Curator's Take
AI Commentary
This article reveals that the shape of entanglement in a quantum circuit fundamentally limits how cheaply it can be “cut” into smaller pieces without losing any quantum advantage, showing that popular MPS‑ and TTN‑based ansätze are always classically simulable once they become efficiently cuttable. By engineering circuits where seam entanglement stays low while global bond dimension explodes, the authors demonstrate a concrete route to retain hardness even under aggressive cutting, and they further point out that using magic (T‑gates) rather than raw entanglement sidesteps the depth trade‑off between hardness and trainability. The results give algorithm designers a clear diagnostic—entanglement geometry versus magic content—for building scalable variational workloads that remain both cuttable on near‑term hardware and resistant to classical simulation.
— Mark Eatherly
Summary
Circuit cutting promises to scale quantum computations beyond current hardware, but variational quantum advantage also requires low cutting overhead, classical hardness, and trainability. We show that these properties are strongly constrained by entanglement geometry. Matrix product state (MPS) and tree tensor network (TTN) circuits with constant seam bond dimension can be cut with \(O(1/\varepsilon^2)\) sampling overhead, but remain efficiently classically simulable, ruling out asymptotic quantum advantage within these families. By independently controlling seam and intra-block entanglement, we construct a two-block circuit family that remains cheaply cuttable while requiring a super-polynomial global MPS bond dimension, as supported numerically up to \(n=100\). However, MPS hardness and trainability require incompatible depth regimes, \(d=ω(\log n)\) and \(d=O(\log n)\), respectively. Using magic rather than entanglement as the hardness resource avoids this conflict: shallow Clifford+\(T\) circuits remain cuttable and trainable while their stabiliser-simulation cost grows exponentially with the \(T\)-count.