From eb0dcbd241ed5f79b924978ca2b574e8b4604520 Mon Sep 17 00:00:00 2001 From: Ruge Lin Date: Sun, 8 Nov 2026 16:12:40 +0800 Subject: [PATCH 1/2] Explain exact canonical-purification transfer boundaries --- README.md | 5 + learning/MILESTONES.md | 18 +- learning/README.md | 3 + learning/TUTORIAL.md | 11 ++ learning/WATROUS_BRIDGE.md | 8 + llms.txt | 4 +- references/RELATED_WORK.md | 8 + references/bibliography.json | 24 +++ science/README.md | 1 + science/canonical_purification_boundary.md | 194 +++++++++++++++++++++ science/interpretation.md | 6 + 11 files changed, 272 insertions(+), 10 deletions(-) create mode 100644 science/canonical_purification_boundary.md diff --git a/README.md b/README.md index 71d2f30..fd812d3 100644 --- a/README.md +++ b/README.md @@ -12,6 +12,10 @@ $$ where $m$ and $n$ are the two copy budgets. Balanced supplies therefore require and suffice with $\Theta(d)$ copies of each program. The construction additionally returns the remaining programs within one global fidelity guarantee. +## Sister connection: exact canonical purification + +[Canonical-Purification-Cost](https://github.com/GoGoKo699/Canonical-Purification-Cost) studies a related contraction problem with an exact prescribed output and independent state/conjugate samples. Its sharp unequal-budget frontier is derived from exact-output rigidity and a finite LP, with mixed-source costs given by an exact binomial transform. The shared mechanism is coherent contraction with physical whole-space normalization; the unitary-Choi theorem below is not a premise of that frontier. The [comparison and transfer boundary](science/canonical_purification_boundary.md) fixes the different input correlations, exactness requirements and ownership. + ## Principal results | Result | Established guarantee | Scientific account | @@ -42,6 +46,7 @@ Choose a route according to the question you want to answer: | Which foundations connect to the proofs? | [Watrous bridge](learning/WATROUS_BRIDGE.md) | | How do the ideas fit into one learning route? | [Learning guide](learning/README.md) and [30 milestones](learning/MILESTONES.md) | | How can I reproduce the finite checks? | [Verification guide](verification/README.md) | +| Which contraction ideas apply to exact purification? | [Canonical-purification boundary](science/canonical_purification_boundary.md) and [sister project](https://github.com/GoGoKo699/Canonical-Purification-Cost) | The single educational anchor is John Watrous's *The Theory of Quantum Information* (2018). The bridge identifies relevant sections and translates the Choi normalization and fidelity conventions. The specialist proofs remain in the scientific account. diff --git a/learning/MILESTONES.md b/learning/MILESTONES.md index 2a3b84a..209fdbe 100644 --- a/learning/MILESTONES.md +++ b/learning/MILESTONES.md @@ -169,13 +169,13 @@ Three source conventions matter throughout: - **Conceptual connection:** Uniformity over unknown $U,V$ follows from the marginal structure, not from averaging the accuracy guarantee over those unitaries. - **Sources:** [Watrous §§2.1.3][w213] and [1.2.2][w122]; [exact moment derivation](../science/construction.md#3-exact-first-and-second-moments). -### M19 — Polar normalization preserves the full joint target +### M19 — Polar normalization controls the full joint target - **Central question:** How do the moments certify the composed program and every returned specimen together? - **Understanding target:** Connect the polar overlap to $(\mathbb E\sqrt X)^2$, with $X=G/\mathbb EG$, and follow the Hölder step yielding $F_{\rm global}\geq1/(1+r)=mn/(mn+d^2-1)$. Keep this unconditional contribution distinct from the support probability $p$ and the conditional fidelity $f_{\rm pol}/p$. - **Dependencies and story role:** M08, M11, M16–M18. Completes the constructive argument by joining physical branches to a global error guarantee. -- **Conceptual connection:** The bound remains informative when $1-r$ is negative. It certifies approximate joint return without asserting that every successful output is exact or that the bound is optimal. -- **Sources:** [Watrous §§1.1.3][w113], [1.2.2][w122] and [3.2.1][w321] for the spectral and fidelity language; [local Hölder derivation and success probability](../science/construction.md#4-fidelity-and-success-probability), [global theorem](../science/model.md#theorem-1-a-universal-collective-construction). +- **Conceptual connection:** The bound remains informative when $1-r$ is negative. It certifies approximate joint return without asserting that every successful output is exact or that the bound is optimal. Compare this tolerance for distortion with a prescribed pure target that must remain exact on success. +- **Sources:** [Watrous §§1.1.3][w113], [1.2.2][w122] and [3.2.1][w321] for the spectral and fidelity language; [local Hölder derivation and success probability](../science/construction.md#4-fidelity-and-success-probability), [global theorem](../science/model.md#theorem-1-a-universal-collective-construction), [polar versus exact normalization](../science/canonical_purification_boundary.md#the-shared-physical-normalization-problem). ### M20 — Correlated and independent program priors @@ -252,18 +252,18 @@ Three source conventions matter throughout: ### M29 — Returned programs and operational limits - **Central question:** Which further capabilities do the established promises leave unresolved? -- **Understanding target:** Explain why approximate joint return is a one-shot statement, why successive reuse needs accumulated-error accounting, and why a Choi output does not itself execute the product on arbitrary fresh data. Retain the pure-unitary-program and full-access assumptions when considering noise, mixed programs, many outputs or expected budgets. Translate the gate count into its dependence on a qubit operation's dimension. +- **Understanding target:** Explain why approximate joint return is a one-shot statement, why successive reuse needs accumulated-error accounting, and why a Choi output does not itself execute the product on arbitrary fresh data. Retain the pure-unitary-program and full-access assumptions when considering noise, mixed programs, many outputs or expected budgets. Translate the gate count into its dependence on a qubit operation's dimension. Audit the canonical-purification