<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title>Ziqi Xu</title>
  <link href="https://helioxv022.github.io/feed.xml" rel="self"/>
  <link href="https://helioxv022.github.io/"/>
  <id>https://helioxv022.github.io/</id>
  <updated>2026-09-24T00:00:00Z</updated>
  <author><name>Ziqi Xu</name></author>
  <entry>
    <title>Expressive is not the same as learnable</title>
    <link href="https://helioxv022.github.io/posts/2026-09-structure-matched-learning/"/>
    <id>https://helioxv022.github.io/posts/2026-09-structure-matched-learning/</id>
    <updated>2026-09-24T00:00:00Z</updated>
    <summary>A rotated cluster state is an exact bond-dimension-2 matrix product state, so the complex variational models I trained never lacked the capacity to represent it. How easily each one learned it tracked whether its hidden update had the shape of that two-dimensional memory.</summary>
  </entry>
  <entry>
    <title>What phase disorder does to the GHZ advantage</title>
    <link href="https://helioxv022.github.io/posts/2026-09-ghz-phase-disorder/"/>
    <id>https://helioxv022.github.io/posts/2026-09-ghz-phase-disorder/</id>
    <updated>2026-09-23T00:00:00Z</updated>
    <summary>With squint, a JAX library for differentiable quantum metrology, the quantum Fisher information of a GHZ sensor array under random signal phases takes a few lines. The Heisenberg advantage disappears, and a handful of X gates brings most of it back.</summary>
  </entry>
  <entry>
    <title>How far do measurement correlations reach in a cluster state?</title>
    <link href="https://helioxv022.github.io/posts/2026-09-cluster-state-cmi/"/>
    <id>https://helioxv022.github.io/posts/2026-09-cluster-state-cmi/</id>
    <updated>2026-09-22T00:00:00Z</updated>
    <summary>Measuring a 1D cluster state in a tilted basis correlates distant outcomes. An exact matrix-product-state calculation shows how far those correlations reach.</summary>
  </entry>
</feed>
