. What if we could turn it into a scalpel?

By applying the Timing Argument to the full population of Milky Way satellite galaxies at different radii, this project constructs a mass profile of the Local Group — enclosed mass as a function of distance from the centre — from kinematics alone.

Methods: satellite galaxy catalogues · dynamical modelling · numerical methods
Cosmic web · Topology Open

Topology of the cosmic web

The cosmic web is not random. Filaments connect nodes, voids are bounded by walls, and these structures persist — or dissolve — across cosmic time. Topology gives us a language to describe their shape and their fate.

This project uses persistent homology and caustic theory to characterise when and where structures in the cosmic web form, merge, and disappear — asking which topological features are robust signals of the underlying cosmology, and which are noise.

Methods: persistent homology · N-body simulations · cosmic web segmentation
Dark energy · Observations Open

Reconstructing dark energy from cosmic chronometers

We don't need to assume a model for dark energy to measure it. Cosmic chronometers — old, passively evolving galaxies whose relative ages tell us how fast the universe was expanding at different moments — give us H(z) directly from observations. What does that expansion history tell us about the nature of gravity itself?

This project uses H(z) measurements from cosmic chronometers to reconstruct the effective equation of state of dark energy, ω_x(z), within the framework of f(R) gravity — without restricting to any specific functional form. The goal: what family of f(R) theories is consistent with what we actually observe?

Methods: Gaussian process regression · H(z) data · f(R) gravity · statistical inference
Dark energy · Bayesian inference Open

Bayesian comparison of dark energy models

The standard cosmological model has a well-known theoretical problem: nobody knows why the cosmological constant has the value it does. Dozens of alternatives have been proposed. But rather than testing them one by one, can we let the data itself rank them objectively?

This project applies Bayesian model comparison to a set of physically motivated dark energy equations of state — CPL, SEoS, JJE, and the ONE parameterisation family — using CMB, BAO, Type Ia supernovae, and cosmic chronometer data. The goal is to compute Bayes factors and ask which models the data actually prefers.

Methods: Bayesian inference · MCMC (Cobaya) · Savage–Dickey ratio · CLASS Boltzmann solver

Interested? Get in touch.

Don't see exactly what you're looking for? Write anyway — I'm always open to discussing ideas that connect to the group's research themes.

Contact →