comparison: its inaccessible source environment, rank-one signal marginal, exact accepted target and fully counted mixed caps differ from these program promises. - **Dependencies and story role:** M03, M08, M13, M19, M23, M26–M28. Interprets the integrated result under its material assumptions and resource costs. -- **Conceptual connection:** The theorem establishes a capability for manipulating supplied quantum information. It does not establish an indefinite catalyst, a noisy-program theorem or a practical runtime advantage. -- **Sources:** [Watrous §§2.1.3][w213], [2.2.1][w221] and [3.2.1][w321] for joint states, channels and fidelity; [interpretation and limits](../science/interpretation.md), [converse boundaries](../science/converse.md#3-verification-and-boundaries), [charged resources](../science/implementation.md#5-charged-circuit-resources). +- **Conceptual connection:** The theorem establishes a capability for manipulating supplied quantum information. It does not establish an indefinite catalyst, a noisy-program theorem or a practical runtime advantage. Shared coherent contraction is a valid methodological relation to canonical purification, while a theorem transfer still requires matching the actual source and output hypotheses. +- **Sources:** [Watrous §§2.1.3][w213], [2.2.1][w221] and [3.2.1][w321] for joint states, channels and fidelity; [interpretation and limits](../science/interpretation.md), [converse boundaries](../science/converse.md#3-verification-and-boundaries), [charged resources](../science/implementation.md#5-charged-circuit-resources), [canonical-purification transfer boundary](../science/canonical_purification_boundary.md#why-the-composition-theorem-does-not-transfer). ### M30 — The integrated result, evidence, and inherited tools - **Central question:** How do the mechanism, obstruction, implementation and attribution support one coherent scientific claim? -- **Understanding target:** Trace the complete chain from Bell contraction and collective normalization through moments and joint fidelity, then from correlated priors and harmonic blocks to the universal converse, and finally to charged gates. Identify which finite checks test consequential identities, which arguments establish the all-dimensional theorem, and which tools are inherited from prior research. +- **Understanding target:** Trace the complete chain from Bell contraction and collective normalization through moments and joint fidelity, then from correlated priors and harmonic blocks to the universal converse, and finally to charged gates. Identify which finite checks test consequential identities, which arguments establish the all-dimensional theorem, and which tools are inherited from prior research. Explain why the exact canonical frontier has its own rigidity, harmonic LP and binomial source accounting rather than a dimension substitution in the Choi theorem. - **Dependencies and story role:** M14–M29, drawing on the complete background phase. Integrates the narrative into the project's final scientific interpretation. -- **Conceptual connection:** The result is an optimal fixed-error copy-allocation law for the stated program task, with a stronger constructive return guarantee and an explicit implementation. PGM, representation theory and QSVT retain their prior attribution; finite numerical agreement does not replace proof or determine priority. -- **Sources:** [Watrous §§2.2.2][w222], [3.1.1][w311], [3.1.2][w312] and [7.1.2][w712] as foundations; [canonical model](../science/model.md), [scientific account](../science/README.md), [verification scope](../verification/README.md), [operational comparisons and inherited ingredients](../references/RELATED_WORK.md). +- **Conceptual connection:** The result is an optimal fixed-error copy-allocation law for the stated program task, with a stronger constructive return guarantee and an explicit implementation. PGM, representation theory and QSVT retain their prior attribution; finite numerical agreement does not replace proof or determine priority. The sister relation identifies a shared physical normalization problem with different operational consequences and a clear authoritative home for each proof. +- **Sources:** [Watrous §§2.2.2][w222], [3.1.1][w311], [3.1.2][w312] and [7.1.2][w712] as foundations; [canonical model](../science/model.md), [scientific account](../science/README.md), [verification scope](../verification/README.md), [operational comparisons and inherited ingredients](../references/RELATED_WORK.md), [shared method and result ownership](../science/canonical_purification_boundary.md#ownership-and-reading-route). [watrous]: https://cs.uwaterloo.ca/~watrous/TQI/TQI.pdf [w111]: https://cs.uwaterloo.ca/~watrous/TQI/TQI.pdf#page=9 diff --git a/learning/README.md b/learning/README.md index 5ef1a80..95c5028 100644 --- a/learning/README.md +++ b/learning/README.md @@ -21,6 +21,7 @@ research-level dependencies have their own cited scientific sources. | Connect familiar quantum-information foundations to this task | [Watrous bridge](WATROUS_BRIDGE.md) | | Inspect exact quantifiers and proofs | [Scientific account](../science/README.md) | | Reproduce the small-system checks | [Verification guide](../verification/README.md) | +| Compare the method with exact state preparation | [Canonical-purification boundary](../science/canonical_purification_boundary.md) | The milestone map is defined by the scientific project and is shared by all readers. Its background phase builds the operational and mathematical @@ -43,3 +44,5 @@ questions. A good single-use return guarantee does not establish a repeated-reuse rate. These conventions and boundaries are carried through the learning route and stated precisely in the [model](../science/model.md) and [interpretation](../science/interpretation.md). + +The same Watrous foundations support the sister canonical-purification task, but its learning route has different understanding targets. The final milestones here use the [comparison](../science/canonical_purification_boundary.md) to audit a proposed transfer: check the accessible correlations, the exact or approximate output promise, every source charge, and the residual systems. Shared contraction machinery does not make the two resource laws interchangeable. diff --git a/learning/TUTORIAL.md b/learning/TUTORIAL.md index 9c7989e..dc42d29 100644 --- a/learning/TUTORIAL.md +++ b/learning/TUTORIAL.md @@ -301,3 +301,14 @@ The project's conclusion is the composition resource law under the stated copy access model. Return to the [milestone map](MILESTONES.md) to locate the individual conceptual dependencies, or to the [scientific account](../science/README.md) for the complete proofs. + + +## 8. Test the scope of a shared contraction method + +**Milestones M19, M29–M30.