Further Reading: Measurement, Shots, and Statistics

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

This chapter sits on two literatures that rarely meet: the quantum-mechanical account of measurement, and ordinary applied statistics. You need much more of the second than most quantum resources acknowledge.

On measurement, quantum-mechanically

• Nielsen and Chuang, §2.2.3 through §2.2.6. Projective measurement, POVMs, and observables done properly. §2.2.5 on "distinguishing quantum states" is the formal version of §5.6's complementarity table, and it is worth reading for the precise statement of what cannot be distinguished. Tier 1.
• John Preskill, lecture notes on quantum information, Chapter 3. Measurement, density operators, and the partial trace — the machinery behind §5.7 and behind Chapter 4's reduced-state test. Free, careful, and the best treatment of why the partial trace is the right operation rather than merely a definition. Tier 1.
• The Qiskit quantum_info documentation on SparsePauliOp and the Statevector.expectation_value method. The API for everything §5.5 does by hand. Once you understand the parity computation, use the library. Tier 1.

On the statistics you actually need

This is the gap. Almost no quantum computing resource teaches it, and almost every quantum result depends on it.

• David Spiegelhalter, The Art of Statistics. The best available introduction to statistical reasoning for a technical reader who is not a statistician. Chapters on uncertainty intervals, hypothesis testing, and — crucially — the multiple-comparisons problem that Case Study 2's Problem 5 describes. Non-mathematical, and it will improve every result you ever report. Tier 1.
• Regina Nuzzo, "Scientific Method: Statistical Errors" (2014), Nature 506, 150. A short, widely read account of what p-values do and do not mean. If you read nothing else on statistics, read this; it takes fifteen minutes and it inoculates you against the most common misreading in §5.8. Tier 1.
• The American Statistical Association's 2016 "Statement on p-Values." Six principles, two pages, from the professional body. Principle 3 — "scientific conclusions should not be based only on whether a p-value passes a specific threshold" — is the formal version of §5.8's 🔬 Honest Assessment. Tier 1.
• Jacob Cohen, Statistical Power Analysis for the Behavioral Sciences. The standard reference on statistical power — the concept §5.8 introduces and Case Study 2 turns into a design decision. The examples are from psychology; the mathematics is identical for quantum experiments. Consult rather than read. Tier 2 — check for the current edition.
• The SciPy stats documentation, particularly chisquare, binomtest, and bootstrap. The tools. binomtest is often the better choice for the two-outcome case in §5.8, and bootstrap is the most practical way to get an error bar on a quantity whose analytic variance you do not want to derive. Tier 1.

On shot budgeting and measurement cost (Case Study 1)

• Wecker, Hastings, and Troyer, "Progress Towards Practical Quantum Variational Algorithms" (2015), Physical Review A 92, 042303. The paper that made the shot-count problem visible: it works out what VQE actually costs in measurements and finds the numbers alarming. The direct ancestor of Case Study 1's arithmetic, and it is still the right framing. Tier 1.
• The literature on Pauli grouping and commuting-set partitioning — search for "qubit-wise commuting grouping VQE" and "measurement reduction variational quantum eigensolver." An active area with real results; Case Study 1's Reduction 1 is the crude version. Worth a survey read before any serious VQE project. Tier 2 — a moving target, so read recent work.
• Brassard, Høyer, Mosca, and Tapp, "Quantum Amplitude Amplification and Estimation" (2002). The $1/N$ scaling referenced in §5.4 and Question 7 of Case Study 1. Understanding what it demands of the circuit — deep coherent evolution — is understanding why it is not currently usable, and why it is the thing to reach for once error correction arrives. Tier 1.

On reporting quantum results honestly

• The reproducibility guidance issued by major physics journals for quantum computing submissions. Increasingly explicit about reporting device, layout, calibration window, shot count, and error bars. Read your target journal's version before you write up anything; the requirements are exactly Case Study 2's lessons, formalized. Tier 2 — journal-specific and evolving.
• Chapter 30 of this book. Everything in Case Study 2 generalized: what a benchmark must control for, how to design a fair comparison between devices or strategies, and how to read a vendor's claim. Tier 1.
• Chapter 27. The testing counterpart — statistical assertions, tolerances, and CI for probabilistic output. §5.8's warning that a test cannot establish correctness is what that chapter is built around. Tier 1.

If you want the foundations

Optional, and it will not help you write better code — but the measurement problem is genuinely unresolved and it is worth knowing that.

• Maximilian Schlosshauer, Decoherence and the Quantum-to-Classical Transition. The careful modern treatment of what "measurement" means physically, and of what decoherence does and does not explain. Relevant to Chapter 11's noise models as well as to the philosophy. Tier 2.
• Sean Carroll, Something Deeply Hidden. A readable argument for one particular resolution (many-worlds), and — more useful — a clear account of why the measurement problem is a problem at all. Take the conclusion as one position among several. Tier 2.

Where to go next. If one thing: Regina Nuzzo's Nature piece on p-values. Fifteen minutes, free, and it will change how you read every quantum result you encounter — including the ones in this book.

Then Chapter 6, which closes Part I with the assembly language underneath every framework.