Benjamin Suzzoni

Researcher in theoretical physics

Suzzoni Picture
Mount Fuji

Research interests

Neural-Network Field-Theory (NNFT)

Neural-networks have emerged as powerful approximators of quantum fields. Taking the limit where the number of parameters in a neural-network becomes infinite (eg. infinite width, or infinite depth), these turn out to approximate any given function in a given class of functions. Using this limit, one can build an approximation to the path integral of a given quantum field theory. The neural networks then play the role of fields, and their parameter distribution plays the role of the action.

NNFT then provides a fruitful interface between Physics, or more specifically field theory, and Machine Learning, or more specifically the theory of neural-network. I am interested in studying this interplay between the two fields, and see what our knowledge of quantum field theory can tell us about neural-networks.

Superconformal field theory (SCFT)

Superconformal field theories are uniquely positioned in the space of field theories. The cancellation between fermionic and bosonic contributions leads to many computations being attainable analytically. Being interconnected by webs of dualities across energy scales and dimensions, these theories enable us to tackle old problems in novel ways.

Their tight connection with topological invariants of manifolds, and mirror symmetry of Calabi-Yau varieties also makes them uniquely interesting from a Mathematician's perspective.

I've been interested in finding new properties of the superconformal index of certain SCFTs. This can be approached from a background supergravity perspective (Festuccia-Seiberg), or by looking at properties of its UV construction, say for D3-branes probing a toric Calabi-Yau 3-fold.

Holography or AdS/CFT

The anti-de-Sitter/conformal field theory (AdS/CFT) correspondence has introduced novel ways of studying the strong-coupling regime of SCFTs. By leveraging its weak form, we can study strongly-coupled SCFTs at large rank using classical supergravity.

This is the standard approach used in computing holographic entanglement entropy and large-\(N\) limits of the superconformal index.

Publications



Auroras in Norway

Past positions



2025 – now

Postdoctoral researcher, UNIST.

2021 – 2025

PhD, Southampton.