** A related exact task is to prepare the canonical purification of a known-vacuum rank-two family from state/conjugate samples. The [comparison page](../science/canonical_purification_boundary.md) supplies its source states and target. + +Both constructions begin with a linear map that has the desired coherent action. A physical success operator must also be a contraction on the whole input space. For the program composer, polar normalization produces a legal map and the Choi marginal moments control its distortion. Exact purification requires the accepted target direction to be preserved, so the sister proof first derives a residual-polynomial restriction from exactness and then optimizes the physical residual effect. + +This gives a useful way to assess a proposed transfer. Check what correlations are supplied, what systems remain inaccessible, whether successful output is exact, and whether every rejected source request counts. Unitary Choi halves are maximally mixed, but extracted signal specimens are rank one. The corresponding moment proof therefore does not transfer. With one conjugate mixed specimen, a vacuum outcome alone prevents exact success and imposes a ceiling of one-half, regardless of the other budget. + +The [canonical pure frontier](https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/exact_frontier.md) and [mixed-source frontier](https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/mixed_frontier.md) own those exact proofs. This repository's result remains the fixed-error unitary-program composition law. Shared methods explain the relationship without turning one theorem into a premise of the other. diff --git a/learning/WATROUS_BRIDGE.md b/learning/WATROUS_BRIDGE.md index 0910e68..43ee815 100644 --- a/learning/WATROUS_BRIDGE.md +++ b/learning/WATROUS_BRIDGE.md @@ -124,6 +124,14 @@ $\mathbb C^{d^2}$. Those are different ensembles. The local [converse](../science/converse.md) explains the correlated and independent priors before applying the discrimination bound. +## The same foundations in exact canonical purification + +Watrous §2.2.2 also describes the sister task's success branch as a Stinespring contraction. Section 7.1.1 supplies the language of repeated specimens, while §7.2.1 justifies Haar twirling when passing from arbitrary exact processors to uniform success. Here a symmetric factor is a whole two-register program; in the sister pure-input problem it is one signal specimen. The fixed basis defining a supplied conjugate is part of that other model. + +The [canonical-purification comparison](../science/canonical_purification_boundary.md) connects these foundations to the difference between polar and exact normalization. A contraction can be a physical success operator without preserving a prescribed target direction. The composition analysis allows distortion and bounds fidelity through its Choi marginal moments. The canonical task instead uses exact-output rigidity to restrict the residual polynomial space before normalizing a filter. + +Its pure n-by-one success optimum and finite all-allocation LP, followed by the exact binomial accounting for mixed hard caps, are developed in [the canonical project's scientific account](https://github.com/GoGoKo699/Canonical-Purification-Cost/tree/f2dc023a98242739929ef01eb4f9478d4bb96782/science). They are task-specific research bridges; they neither replace this repository's unitary theorem nor add a second educational anchor. + ## Research bridges supplied by the project The following are research dependencies, not additional educational anchors. diff --git a/llms.txt b/llms.txt index 0b248c8..d2a93bd 100644 --- a/llms.txt +++ b/llms.txt @@ -9,6 +9,7 @@ Physical instrument, PGM and global returned-program fidelity: science/construct All-dimensional mixed-moment comparison and universal converse: science/converse.md Projected QSVT, known symmetry phases and charged gate costs: science/implementation.md Meaning, precision alternatives and limitations: science/interpretation.md +Shared contraction method and exact canonical-purification transfer boundary: science/canonical_purification_boundary.md Closest work and inherited ingredients: references/RELATED_WORK.md Structured source metadata: references/bibliography.json Reproduction commands and tested limits: verification/README.md @@ -21,7 +22,8 @@ Story route: README.md -> learning/TUTORIAL.md -> science/interpretation.md. Study route: learning/README.md -> learning/MILESTONES.md, using learning/WATROUS_BRIDGE.md for targeted Watrous readings and notation conversions. Theorem route: science/model.md -> science/construction.md -> science/converse.md -> science/implementation.md. Reproduction route: verification/README.md -> the three documented Python scripts. +Sister route: science/canonical_purification_boundary.md -> Canonical-Purification-Cost/science/exact_frontier.md and science/mixed_frontier.md. The canonical repository owns those proofs; this project owns the unitary-Choi composition law. Educational anchor: John Watrous, The Theory of Quantum Information (2018), official author-hosted prepublication manuscript. Its Choi operator is unnormalized and its fidelity is unsquared. The project divides a unitary channel's Choi operator by d and uses squared pure-target overlap. The learning map is a common route through the scientific understanding targets. -Essential distinctions: Fidelity is pure-target overlap (squared convention). Inputs are complete pure unitary Choi copies; no unknown-unitary query or preparation oracle is supplied. The main theorem is a bounded-copy, constant-error resource law. The output is a program state, not application to fresh data. Global approximate program return does not establish indefinite reuse. Copy complexity and known-gate complexity are distinct. +Essential distinctions: Fidelity is pure-target overlap (squared convention). Inputs are complete pure unitary Choi copies; no unknown-unitary query or preparation oracle is supplied. The main theorem is a bounded-copy, constant-error resource law. The output is a program state, not application to fresh data. Global approximate program return does not establish indefinite reuse. Copy complexity and known-gate complexity are distinct. Exact canonical purification is a different task: rank-one signal marginals do not satisfy the maximally mixed Choi-moment premise, and polar normalization need not preserve exact accepted coherence. Its unequal-budget exact LP and mixed binomial frontier do not import this project's theorem; no canonical approximate-fidelity product law or general nm-of-order-d theorem is asserted. diff --git a/references/RELATED_WORK.md b/references/RELATED_WORK.md index b2d2856..0f12cc5 100644 --- a/references/RELATED_WORK.md +++ b/references/RELATED_WORK.md @@ -41,6 +41,14 @@ The public research repository [Quantum-Coherent-Collisions](https://github.com/ Those collective-copy ideas are shared. Matrix multiplication additionally sums coherently over a contracted index between two sources, and its Gram operator is a sum of noncommuting Bell projectors. The amplitude-power theorem does not directly implement this contraction or supply the mixed-unitary moment converse. The source is a scientific repository at a pinned revision, not a claim of an independently published article. +## Exact canonical purification and the normalization boundary + +[Canonical-Purification-Cost](https://github.com/GoGoKo699/Canonical-Purification-Cost) is a sister task rather than a specialization of the unitary-Choi theorem. Its [earlier model and balanced construction](https://github.com/GoGoKo699/Canonical-Purification-Cost/tree/823c42ab394e6764263aeb670a7d86292cec5341/science) prescribe the canonical purification of a flat rank-two state with a known vacuum, supplied through independent state/conjugate streams. Their inaccessible environments are not part of the input. + +The shared method is coherent contraction across separately symmetric copy pools, exact target/residual factorization before normalization, and a norm check on the whole physical input space. The [pure exact frontier](https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/exact_frontier.md) additionally uses exact-output polynomial rigidity and charged harmonic kernels to characterize every pure allocation by a finite LP. Its [mixed frontier](https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/mixed_frontier.md) is the lossless binomial average over known vacuum/signal occupancy counts. These proofs belong in that repository; they do not use the unitary-Choi composition theorem as a premise. + +The [local comparison](../science/canonical_purification_boundary.md) gives the task definitions, the one-conjugate exact optimum, the mixed-source success ceiling, and the transfer obstruction. Choi inner marginals are maximally mixed; canonical signals are rank one. The polar normalization here generally changes the accepted direction and supports an approximate joint-fidelity guarantee; exact purification requires the specified coherent direction on every accepted output. Neither a substitution of $d$ for $d^2$ nor the common symmetry notation supplies the missing proof. No exact canonical frontier is credited as a second unitary-program theorem, and no approximate-fidelity product law is inferred. + ## Representation theory used by the converse The [converse](../science/converse.md) compares correlated and independent Haar-unitary program ensembles. Its representation-theoretic foundation is inherited: diff --git a/references/bibliography.json b/references/bibliography.json index c5e42f1..6840565 100644 --- a/references/bibliography.json +++ b/references/bibliography.json @@ -335,6 +335,30 @@ "https://cs.uwaterloo.ca/~watrous/TQI/TQI.3.pdf", "https://cs.uwaterloo.ca/~watrous/TQI/TQI.7.pdf" ] + }, + { + "id": "canonical_purification_cost", + "type": "research_repository", + "title": "Canonical Purification Cost", + "repository_owner": "GoGoKo699", + "repository": "GoGoKo699/Canonical-Purification-Cost", + "baseline_revision": "823c42ab394e6764263aeb670a7d86292cec5341", + "frontier_revision": "f2dc023a98242739929ef01eb4f9478d4bb96782", + "url": "https://github.com/GoGoKo699/Canonical-Purification-Cost", + "inspected_locations": [ + "science/model.md", + "science/construction.md", + "science/harmonic_analysis.md", + "science/converse.md", + "science/exact_frontier.md", + "science/mixed_frontier.md" + ], + "role": "Sister exact canonical-purification task. Shared coherent contraction/full-space normalization; task-specific arbitrary-processor exact LP and fully counted mixed-source binomial frontier. The unitary-Choi theorem is not a premise.", + "primary_sources": [ + "https://github.com/GoGoKo699/Canonical-Purification-Cost/tree/823c42ab394e6764263aeb670a7d86292cec5341/science", + "https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/exact_frontier.md", + "https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/mixed_frontier.md" + ] } ] } diff --git a/science/README.md b/science/README.md index a1f1673..e36bb71 100644 --- a/science/README.md +++ b/science/README.md @@ -9,6 +9,7 @@ Read the account in this order: 3. [Converse](converse.md): correlated and independent Choi ensembles, mixed harmonic tensors, and the all-dimensional lower bound. 4. [Implementation](implementation.md): projected singular-value transformation, known symmetry phases, finite precision and charged gates. 