Stacking the Deck: Tunable Trainability in Stacked LCUs
2026-07-27 • Machine Learning
Machine Learning
AI summaryⓘ
The authors study a type of quantum circuit used for near-term quantum computing called variational quantum circuits, which often face a problem called barren plateaus that make training difficult. They introduce a new design called stacked linear combination of unitaries (S-LCU) that balances how hard the circuit is to simulate on classical computers and how easy it is to train. By adjusting layers in their design, they provide a way to control this balance with proven mathematical bounds. This method helps users choose circuit designs that best match their computing needs and hardware limitations.
Variational quantum circuitsBarren plateausQuantum advantageClassical simulabilityUnitary operatorsStacked linear combination of unitaries (S-LCU)Fermionic Gaussian unitariesLoss landscapeQuantum gate complexityFree Fermion model
Authors
Nikhil Khatri, Stefan Zohren, Gabriel Matos
Abstract
Variational quantum circuits have been central to many proposed near-term applications of quantum computing, but a growing body of evidence suggests that trainability and quantum advantage are fundamentally at odds: ansätze expressive enough to resist efficient classical simulation tend to exhibit barren plateaus, while structures that provably rule out barren plateaus typically render them classically simulable. We propose a stacked linear combination of unitaries (S-LCU) as a variational ansatz which provides a tunable trade-off between barren plateaus and classical simulability. Using a diagrammatic analysis, we bound the loss-landscape variance of the Free Fermion S-LCU, whose elements are fermionic Gaussian unitaries. We prove a variance lower bound of $Ω(1/(n k^{3l}))$, with a simulation cost of $O(k^{2l} n^3)$ using the best known classical algorithm, compared to a quantum gate complexity of only $O(lkn^2)$. The number of layers $l$ serves as a single dial that trades computational complexity against the rate of cost concentration. This offers practitioners a systematic method for constructing ansätze with a complexity-trainability trade-off that best suits their application and hardware.