ACM Doctoral Dissertation Award: NYU’s Allen Liu, Three‑Time IMO Gold Medalist, Wins

ACM Doctoral Dissertation Award 2025

2025 ACM Doctoral Dissertation Award (US$20,000) and two honor‑mention awards (US$10,000 each) were announced on 10 June. The winning dissertation is published in the ACM Digital Library series.

Winner – Allen Liu (NYU)

Dissertation: Learning Theoretic Foundations for Understanding Quantum Systems . Focuses on two fundamental problems for local‑interaction many‑body Hamiltonians:

Characterizing equilibrium properties of a given Hamiltonian.

Reconstructing the Hamiltonian from measurement data.

Key technical contributions:

Proof of a universal law: at a geometry‑dependent critical temperature, entanglement undergoes a sudden death that is independent of system size.

First efficient algorithm that recovers an arbitrary Hamiltonian at any temperature, overcoming previous barriers at low temperature.

New learning paradigm that overturns several previously held beliefs about quantum system behavior.

Analysis of learning and testing general quantum states under realistic device constraints, establishing optimal sample complexities for single‑copy and multi‑copy measurements.

Dissertation URL: https://dspace.mit.edu/entities/publication/86bf5543-05b9-45e0-9cfc-2cc342559582

Honorary Mention – Gal Arnon (Bocconi University)

Dissertation: New Advancements in Interactive Oracle Proofs: Theory, Practice, and Limitations . Advances the theory of interactive oracle proofs (IOPs).

Shows equivalence between IOPs with small query complexity and traditional interactive proofs, establishing an IOP analogue of the PCP theorem.

Constructs a new NP‑complete IOP with low soundness error and reduced query complexity.

Develops an improved proximity IOP for Reed–Solomon codes.

Identifies fundamental obstacles to building highly efficient IOPs and PCPs.

Dissertation URL: https://galarnon42.github.io/gal_thesis.pdf

Honorary Mention – Rachit Nigam (MIT)

Dissertation: Modular Abstractions for Efficient Hardware Design . Investigates the tension between modularity and efficiency in hardware design and argues that explicit time‑sensitive reasoning is essential.

Three systems illustrate the approach:

Dahlia : an imperative language compiled to hardware that uses time‑sensitive reasoning to generate efficient circuits.

Calyx : a compiler and intermediate language that bridges software‑style control flow with hardware‑style structural constructs, enabling fine‑grained, time‑aware scheduling.

Filament : a hardware description language that models cycle‑level constraints directly in module interfaces and guarantees conflict‑free designs at compile time.

Collectively, the systems demonstrate that modeling time at each abstraction layer is crucial for scaling hardware design tools in the era of specialized accelerators.

Dissertation URL: https://people.csail.mit.edu/rachit/files/pubs/dissertation.pdf