Further Reading: Multi-Qubit Programming

Tagged Tier 1 (confident it exists and recommended) and Tier 2 (real and worth seeking, but verify the current edition or URL).

Entanglement has an enormous literature spanning foundations, information theory, and engineering. The selection below is deliberately narrow: what a programmer benefits from, roughly in the order you would want it.

The essentials

• Nielsen and Chuang, Quantum Computation and Quantum Information, §1.3.6, §2.2.8, and Chapter 2's discussion of the Schmidt decomposition. The standard treatment of the tensor product, reduced density matrices, and what entanglement means formally. §2.5's Schmidt decomposition is the tool behind Exercise 4.16 and behind matrix-product-state simulation in Chapter 11. Tier 1.
• Appendix D of this book, §D.4. The tensor product, with the numpy.kron correspondence and the endianness discussion spelled out. If §4.1 moved fast, read this. Tier 1.
• Appendix B. Every two- and three-qubit gate with its matrix in Qiskit's convention, its circuit symbol, and its decomposition cost. The page to keep open while writing Part IV. Tier 1.

On what entanglement actually is

• John Preskill, lecture notes on quantum information, Chapter 4 ("Quantum Entanglement"). The best available treatment for someone who wants rigor without a full quantum information course. Careful about exactly the distinctions §4.4 makes — pure versus mixed, reduced states, and why a subsystem of a pure entangled state is maximally mixed. Free online. Tier 1.
• Ryszard Horodecki, Paweł Horodecki, Michał Horodecki, and Karol Horodecki, "Quantum Entanglement" (2009), Reviews of Modern Physics 81, 865. The comprehensive review. Far more than a programmer needs, and the right place to go when you want the real answer about separability criteria, entanglement measures, or witnesses. Sections on witnesses are directly relevant to Case Study 2. Tier 1.
• Asher Peres, "Separability Criterion for Density Matrices" (1996), Physical Review Letters 77, 1413, and the Horodecki follow-up. The PPT (positive partial transpose) criterion: a genuinely practical, computable test for whether a two-qubit state is entangled, stronger than the reduced- state test for mixed states. Worth implementing once. Tier 1.

On multipartite entanglement (GHZ vs W)

• W. Dür, G. Vidal, and J. I. Cirac, "Three Qubits Can Be Entangled in Two Inequivalent Ways" (2000), Physical Review A 62, 062314. The paper behind §4.7. It proves that GHZ and W states cannot be converted into each other by local operations and classical communication — that they are genuinely different resources, not two points on a spectrum. Readable, and the result is more surprising than the abstract makes it sound. Tier 1.
• Greenberger, Horne, and Zeilinger, "Going Beyond Bell's Theorem" (1989). The original GHZ paper. Its argument against local realism is, unlike Bell's, a deterministic contradiction rather than a statistical one — a single measurement outcome that no local theory can produce. One of the most elegant arguments in physics. Tier 2 — reprinted in several collections; find whichever version you can.

On witnesses and proving entanglement (Case Study 2)

• Otfried Gühne and Géza Tóth, "Entanglement Detection" (2009), Physics Reports 474, 1. The reference on entanglement witnesses: how to construct them, what they prove, and their limits. If you ever need to make a defensible entanglement claim about experimental data, this is the source. Tier 1.
• J. S. Bell, "On the Einstein Podolsky Rosen Paradox" (1964), Physics 1, 195. Six pages that changed the subject. This book never runs CHSH — Case Study 2's two-correlator witness is a simpler relative of it, and Chapter 38 §38.6 notes the family resemblance while measuring something else. Read Bell's original once for the argument's shape, then run Case Study 2's witness for a number you can defend. Tier 1.
• Clauser, Horne, Shimony, and Holt (1969), Physical Review Letters 23, 880. The CHSH inequality — the experimentally practical version of Bell's argument, and the direct ancestor of the two-correlator witness in Case Study 2. Tier 1.
• The 2022 Nobel Prize in Physics scientific background document (Aspect, Clauser, Zeilinger). A clear, authoritative summary of the experimental history of Bell tests, including the loopholes that took decades to close. Free from the Nobel Foundation, and unusually well written. Tier 2 — check the current URL.

On the hardware side (Case Study 1)

• The Qiskit transpiler documentation on preset pass managers and layout stages. What each optimization level actually does about layout — TrivialLayout, VF2Layout, SabreLayout — and how noise-awareness enters. The authoritative answer to "why did level 3 choose different qubits?" Tier 1.
• The VF2PostLayout pass documentation specifically. The scoring mechanism that finds a better-quality isomorphic layout using error data. This is the pass that saved the $n=7$ GHZ state. Tier 2 — pass names have moved across releases.
• Chapter 29 of this book. Case Study 1 is a preview of that chapter; the full treatment includes noise-aware layout selection, routing cost, idle-time management, and the general "same circuit, two layouts, two answers" demonstration. Tier 1.

If you want the philosophy

Optional, and genuinely interesting once you have the operational picture.

• N. David Mermin, "Is the Moon There When Nobody Looks?" (1985), Physics Today 38(4), 38. The best popular-level explanation of Bell's theorem ever written, by someone who understood it completely. Uses a device with three switches and two lights and requires no mathematics. Tier 1.
• Tim Maudlin, Quantum Non-Locality and Relativity. For readers who want to know what entanglement implies about the structure of the world, and are prepared for a careful philosopher to tell them that most of the popular answers are wrong. Not needed to program anything. Tier 2 — check for the current edition.

Where to go next. If one thing: read Mermin's "Is the Moon There When Nobody Looks?" It takes twenty minutes and it will permanently improve how you explain entanglement to other people, which turns out to be a large part of this job.

If you would rather build: Chapter 5, where several claims deferred here — how many shots you need, how to measure in another basis, and how to tell whether a distribution is the one you expected — finally get settled.