This paper explains how quantum many-body chaos emerges when an otherwise solvable system is slightly perturbed. The authors study a one-dimensional circuit made from two kinds of two-qubit gates. One kind (matchgates) makes the circuit equivalent to free fermions and is integrable, meaning the dynamics remain simple. The other kind (SWAP gates), inserted at a tunable density λ, breaks integrability and can produce chaotic operator spreading. The key result is a concrete, quantitative picture of the crossover from early-time integrable behaviour to late-time chaos, including the characteristic time and length scales that control it.

The model is a brickwork circuit on a chain of qubits, which can also be described in terms of Majorana fermions. When λ = 0 the circuit has only matchgates and is exactly integrable. At finite λ some gates are SWAPs, which generate quartic-in-fermions interactions and act as local “hotspots” for chaos. By averaging over the random choices of matchgates the authors map the averaged out-of-time-ordered correlator (OTOC) — a standard probe of how quickly quantum information scrambles — to an exact classical Markov process. That mapping lets them obtain statistically exact results for arbitrarily large systems.

The OTOC behaviour changes qualitatively with λ. At λ = 0 the OTOC spreads diffusively with a Gaussian profile and no ballistic front. This matches the picture of free particles wandering in a noisy environment. At any nonzero λ, the SWAP gates seed local amplifications of the OTOC. Over time these spots accumulate and the average OTOC develops a ballistic front that moves outward with a butterfly velocity vB, while the front itself broadens diffusively. The authors give a scaling form for this chaotic regime and report that vB scales roughly like √λ (with logarithmic corrections). The diffusion constant associated with the front broadening depends little on λ.