Skip to content
NewA calmer, faster ZeroHour

Barren plateaus analyzed for quantum linear combinations of unitaries

Original title: Barren Plateaus in Quantum Linear Combinations: Interference and Lie-Algebraic Control

arXiv cs.AI / cs.LG / cs.CLby Amir Koutahi ()1 min read
AI summary · x-ai/grok-4.7Summaries by Vani AI

A single-pass LCU keeps a gradient-variance lower bound, while interleaved Haar sampling can concentrate exponentially.

The paper separates a single linear combination of unitaries from circuits that repeatedly interleave branch operations and ancilla mixing. For a single-pass LCU, independent zero-mean branches cannot erase a differentiated branch’s gradient variance, and the bound does not require Haar-random unitaries. Exact loss and gradient formulas include interference terms, one of which depends on relative coefficient phases. Matchgate examples show polynomial single-pass variances and exponential concentration under joint Haar-random sampling.

Key facts
  • The variance lower bound is the branch variance times the fourth power of its coefficient magnitude.
  • The bound holds at any gate position without Haar-random circuit unitaries.
  • A four-qubit matchgate example shows a phase-dependent interference term can be nonzero.
  • The paper also gives coefficient-selection rules and bounds for postselection and measurement error.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

From the original article

1 min read at arxiv.org

A coherent sum of quantum circuits introduces interference that can alter the training signal. We distinguish a single linear combination of unitaries (LCU) from a circuit that repeatedly alternates branch operations and ancilla mixing.

arXiv cs.AI / cs.LG / cs.CL

By Amir Koutahi · arxiv.org

Read the full article at arXiv cs.AI / cs.LG / cs.CL

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.