Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

The State Equation: Causal Cone Rank and Minimal Physical Memory

Максим Чурилов

Published Nov 3, 2026• 0 citations
Publisher

Abstract

We ask which finite classical or quantum memory is sufficient to generate the predictions of a stochastic process. Starting from its Hankel kernel \(H_p(u,v)=p(uv)\), we define causal cone rank by globally shift-compatible cone lifts of the canonical predictive realization. Within a specified realization class, the physical carrier minimum equals the corresponding minimum compatible lift dimension. The framework separates unconstrained Hankel rank, classical nonnegative realizations, and quantum completely positive instruments, and extends to action--outcome processes with interventionwise normalization. A declared linear coarse state has exactly \(\dim\ker(C|_{\operatorname{im} H_p})\) missing predictive coordinates. Finite temporal cuts give nonnegative- and positive-semidefinite-rank witnesses; an explicit two-time family exhibits a growing exact classical--quantum dimension gap, while its fixed-tolerance classical cost remains bounded. For an exact finite-dimensional time-homogeneous Markovian dilation, reset-conditioned closure rank obeys \(r_{\rm cl}\le d_S^2(d_E^2-1)\), with a saturating qubit--qubit example. We give nonasymptotic singular-value tests, while emphasizing that numerical rank and unconstrained truncated singular expansions do not themselves certify a physical realization.

DOI
10.5281/zenodo.23918696

Related papers

Intervention-Relative Predictive Memory

Alexandre Dumas

This preprint develops a framework for intervention-relative predictive memory: the minimum information that must be retained about the past in order to predict every future experiment allowed by a declared intervention set.For finite-dimensional linear processes, predictive memory is characterized exactly by a reachable-observable quotient and the rank of the corresponding past-future Hankel response matrix. The paper derives finite certification procedures, monotonicity under richer control, a decomposition of memory growth into reachability expansion and visibility activation, symmetry-protected compression, and no-compression results under sufficiently rich intervention algebras.The framework is connected to process-tensor bond rank, hidden-memory realization, and physical memory costs. At a prepare-and-measure temporal cut, linear, classical, and quantum memory are distinguished through ordinary rank, nonnegative rank, and positive-semidefinite rank. A robust quantum-dimension witness is derived for an n-message decoding task.A proposed single-photon spatial-mode experiment provides a concrete test. Phase-coded histories have identical present path-intensity distributions but can be perfectly distinguished by permitted future phase and Fourier operations, giving an ideal response matrix equal to the identity. For 17 modes, sufficiently high decoding success directly certifies a 17-dimensional quantum memory.The paper distinguishes established realization, automata, process-tensor, quantum-control, and factorization results from the proposed synthesis and experimental interpretation. It concludes with two natural next directions: experimental implementation and a multitime completely-positive / PSD-rank theory of intervention-relative quantum memory.

•Sep 29, 2026•0 citations

Robust State Equation: Calibrated Hankel Recovery and Finite-Precision Classicalization

Максим Чурилов

Exact physical memory is an unstable target: a process of Hankel rank one can be arbitrarily close to processes requiring larger classical and quantum carriers. We make this obstruction quantitative and then identify the extra structure under which a physical finite-sample upper theorem is possible. First, on a stationary binary Markov family whose mixing gap is uniformly at least one half, every estimator of exact linear rank, minimum classical memory, or minimum quantum Hilbert dimension has worst-case error at least 3/8 at local separation \(\Theta([N(L-1)]^{-1/2})\) from the rank-one model. The smallest nonzero singular value of an explicit Hankel core is proportional to the same parameter. Thus a spectral or dynamical margin is necessary even under rapid mixing. Second, finite Hankel data determine a minimum linear realization only up to similarity, whereas nonnegativity and complete positivity are not invariant under a generic similarity. We give an exact two-state counterexample and then formulate the missing assumption as a calibrated physical core. For an invertible core \(H\), all shifts are recovered in one basis by \(\tilde{A}_a = \tilde{H}^{-1}\tilde{H}_a\). An explicit perturbation theorem, followed by Choi positive-part projection and one common trace-preserving correction, returns a genuine quantum instrument. A direct-query finite-sample corollary controls its error uniformly over adaptive policies and exposes every dependence on core conditioning, physical calibration, horizon, and confidence. The resulting linear-in-horizon coefficient is first-order sharp: a one-state classical process attains the exact discrepancy \(1-(1-\Delta/2)^L\) at one-step diamond radius \(\Delta\). Third, we prove a compact-state classicalization theorem. Any finite-dimensional physical memory described by a compact base-norm state space has, at horizon \(L\) and operational error \(\eta>0\), a classical realization whose number of states is the seeded covering number at radius \(2\eta/L\), uniformly over adaptive policies. For a \(d\)-dimensional quantum instrument this gives, for \(d\geq 2\), \[ \left(1+\frac{2(1-d^{-1})L}{\eta}\right)^{d^2-1} \] states, with no extra transient state; one state suffices for \(d=1\). The ambient exponent \(d^2-1\) is optimal for universal trace-norm discretization, while a compact branch-invariant reachable set replaces it by its own metric entropy. Consequently, an unbounded exact quantum/classical memory separation cannot remain unbounded at fixed \(d\), \(L\), \(\eta\). Robust quantum advantage must instead be quantified as a joint precision–horizon tradeoff. Deterministic verification code audits the mixing lower bound, the similarity-gauge counterexample, common-shift recovery, physical repair, and finite-net simulation.

