PI: Matteo Vorabbi

Ab initio nuclear theory has achieved significant advances, enabling predictive calculations of structure observables in medium-, heavy-, and deformed nuclei directly from QCD consistent theories of the nuclear force. However, the microscopic description of nuclear reactions remains a major open challenge and requires new cutting-edge computational techniques on that can exploit large scale supercomputing resources.

The self-consistent Green’s function (SCGF) method provides a particularly natural framework to address structure and reactions within a unified formalism. In SCGF, the nuclear elf-energy generalizes the exact optical potential and is constructed through a systematic expansion in Feynman diagrams that encode the underlying interaction processes.

In standard truncation schemes, diagrams are included consistently up to third order. While these approximations have proven accurate for structure calculations, they lack higher-order correlations that are essential for describing absorption due to inelastic reaction channels. To solve this limitation the entire infinte series of Feynman diagrams is reformulated as a statistical sampling of diagrammatic contributions (both in configuration and topological spaces).

Ref. [1] developed the first Diagrammatic Monte Carlo (DiagMC) algorithm for a finite model at zero temperature, and exploited DiRAC3 facilities to apply it to the Richardson model of nuclear pairing.

DiagMC resums the self-energy by stochastically sampling diagram topologies and quantum numbers at the same time, thereby including processes of arbitrarily high order. Our implementation sampled the ladder expansion up to order eight, achieving better precision than the state-of-the-art ADC(3) method. DiagMC is currently being extended to true nuclei with QCD-based EFT interactions, with promising results already obtained up to fifth order.

[1] Stefano Brolli, Carlo Barbieri, and Enrico Vigezzi. “Diagrammatic Monte Carlo for Finite Systems at Zero Temperature.” Phys. Rev. Lett. 134, 182502 (2025)