Skip to main content
Back to KTH start page

Master/Bachelor degree research projects

I warmly welcome KTH master students as well as exchange students to work with us on different subjects of the many-body theory.  

 

 

Master thesis project (Fall 2026 until position is filled): A machine-learning emulator for fission and the de-excitation of fission fragments

 

Background.

Fission is central to nuclear energy and applied nuclear physics, and it remains a key open problem in fundamental nuclear physics. There is still no complete theory of fission. Event generators such as CGMF follow how each fragment cools by emitting neutrons and γ rays in a Hauser–Feshbach cascade. Other codes, such as FREYA and GEF, combine empirical or semi-empirical fragment yields with simpler evaporation models. Full calculations are slow, which limits their use in uncertainty quantification and transport simulations. Our aim is to build fast emulators for two stages: the potential energy surface that governs the splitting, and the single emission step in the de-excitation. For the second stage, a network learns the probability of the next emission, its type (neutron or γ), its energy and its change in spin, given the current state of the nucleus (A, Z, U, J, π). The cascade is then generated by repeating this step until the fragment reaches its ground state.

We expect up to two students working on one of the following directions:

Fission emulationbased on the potential energy surface. The emulator can use input from nuclear models developed in our group, or from a new energy functional based on an energy theorem we have recently proposed.

Emulation of the de-excitation.Extract step-by-step emission data from CGMF. Train a conditional density model, such as a mixture density network or a normalizing flow, for the emission step. Validate against full CGMF calculations for ²⁵²Cf(sf) and ²³⁵U(n,f): neutron and γ multiplicity distributions, prompt neutron and γ spectra, and n–γ correlations. Test on reactions not used in training. Vary the TKE distribution and the spin cutoff parameter, and check that the emulator follows CGMF. Measure the speed-up.

Supervisor: Chong Qi, chongq@kth.se


Profile picture of Chong Qi

Portfolio