•Nov 3, 2026•0 citations

Quantum Source Completions and Memory Environment Tradeoffs

Jeremy Rodgers

How much internal quantum memory is needed to reproduce a stationary stochastic process, and how does reducing that memory change the environmental resources required when output records are retained? This paper establishes exact quantum-source constructions and resource bounds under explicitly specified readout and retained-output conditions. A general completion theorem identifies the minimum Hilbert-space dimension for stationary triple data in the orthogonal-current-symbol source class: it equals the sum of the complex positive-semidefinite ranks of the middle-symbol slices. An attaining construction reproduces the entire canonical stationary order-two process. For the delayed trine process, the minimum classical hidden Markov model requires nine states at every non-independent parameter value. Quantum instruments reproduce the same complete output law using four-dimensional memory throughout the parameter range, with four dimensions proved necessary in a certified low-noise regime and three dimensions both sufficient and necessary on an explicit high-noise interval. The paper then separates reusable memory dimension from environmental dilation cost. On the certified high-noise interval, the exact one-step Pareto frontier consists of a three-dimensional source with recorded-channel Choi rank nine and a four-dimensional source with rank three. Here, Choi rank measures the minimum initially pure environmental remainder dimension for the channel retaining both the emitted classical symbol and final quantum memory. A separate mixed-support qutrit construction at noise parameter 0.99 improves complete-history costs and gives upper bounds close to universal lower bounds. These results show why optimizing a single step need not optimize a retained multi-step history. The analysis also distinguishes complete channels, fixed-input states and archive-only outputs, and demonstrates how access to a retained reference can distinguish generators with identical passive output statistics. Full-word proofs, algebraic complete-positivity certificates, a technical companion and reproducibility materials accompany the results. The exact intermediate memory boundary and complete-history optimum remain open. All results use standard quantum probabilities and apply to the stated operational resource models; they do not establish total hardware, energy or thermodynamic minima.

•Oct 28, 2026•0 citations

Finite Realizations and Effective Memory in Nonlinear Quantum Filtering

Jacob Emerson

Nonlinear dynamics generate non-Gaussian states and operations, but they also make continuously monitored quantum systems difficult to track. A quantum filter performs this task by compressing a noisy measurement record into a set of evolving variables that predict future observables. For Gaussian dynamics, this compression is exact and finite. In nonlinear systems, conventional moment equations generally expand without bound, but the failure of one representation does not prove that every finite description is impossible. Here we formulate quantum filtering as a realization problem for general quantum input--output maps. We introduce the linearized observable Hankel operator $K_O$, which maps changes in the past measurement record to changes in a future conditional observable. Any finite-dimensional smooth filter forces $K_O$ to have finite rank, so infinite rank provides a coordinate-independent proof that no exact finite-dimensional filter exists. Applied to single-mode polynomial bosonic systems under continuous quadrature monitoring, this yields a sharp boundary: Gaussian dynamics and the measurement-aligned Conditional Momentum Moment class admit finite realizations, whereas dynamics outside these classes produce infinitely many response directions and admit no finite-dimensional smooth exact filter. Exact impossibility, however, need not imply large practical complexity. The singular values of $K_O$ quantify how many response directions matter at a chosen local accuracy, defining the effective memory of an observable. For the Kerr and Duffing oscillators, these singular values decay rapidly despite infinite exact rank, showing that infinite exact memory can coexist with low finite-accuracy complexity. The framework therefore separates exact realizability from effective complexity in nonlinear quantum filtering.