5. [Interpretation and limits](interpretation.md): what program composition enables, how the copy advantage arises, and the boundaries of the result. +6. [Coherent contraction and exact canonical purification](canonical_purification_boundary.md): the sister project's exact pure and mixed frontier, the shared normalization problem, and the input/output assumptions that prevent a theorem transfer. The [related-work comparison](../references/RELATED_WORK.md) identifies inherited techniques and the distinct input models of earlier teleportation and program-processing tasks. The [verification guide](../verification/README.md) provides the commands and the tested scope of the finite checks. Those checks complement the analytical proofs; they do not replace them. diff --git a/science/canonical_purification_boundary.md b/science/canonical_purification_boundary.md new file mode 100644 index 0000000..96b3bb8 --- /dev/null +++ b/science/canonical_purification_boundary.md @@ -0,0 +1,194 @@ +# Coherent contraction and exact canonical purification + +[Scientific account](README.md) · [Composition construction](construction.md) · [Interpretation](interpretation.md) · [Canonical-purification sister project][canonical] + +The [Canonical-Purification-Cost][canonical] project and quantum program composition share a physical design question: a linear contraction has the desired action on unknown supplied states, but how can it be normalized into a legal quantum operation on the whole input space? They answer that question for different input orbits and different correctness requirements. The exact canonical-purification frontier belongs to the sister project; it is not a consequence of the unitary-Choi composition theorem. + +## Two tasks with different supplied correlations + +In this repository, a program is the normalized pure Choi vector of an arbitrary unknown unitary: + +```math +|U\rangle=\frac{\mathrm{vec}(U)}{\sqrt d}, +\qquad +\mathrm{Tr}_{1}|U\rangle\langle U| +=\mathrm{Tr}_{2}|U\rangle\langle U|=\frac{I}{d}. +``` + +Both registers of every supplied program are accessible. Contracting the inner halves of one program from each pool produces + +```math +(\langle\Phi_d|_{BC}\otimes I_{AD}) +\bigl(|U\rangle_{AB}|V\rangle_{CD}\bigr) +=\frac{|UV\rangle_{AD}}{d}. +``` + +The [model](model.md) specifies the register order and hard program budgets. The collective construction controls squared fidelity with the composed program and all returned specimens. Its successful state is generally approximate. + +For the sister task, fix a known signal space $`V\cong\mathbb C^d`$ orthogonal to a known vacuum. Its mixed sources and prescribed target are + +```math +\rho_\psi=\frac{|\mathrm{vac}\rangle\langle\mathrm{vac}| ++|\psi\rangle\langle\psi|}{2}, +\qquad +\rho_\psi^*=\frac{|\mathrm{vac}\rangle\langle\mathrm{vac}| ++|\bar\psi\rangle\langle\bar\psi|}{2}, +``` + +```math +|T_\psi\rangle +=\frac{|\mathrm{vac},\mathrm{vac}\rangle ++|\psi\rangle|\bar\psi\rangle}{\sqrt2}. +``` + +Conjugation is in the fixed supplied basis. The two sources are independent conditional on the unknown ray. Their inaccessible purification environments are not supplied. The target's vacuum/signal relative coherence is part of its definition; producing its two marginals separately does not prepare this target. + +| Operational feature | Unitary-program composition | Exact canonical purification | +|---|---|---| +| Supplied pure objects | Two-register unitary Choi states | State/conjugate vectors after extracting known signal sectors | +| Key marginal | Every inner Choi half is maximally mixed | A retained signal is rank one | +| Unknown parameters | Two arbitrary unitaries | One ray and its supplied conjugate | +| Useful output | Program for the ordered matrix product | Prescribed vacuum/signal coherent purification | +| Collective correctness | Approximate global squared fidelity | Exact target conditional on success | +| Remaining registers | Included in a global approximate return promise | May be discarded after exact target/residual factorization | +| Mixed-source environment | No mixed-source environment is needed | Inaccessible; not an available record or program | + +These distinctions are assumptions at the boundary of a proposed transfer, not changes of notation. + +## The shared physical normalization problem + +The program construction restricts a known Bell contraction to the separately symmetric space of whole programs. With the notation of [Construction](construction.md#2-symmetry-identifies-the-common-successful-state), its scaled map satisfies + +```math +Mv=\sqrt{\mu}\,w, +\qquad M^\dagger M=G, +\qquad \mu=\frac{mn}{d}. +``` + +This target-action identity alone does not make $`M`$ a success operator. The physical processor uses the polar partial isometry + +```math +K=MG^{-1/2}, +\qquad K^\dagger K=\Pi_{\mathrm{supp}G}\leq I. +``` + +The inverse vanishes on the kernel. It acts differently on different singular directions, so $`Kv`$ need not remain parallel to $`w`$. The exact moment calculation bounds the resulting distortion: + +```math +F_{\mathrm{global}}\geq\frac{mn}{mn+d^2-1}. +``` + +An exact mathematical definition of this channel specifies its operation before gate-synthesis error; it does not mean exact accepted output. The [success and fidelity calculation](construction.md#4-fidelity-and-success-probability) keeps these quantities separate. + +The canonical task instead combines a coherent scalar contraction and extraction of the signal pair. On the pure input space + +```math +R_{n,m}=\mathrm{Sym}^n(V)\otimes +\mathrm{Sym}^m(\overline V), +``` + +let $`a_i,b_i`$ be normalized symmetric-tensor annihilation maps and define + +```math +C=\sum_i a_i b_i, +\qquad +E_{n,m}=\frac1{\sqrt{nm}}\sum_{i,j}|i,j\rangle\otimes a_i b_j, +``` + +```math +W_{n,m} +=\frac1{\sqrt2} +\left( +|\mathrm{vac},\mathrm{vac}\rangle\otimes\frac{C}{\sqrt{nm}} ++E_{n,m} +\right). +``` + +For positive $`n,m`$, $`E_{n,m}`$ is an isometry and + +```math +W_{n,m}\bigl(\psi^{\otimes n}\bar\psi^{\otimes m}\bigr) +=T_\psi\otimes +\psi^{\otimes(n-1)}\bar\psi^{\otimes(m-1)}. +``` + +Again, this exact action on the promised orbit is not a contractivity proof. The [canonical exact-frontier account][exact] characterizes all admissible residual positive operators $`X`$ through + +```math +W_{n,m}^\dagger(I_{\mathrm{out}}\otimes X)W_{n,m}\leq I, +\qquad X\geq0. +``` + +This condition is imposed on the entire symmetric input space and extended by zero on its orthogonal complement. Every feasible operator gives a completely positive trace-nonincreasing success map, with a complementary failure effect. Normalized occupation notation represents ordinary finite-dimensional symmetric tensors; it does not supply free bosonic hardware. + +For one conjugate pure specimen, the residual space is $`\mathrm{Sym}^{n-1}(V)`$ and + +```math +CC^\dagger=(d+n-1)I, +\qquad +\|W_{n,1}\|^2=\frac{d+2n-1}{2n}. +``` + +Scaling by this norm gives success $`2n/(d+2n-1)`$. The sister project's exact-output rigidity argument proves that this is optimal over