Thèophile Bonnet

Dr Théophile Bonnet

Post Doctoral Researcher · University of Cambridge

Focus - Monte-Carlo methods for reactor physics 2024 – Present

Applied mathematician and reactor physicist working on Monte-Carlo methods for neutron transport. My work develops faster, more reliable ways to quantify sensitivity, noise and convergence in reactor calculations — turning stochastic-process theory into tools that nuclear codes can run in practice. Within MaThRad I connect the programme's probabilistic core to reactor-physics practice with industrial and national-lab partners (UKAEA, CEA).

Body of work

5 items

Year

Software 2025 → present

Generalized polynomial chaos for reactor-physics sensitivity

SCONE Monte-Carlo code · prototype branch · with Dr Dubey & Prof. Shwageraus

A generalized polynomial-chaos algorithm for k-eigenvalue and fixed-source calculations that computes the sensitivity of reactor calculations to geometrical and density perturbations. Implemented as a prototype branch of the SCONE Monte-Carlo code, aimed at sharper sensitivity analysis for reactor cores and fusion breeding blankets.

›Published — "Addressing geometrical perturbations by applying generalized polynomial chaos to virtual density in continuous-energy Monte-Carlo power iteration", Nuclear Science & Engineering (2026) · arXiv:2506.06046. DOI
›Presented as an invited seminar at NTU Singapore (Oct 2025) and at a workshop with UKAEA (Oct 2025).

Linked · code Reactor physicsSensitivityPrototype

Model 2025 · Apr

Stochastic models of neutron-population dynamics at reactor start-up

M&C 2025 · stochastic transport · martingale formulation

A stochastic-differential-equation (martingale) formulation of the neutron population during reactor start-up that captures reactor noise and is benchmarked against time-dependent Monte-Carlo solutions of the neutron transport equation. The M&C 2025 paper analyses how negative sample paths affect the accuracy of stochastic point-kinetics solutions.

›Conference paper submitted to PHYSOR 2025 (with Saltfoss).

Linked · M&C 2025 Stochastic processesReactor noise

Model 2025 · in progress

Clustering & source-convergence diagnostics in heterogeneous media

Collaboration with A. Cox, O. Tough & T. Sutton

Characterises clustering of the neutron population in heterogeneous media and develops a source-convergence diagnostic for Monte-Carlo calculations with large dominance ratio, improving the reliability of such simulations. Early-stage collaborative work, currently documented in correspondence.

Asserted Monte CarloIn progress

Industrial co-design 2025 → present

Sensitivity tools for fusion-blanket modelling with UKAEA

UK Atomic Energy Authority · breeding-blanket modelling

UKAEA needs accurate, efficient sensitivity calculations for breeding-blanket modelling. The polynomial-chaos sensitivity method is being co-designed with them through a validation exercise: David Foster (UKAEA) has confirmed interest by email, and the method was presented to UKAEA at an October 2025 workshop.

Impact trajectory

Prototype

Engaged

Validation

Adoption

Asserted UKAEAFusion

Industrial co-design 2025 → present

Coupled Monte-Carlo depletion with CEA

CEA-Saclay · coupled Monte-Carlo calculations

CEA is interested in improving its coupled Monte-Carlo calculations, which the branchless-collision work is expected to make more stable and efficient. The collaboration is at the scoping stage, supported by email exchanges and a research visit to CEA-Saclay.

Impact trajectory

Concept

Scoping

Co-design

Adoption

Asserted CEADepletion

Output over time

2024 — 2026

2024

2025

2026

Research

Joined MaThRad · reactor-physics thread

Polynomial-chaos sensitivity in SCONE

SDE model of reactor noise · M&C 2025

Polynomial-chaos paper published · NSE

Clustering & depletion studies · in progress

Collab

Impact

UKAEA sensitivity co-design begins

CEA-Saclay visit · coupled-MC scoping