quant-pharXiv:2609.09663•Sep 10, 2026•1 citation

Predictive Holonomy and Minimal Completion of Coarse Dynamics

Maxim V. Churilov

A coarse description of a dynamical system generally fails to evolve autonomously: evolving and coarse-graining in different orders gives different predictions, and the compressed propagator fails the semigroup law. We show that this path dependence is the observable boundary of missing state. For a finite continuous-time Markov chain with generator $L$, coarse channel $C$, and stochastic section $R$ satisfying $CR=I$, define $P=RC$, $Q=I-P$, and $K=CLR$. The exact coarse propagator $T_t=C\mathrm{e}^{tL}R$ obeys a Volterra equation with memory kernel $\mathcal{M}(t)=CLQ\,\mathrm{e}^{tQLQ}\,QLR$. Every memory jet $\mathcal{M}^{(k)}(0)$ is recursively recoverable from the short-time derivatives of $T_t$ alone. The rank of the block Hankel operator $\mathbb{H}_{\infty}=(\mathcal{M}^{(i+j)}(0))_{i,j\geq 0}$ is exactly the smallest number of auxiliary linear state variables required to turn the coarse dynamics into an exact autonomous, time-homogeneous system. A minimal completion is constructible by Ho--Kalman factorization and is unique up to similarity. If $d=\min\{k:\mathcal{M}^{(k)}(0)\neq 0\}$, then every smooth classical $f$-divergence generically scales as $t^{2d+4}$. Exact Markov examples realize exponents $4,6,8$ and closure ranks $1,2,3$. We also derive finite-sample singular-value certificates for hidden predictive coordinates, including a matrix-Bernstein bound with intrinsic-variance control. Finally, histories are identified when they induce the same future law under every admissible intervention, yielding the minimal counterfactual predictive state. The results turn coarse-graining into a constructive state-completion problem and separate ordinary linear realization from the harder requirement of a positive or Markov realization.

quant-pharXiv:2610.13951•Aug 10, 2026

Task-Sufficient Memory under Stochastic Clocks: Coarse-graining operators, fixed points, and finite-policy certification

Marcos Eduardo — Elias

What information about environmental memory must a quantum controller retain to guarantee a specified computational output? This preprint formulates that question in terms of task-sufficient process representations, intervention-valid predictions, stochastic clock laws, and explicit resource accounting. The central requirement is to preserve the outcomes and costs of the permitted decisions, with a stated accuracy contract and a finite-sample certificate. The paper connects two complementary lines of analysis. An operational construction studies Gaussian dephasing and controlled qubit tasks. A separate stochastic-clock analysis studies direct and inverse time changes, fractional relaxation, and the information that temporal coarse graining must transport. These constructions address related modelling questions but are not identified as a single microscopic noise model. The main elements are: Identical uncontrolled channels, different controlled responses. A continuum of Gaussian phase processes has the same uncontrolled coherence at every time and zero Breuer-Laine-Piilo backflow, yet different spin-echo responses. The construction shows why an uncontrolled channel family alone can fail to determine an intervention-dependent control decision. Task-sufficient equivalence and resource contracts. The manuscript defines a contract functional collecting accepted-output moments, acceptance probabilities, and physical costs over a declared policy family. Its finite linear task quotient identifies distinctions relevant to those predictions. Approximate sufficiency is accompanied by stability bounds that expose the sensitivity of conditional guarantees to low acceptance probabilities and small feasibility margins. Simultaneous finite-policy certification. Conservative confidence bounds certify conditional task bias and expected cost per accepted outcome after selection among a fixed, finite set of candidate policies. The assumptions include independent reset attempts, bounded observations and costs, and a trusted ideal reference. Failure to certify a candidate is explicitly distinguished from proof that no feasible physical policy exists. An explicit coarse-graining framework for stochastic clocks. The proposed decorated phase space records the process, permitted policy family, observable set, and clock law. The clock component transports stable laws under their natural scaling and relates domains of attraction to regular variation of waiting-time tails. Within the stated two-parameter clock-and-coupling family, the relaxation-rate direction scales with eigenvalue c^alpha. These clock results are placed in the established probabilistic renormalization literature; they do not establish hyperbolicity or convergence for the full decorated operator. A reset-sensitive fractional-relaxation calculation. Direct subordination of a quantum dynamical semigroup is distinguished from inverse-stable time change. For the latter, the paper compares uninterrupted evolution with independent renewal of the hidden clock and relevant environment. At half order and dimensionless rate one, the exact erfcx representation gives the short-time defect Delta(h) ~ 0.660989212585294 sqrt(h). The difference concerns the declared reset protocols, rather than a general assertion that every physical gate resets its environment. The reported classical benchmark uses ten candidate policies with explicit pulse depolarisation and separate X- and Z-expectation contracts. At the quasistatic endpoint, midpoint echo increases average identity-gate fidelity from 0.789293 to 0.990050. The fidelity crossover near r = 0.0467883 is distinct from the stricter X-task feasibility threshold r >= 0.9688074. For the selected Z-task input, dephasing is harmless and unnecessary echo pulses reduce the expectation. Conventional calibrated echo matches the best tested control performance; calibration, sampling, rejected attempts, and amortisation remain part of the resource account. Two questions are isolated as conjectures: boundedness of the task-quotient dimension along the coarse-graining orbit, and rigidity of the task quotient as a complete contract invariant. The manuscript specifies falsifiable numerical and experimental tests, including rescaled relaxation-profile collapse and finite tests of quotient-rank growth. It establishes no hardware advantage, scalable task-compression theorem, logical error-rate improvement, or error-correction overhead reduction. This illustrated version 2.0 contains 30 pages, four scientific figures, derivation and reproduction appendices, a methodological comparison with renormalization in one-dimensional dynamics, a glossary, and 58 references. The manuscript is dated 12 November 2026 and supersedes version 1.0 of 10 November 2026. The four figures show the uncontrolled/control-response contrast, task-dependent protection, finite-policy certification alongside dimensionless fractional relaxation, and the exact scalar clock-reset defect.

