Notes

Notes · PQP digest

PQP vocabulary mapping

Several Nadarasa terms are homemade. This page maps them to the standard vocabulary of Picturing Quantum Processes (Coecke & Kissinger, CUP 2017) so reviewers can locate each operation in the book. Built from the digest at quantum/PQP_DIGEST.md.

Term-by-term mapping

Our termPQP termChapter
modal projectorclassical post-selection on a bastard-spider outputCh. 8 §8.3.3 + §8.4
predictor derived from the gatesspider fusion + doubling-BornCor. 8.35 + Eq. 6.4
structured-reflection promise (SRP)structure on the phase-group of the classical subgroupCh. 9 §9.3.6 + Ch. 11 Thm 11.12
per-window √p improvementamplitude-amplification cost (Grover) — NOT HSPCh. 12 §12.2.3 vs §12.2.4
GAP ledger entryA ⋡ B in (P, F) — resource gap under implemented rewritesCh. 13
kernel cannot signalcausality (discard out = discard in)Def 6.52 + Eq. 10.31
live sampler / shot streamONB measurement as doubled effectDef 7.1, Def 7.6

modal projector: Not a named primitive in PQP — it is a perfectly well-defined operation that the book never calls out by name. Bastard-spider fusion (Thm 8.72) is what gives the closed form.

predictor derived from the gates: v0.3 methodology rule operationalised. Not heuristic discipline — the dodo theorem (Thm 8.41) says the diagram uniquely determines the basis.

structured-reflection promise (SRP): Reads as a constraint on which phase-group elements survive spider fusion. The Z₄ phase-group fingerprint test is the same machinery applied as a litmus.

per-window √p improvement: The original G3 framing confused Grover-style √-cost with HSP-style log-cost. Ex. 12.22 explicitly disambiguates: Deutsch–Jozsa for n > 1 is NOT HSP.

GAP ledger entry: A gap becomes 'certified' when an additive monotone (T-count, spider-count) provably non-increasing under F separates the two sides. See /nadarasa/resources.

kernel cannot signal: Non-signalling is a derived consequence of causality (Thm 6.84), not an extra postulate.

live sampler / shot stream: State collapse as a one-line spider equation; tomography is the Ch. 7 §7.4 picture.

Citations the v0.3 rule depends on

  • Eq. 6.4Born rule from the doubling constructionCh. 6 · Quantum processes

    Every pure quantum map is f ⊗ f̄; composing an effect on a state then yields |⟨ψ|φ⟩|² automatically. This is why deriving the predictor from the gates is not heuristic — it is the definition.

    Anchors: v0.3 methodology rule (predictor from gates) · G1 / G2-real / G12 closed-form derivations

  • Thm 8.34 · Cor 8.35Spider fusion theoremCh. 8 · Classical-quantum processes

    Any connected same-colour spider diagram fuses to a single spider determined only by leg count. Operationally: closed forms collapse to one expression per branch — exactly what the gate-derived predictors compute.

    Anchors: v0.3 methodology rule · modal projector / G1 / G12 simplifications

  • Thm 8.41Dodo theorem (no spider impostors)Ch. 8 · Classical-quantum processes

    Any family of maps satisfying spider fusion IS a real spider family for some ONB. So 'derive the predictor from the gates' is not just heuristic discipline — the diagram uniquely determines the basis up to relabelling.

    Anchors: v0.3 methodology rule

  • Def 6.52 · Eq. 10.31Causality criterion (discard out = discard in)Ch. 6 / Ch. 10

    A quantum process is causal iff discarding its output equals discarding its input. Non-signalling between subsystems is a derived consequence (Thm 6.84), not an extra postulate.

    Anchors: G2-real signalling check · any 'kernel cannot signal' claim

  • Thm 9.128Clifford-ZX completenessCh. 9 · Phases and complementarity

    The 4-rule Clifford ZX-calculus is complete: any equality of Clifford circuits is provable by the 4 rules alone. Justifies treating rewrites as a sound monotone-tracking system on the Clifford fragment.

    Anchors: resource-monotone certification · G13 phase-group test

  • JPV 2017 + NW 2017Full-ZX completeness (post-publication)Successor results, not in PQP

    Jeandel–Perdrix–Vilmart 2017 completed Clifford+T ZX; Ng–Wang 2017 completed full ZX. Both post-date PQP and are what makes modern rewriters (PyZX, quizx) theoretically grounded.

    Post-pub: JPV: arXiv:1705.11151 · Ng–Wang: arXiv:1706.09877

    Anchors: full-ZX rewriting

  • Thm 11.12Phase-group fingerprint for non-localityCh. 11 · Quantum non-locality

    A ZX-style theory exhibits GHZ-style non-locality iff its phase group contains Z₄ (vs. Z₂×Z₂ for Spekkens' toy theory). A single litmus test on the phase angles a kernel actually uses.

    Anchors: G13 phase-group test

  • §12.2.3 vs §12.2.4 · Ex. 12.22HSP exponential ≠ Grover square-rootCh. 12 · Quantum computation

    HSP uses O(log|G/H|) queries (exponential). The √-factor only appears in Grover/amplitude amplification. Ex. 12.22 explicitly notes Deutsch–Jozsa for n>1 is NOT an HSP instance.

    Anchors: G3 split: grover vs hsp

  • Ch. 13Gaps as A ⋡ B in (P, F)Ch. 13 · Quantum resources

    A gap is precisely A ⋡ B in the partial order of diagrams under the implemented rewrites F. Additive monotones (T-count, spider count) provably non-increasing under F certify a gap as real, not conjectured.

    Anchors: resource-monotone certification

  • §14.1–14.3Quantomatic, !-boxes, conjecture synthesisCh. 14 · Automation

    !-boxes finitely encode the infinite spider-fusion family; conjecture synthesis discovers ~170 rules from generators alone. Practical successors (post-PQP): PyZX (Python), quizx (Rust).

    Post-pub: PyZX: arXiv:1904.04735 · quizx: github.com/zxcalc/quizx

    Anchors: scoping note: PyZX / quizx adoption

  • Topel 2026Fourier locking in re-uploading classifiers (spectral homotopy)Related work · post-PQP preprint

    Data re-uploading QML classifiers exhibit 'Fourier locking' — trainable spectrum collapses to a small integer lattice. Remedy is a continuous 'spectral homotopy' of the encoding schedule. Same family of question as our AQFT-under-noise cutoff and rule (M)'s discrete diagram deformations. See /nadarasa/related/fourier-locking.

    Post-pub: Topel, S. (2026), CC-BY-4.0 · github.com/moth-quantum/fourier_locking_study

    Anchors: G1 AQFT under noise · rule (M) as discrete spectral rewrite