arbitrary processors, rather than only over this displayed filter. + +The shared method is therefore useful at the level of coherent contraction, separately symmetric copy pools, target/residual factorization and whole-space physical normalization. The suitable normalization and its performance analysis must be derived for the actual target and source orbit. + +## What the exact frontier adds + +The canonical result starts from arbitrary accepting Kraus operators. Exact output forces their vacuum and signal amplitudes to obey a polynomial identity. After homogenization, the vacuum amplitude is divisible by $`\sum_i z_iw_i`$, leaving residual bidegree $`(n-1,m-1)`$. This is an exact-output restriction absent from an approximate fidelity promise. + +The [authoritative pure-frontier proof][exact] derives the charged harmonic weights and an attained linear program with $`\min(n,m)`$ variables. It covers every pure-copy allocation and arbitrary ancillas, multiple accepted outcomes and discarded systems. In particular, + +```math +q_d^*(n,1)=\frac{2n}{d+2n-1}, +\qquad +q_d^*(n,0)=q_d^*(0,m)=0 +\quad(d\geq2). +``` + +The full finite LP, rather than a guessed product formula, is the exact all-allocation statement. For example, $`q_2^*(2,2)=12/13`$; the earlier top-residual projection filter attains $`1/2`$ for that input. Optimizing the whole admissible residual operator can matter substantially even in a small system. + +For hard mixed-source caps $`A,B`$, the [mixed-frontier proof][mixed] shows that measuring the known vacuum/signal sectors loses no information. Their signal counts are independent, + +```math +K\sim\mathrm{Bin}(A,1/2), +\qquad +L\sim\mathrm{Bin}(B,1/2), +``` + +and the exact optimum is + +```math +Q_d^*(A,B)=\mathbb E\,q_d^*(K,L). +``` + +Every requested specimen counts, including vacuum outcomes, surplus signals and rejected filters. Bounded adaptive source order is included by precollection under these separate caps. An additional stricter total cap requires its own analysis. + +The zero-count sectors imply + +```math +Q_d^*(A,B)\leq(1-2^{-A})(1-2^{-B}). +``` + +Thus a single conjugate mixed specimen limits success to $`1/2`$, however large the ordinary supply becomes. A product of supplied budgets alone cannot describe this full fixed-success frontier. The interpretation that $`nm`$ of order $`d`$ sets a general canonical-purification scale remains conjectural here; no fixed-approximate-fidelity product law is asserted. + +## Why the composition theorem does not transfer + +The [composition moment proof](construction.md#3-exact-first-and-second-moments) replaces expectations of functions of $`G`$ by normalized traces because the complete inner marginal is $`I/d^{m+n}`$. Rank-one signal specimens do not satisfy that hypothesis. The composition converse also compares two Haar-unitary priors, not a state/conjugate orbit. Substituting $`d`$ for $`d^2`$ in its result would supply neither the missing moment identity nor the required converse. + +Likewise, applying a polar normalization to $`W_{n,m}`$ is not an exact-purification protocol merely because it produces a contraction. A direction-dependent rescaling can change the vacuum/signal relative amplitude and the exact accepted state. The residual-operator constraint and polynomial rigidity handle this exactness requirement directly. + +The independent source model supplies neither matched purification gauges nor a coherent record identifying how a mixed specimen was prepared. A proof that introduces environments to represent a completely positive map does not make the source's inaccessible environment physically available. + +Finally, whole-copy symmetry and known register permutations can be used in either setting, but the [program implementation](implementation.md) does not automatically implement an optimal canonical residual filter or prove its gate cost. The canonical frontier is a copy-information result; its efficient gate synthesis is a separate question. + +## Ownership and reading route + +This repository owns the unitary-program construction, its arbitrary-processor composition converse, the fixed-error $`mn=\Theta(d^2)`$ law and its charged implementation. The sister project owns the prescribed-purification model and the [exact pure][exact] and [mixed-source][mixed] frontier proofs. Its inherited balanced total-copy approximate theorem retains its original hypotheses and is distinct from the new exact unequal-budget characterization. + +For the common foundations, use the [Watrous bridge](../learning/WATROUS_BRIDGE.md): completely positive maps and Stinespring contractions in §2.2.2, symmetric tensor spaces in §7.1.1, and Haar averaging in §7.2.1. General positive-kernel stochastic conversion is inherited theory, discussed and attributed in the canonical proof; the task-specific polynomial support and coefficients are derived there. + +[canonical]: https://github.com/GoGoKo699/Canonical-Purification-Cost +[exact]: https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/exact_frontier.md +[mixed]: https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/mixed_frontier.md + diff --git a/science/interpretation.md b/science/interpretation.md index cc2f8a2..fb2255e 100644 --- a/science/interpretation.md +++ b/science/interpretation.md @@ -40,6 +40,12 @@ $$ Assigning an arbitrary output on failure therefore gives infidelity at most $\epsilon$ with $k=\lceil d^2\log(1/\epsilon)\rceil$ pairs. This alternative does not return all but one specimen from each input pool. For extremely small error it can use fewer copies than the sufficient balanced collective allocation $O(d/\sqrt\epsilon)$. A precision-sensitive comparison must retain both alternatives. +## Exact purification as a related contraction task + +The sister project [Canonical-Purification-Cost](https://github.com/GoGoKo699/Canonical-Purification-Cost) must prepare a specified vacuum/signal coherent purification exactly on success. Both projects use contraction on separately symmetric copy pools and require physical normalization on the whole input space. Their normalizations answer different correctness requirements: the polar map here permits controlled distortion, whereas exact purification must retain the prescribed coherence. + +The [self-contained comparison](canonical_purification_boundary.md) explains why maximally mixed Choi marginals and the unitary-prior converse do not apply to independent rank-two state/conjugate samples. Its exact pure frontier is a finite all-allocation LP; mixed hard caps require averaging over the actual binomial signal counts. Those results have their authoritative proof in the sister repository. They do not follow by replacing $d^2$ with $d$ in this project's theorem. + ## What the present theorem leaves open The supported result fixes the order of the copy-product threshold and provides an implementable universal construction. Exact optimal fidelity, sharp small-error rates, improved gate counts, robustness to mixed or noisy programs, expected-budget protocols and repeated composition are separate questions. They are not needed to justify the stated constant-error law, and no result about them is assumed in its proof. From 90f10186a99446b2a1f6a979c4fd5b1b712e3b77 Mon Sep 17 00:00:00 2001 From: Ruge Lin Date: Sun, 8 Nov 2026 16:20:56 +0800 Subject: [PATCH 2/2] Repair inherited TeX commands and pin the checked canonical frontier --- learning/TUTORIAL.md | 2 +- learning/WATROUS_BRIDGE.md | 2 +- references/RELATED_WORK.md | 2 +- references/bibliography.json | 6 +++--- science/canonical_purification_boundary.md | 4 ++-- science/construction.md | 6 +++--- science/implementation.md | 4 ++-- 7 files changed, 13 insertions(+), 13 deletions(-) diff --git a/learning/TUTORIAL.md b/learning/TUTORIAL.md index dc42d29..dab38da 100644 --- a/learning/TUTORIAL.md +++ b/learning/TUTORIAL.md @@ -311,4 +311,4 @@ Both constructions begin with a linear map that has the desired coherent action. This gives a useful way to assess a proposed transfer. Check what correlations are supplied, what systems remain inaccessible, whether successful output is exact, and whether every rejected source request counts. Unitary Choi halves are maximally mixed, but extracted signal specimens are rank one. The corresponding moment proof therefore does not transfer. With one conjugate mixed specimen, a vacuum outcome alone prevents exact success and imposes a ceiling of one-half, regardless of the other budget. -The [canonical pure frontier](https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/exact_frontier.md) and [mixed-source frontier](https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/mixed_frontier.md) own those exact proofs. This repository's result remains the fixed-error unitary-program composition law. Shared methods explain the relationship without turning one theorem into a premise of the other. +The [canonical pure frontier](https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/c0e1cac831e7bc6073042974c5f440bd847b37a0/science/exact_frontier.md) and [mixed-source frontier](https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/c0e1cac831e7bc6073042974c5f440bd847b37a0/science/mixed_frontier.md) own those exact proofs. This repository's result remains the fixed-error unitary-program composition law. Shared methods explain the relationship without turning one theorem into a premise of the other. diff --git a/learning/WATROUS_BRIDGE.md b/learning/WATROUS_BRIDGE.md index 43ee815..d5fc907 100644 --- a/learning/WATROUS_BRIDGE.md +++ b/learning/WATROUS_BRIDGE.md @@ -130,7 +130,7 @@ Watrous §2.2.2 also describes the sister task's success branch as a Stinespring The [canonical-purification comparison](../science/canonical_purification_boundary.md) connects these foundations to the difference between polar and exact normalization. A contraction can be a physical success operator without preserving a prescribed target direction. The composition analysis allows distortion and bounds fidelity through its Choi marginal moments. The canonical task instead uses exact-output rigidity to restrict the residual polynomial space before normalizing a filter. -Its pure n-by-one success optimum and finite all-allocation LP, followed by the exact binomial accounting for mixed hard caps, are developed in [the canonical project's scientific account](https://github.com/GoGoKo699/Canonical-Purification-Cost/tree/f2dc023a98242739929ef01eb4f9478d4bb96782/science). They are task-specific research bridges; they neither replace this repository's unitary theorem nor add a second educational anchor. +Its pure n-by-one success optimum and finite all-allocation LP, followed by the exact binomial accounting for mixed hard caps, are developed in [the canonical project's scientific account](https://github.com/GoGoKo699/Canonical-Purification-Cost/tree/c0e1cac831e7bc6073042974c5f440bd847b37a0/science). They are task-specific research bridges; they neither replace this repository's unitary theorem nor add a second educational anchor. ## Research bridges supplied by the project diff --git a/references/RELATED_WORK.md b/references/RELATED_WORK.md index 0f12cc5..3b6362a 100644 --- a/references/RELATED_WORK.md +++ b/references/RELATED_WORK.md @@ -45,7 +45,7 @@ Those collective-copy ideas are shared. Matrix multiplication additionally sums [Canonical-Purification-Cost](https://github.com/GoGoKo699/Canonical-Purification-Cost) is a sister task rather than a specialization of the unitary-Choi theorem. Its [earlier model and balanced construction](https://github.com/GoGoKo699/Canonical-Purification-Cost/tree/823c42ab394e6764263aeb670a7d86292cec5341/science) prescribe the canonical purification of a flat rank-two state with a known vacuum, supplied through independent state/conjugate streams. Their inaccessible environments are not part of the input. -The shared method is coherent contraction across separately symmetric copy pools, exact target/residual factorization before normalization, and a norm check on the whole physical input space. The [pure exact frontier](https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/exact_frontier.md) additionally uses exact-output polynomial rigidity and charged harmonic kernels to characterize every pure allocation by a finite LP. Its [mixed frontier](https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/mixed_frontier.md) is the lossless binomial average over known vacuum/signal occupancy counts. These proofs belong in that repository; they do not use the unitary-Choi composition theorem as a premise. +The shared method is coherent contraction across separately symmetric copy pools, exact target/residual factorization before normalization, and a norm check on the whole physical input space. The [pure exact frontier](https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/c0e1cac831e7bc6073042974c5f440bd847b37a0/science/exact_frontier.md) additionally uses exact-output polynomial rigidity and charged harmonic kernels to characterize every pure allocation by a finite LP. Its [mixed frontier](https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/c0e1cac831e7bc6073042974c5f440bd847b37a0/science/mixed_frontier.md) is the lossless binomial average over known vacuum/signal occupancy counts. These proofs belong in that repository; they do not use the unitary-Choi composition theorem as a premise. The [local comparison](../science/canonical_purification_boundary.md) gives the task definitions, the one-conjugate exact optimum, the mixed-source success ceiling, and the transfer obstruction. Choi inner marginals are maximally mixed; canonical signals are rank one. The polar normalization here generally changes the accepted direction and supports an approximate joint-fidelity guarantee; exact purification requires the specified coherent direction on every accepted output. Neither a substitution of $d$ for $d^2$ nor the common symmetry notation supplies the missing proof. No exact canonical frontier is credited as a second unitary-program theorem, and no approximate-fidelity product law is inferred. diff --git a/references/bibliography.json b/references/bibliography.json index 6840565..122ea0e 100644 --- a/references/bibliography.json +++ b/references/bibliography.json @@ -343,7 +343,7 @@ "repository_owner": "GoGoKo699", "repository": "GoGoKo699/Canonical-Purification-Cost", "baseline_revision": "823c42ab394e6764263aeb670a7d86292cec5341", - "frontier_revision": "f2dc023a98242739929ef01eb4f9478d4bb96782", + "frontier_revision": "c0e1cac831e7bc6073042974c5f440bd847b37a0", "url": "https://github.com/GoGoKo699/Canonical-Purification-Cost", "inspected_locations": [ "science/model.md", @@ -356,8 +356,8 @@ "role": "Sister exact canonical-purification task. Shared coherent contraction/full-space normalization; task-specific arbitrary-processor exact LP and fully counted mixed-source binomial frontier. The unitary-Choi theorem is not a premise.", "primary_sources": [ "https://github.com/GoGoKo699/Canonical-Purification-Cost/tree/823c42ab394e6764263aeb670a7d86292cec5341/science", - "https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/exact_frontier.md", - "https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/mixed_frontier.md" + "https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/c0e1cac831e7bc6073042974c5f440bd847b37a0/science/exact_frontier.md", + "https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/c0e1cac831e7bc6073042974c5f440bd847b37a0/science/mixed_frontier.md" ] } ] diff --git a/science/canonical_purification_boundary.md b/science/canonical_purification_boundary.md index 96b3bb8..a1b9c5e 100644 --- a/science/canonical_purification_boundary.md +++ b/science/canonical_purification_boundary.md @@ -189,6 +189,6 @@ This repository owns the unitary-program construction, its arbitrary-processor c For the common foundations, use the [Watrous bridge](../learning/WATROUS_BRIDGE.md): completely positive maps and Stinespring contractions in §2.2.2, symmetric tensor spaces in §7.1.1, and Haar averaging in §7.2.1. General positive-kernel stochastic conversion is inherited theory, discussed and attributed in the canonical proof; the task-specific polynomial support and coefficients are derived there. [canonical]: https://github.com/GoGoKo699/Canonical-Purification-Cost -[exact]: https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/exact_frontier.md -[mixed]: https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/f2dc023a98242739929ef01eb4f9478d4bb96782/science/mixed_frontier.md +[exact]: https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/c0e1cac831e7bc6073042974c5f440bd847b37a0/science/exact_frontier.md +[mixed]: https://github.com/GoGoKo699/Canonical-Purification-Cost/blob/c0e1cac831e7bc6073042974c5f440bd847b37a0/science/mixed_frontier.md diff --git a/science/construction.md b/science/construction.md index c5d76b9..4b3c134 100644 --- a/science/construction.md +++ b/science/construction.md @@ -156,13 +156,13 @@ Consequently, **Equation (10).** -$$ +```math Lv=\frac1d w, \qquad -Mv=\sqrt\muw, +Mv=\sqrt{\mu}\,w, \qquad \mu=\frac{mn}{d}. -$$ +``` Let $K=MG_{\mathcal S}^{-1/2}$ be the polar partial isometry, zero on the kernel. It is a contraction, with $K^{\dagger}K=\Pi_{\mathrm{supp}G_{\mathcal S}}$. Since $G^{-1/2}$ preserves $\mathcal S$, every routed physical branch satisfies diff --git a/science/implementation.md b/science/implementation.md index daa2c64..281009a 100644 --- a/science/implementation.md +++ b/science/implementation.md @@ -83,7 +83,7 @@ M^\dagger M=G_{\mathcal S}, A^\dagger A=\frac{G_{\mathcal S}}{dmn}. $$ -Equation (2) also proves $\VertA\Vert\leq1$ and $\VertM\Vert\leq\sqrt{dmn}$. The relevant singular spectrum is that of the **restricted block** $QP$. The unrestricted Bell projector has only zero and unit singular values and cannot by itself implement this polar transformation. +Equation (2) also proves $\Vert A\Vert\leq1$ and $\Vert M\Vert\leq\sqrt{dmn}$. The relevant singular spectrum is that of the **restricted block** $QP$. The unrestricted Bell projector has only zero and unit singular values and cannot by itself implement this polar transformation. ## 2. Approximating the polar map on the supplied states @@ -220,7 +220,7 @@ $$ V^\dagger e^{i\phi(2\Pi_0-I)}V $$ -implements $e^{i\phi(2P-I)}$ on clean ancillas. This follows by expanding the exponential into $\cos\phiI+i\sin\phi(2\Pi_0-I)$ and using (9). Adding the single QSVT control qubit changes the gate count by a constant factor. The Bell-subspace phases are implemented by preparing and unpreparing the known $|\Phi_d\rangle$ and phasing the all-zero state. +implements $e^{i\phi(2P-I)}$ on clean ancillas. This follows by expanding the exponential into $`\cos\phi\,I+i\sin\phi(2\Pi_0-I)`$ and using (9). Adding the single QSVT control qubit changes the gate count by a constant factor. The Bell-subspace phases are implemented by preparing and unpreparing the known $|\Phi_d\rangle$ and phasing the all-zero state. ## 5. Charged circuit resources