•Nov 13, 2026•0 citations
We ask which finite classical or quantum memory is sufficient to generate the predictions of a stochastic process. Starting from its Hankel kernel (H_p(u,v)=p(uv)), we define causal cone rank by globally shift-compatible cone lifts of the canonical predictive realization. Within a specified realization class, the physical carrier minimum equals the corresponding minimum compatible lift dimension. The framework separates unconstrained Hankel rank, classical nonnegative realizations, and quantum completely positive instruments, and extends to action--outcome processes with interventionwise normalization. A declared linear coarse state has exactly (\dim\ker(C|{\operatorname{im} H_p})) missing predictive coordinates. Finite temporal cuts give nonnegative- and positive-semidefinite-rank witnesses; an explicit two-time family exhibits a growing exact classical--quantum dimension gap, while its fixed-tolerance classical cost remains bounded. For an exact finite-dimensional time-homogeneous Markovian dilation, reset-conditioned closure rank obeys (r{\rm cl}\le d_S^2(d_E^2-1)), with a saturating qubit--qubit example. We give nonasymptotic singular-value tests, while emphasizing that numerical rank and unconstrained truncated singular expansions do not themselves certify a physical realization.
This preprint develops a framework for intervention-relative predictive memory: the minimum information that must be retained about the past in order to predict every future experiment allowed by a declared intervention set.For finite-dimensional linear processes, predictive memory is characterized exactly by a reachable-observable quotient and the rank of the corresponding past-future Hankel response matrix. The paper derives finite certification procedures, monotonicity under richer control, a decomposition of memory growth into reachability expansion and visibility activation, symmetry-protected compression, and no-compression results under sufficiently rich intervention algebras.The framework is connected to process-tensor bond rank, hidden-memory realization, and physical memory costs. At a prepare-and-measure temporal cut, linear, classical, and quantum memory are distinguished through ordinary rank, nonnegative rank, and positive-semidefinite rank. A robust quantum-dimension witness is derived for an n-message decoding task.A proposed single-photon spatial-mode experiment provides a concrete test. Phase-coded histories have identical present path-intensity distributions but can be perfectly distinguished by permitted future phase and Fourier operations, giving an ideal response matrix equal to the identity. For 17 modes, sufficiently high decoding success directly certifies a 17-dimensional quantum memory.The paper distinguishes established realization, automata, process-tensor, quantum-control, and factorization results from the proposed synthesis and experimental interpretation. It concludes with two natural next directions: experimental implementation and a multitime completely-positive / PSD-rank theory of intervention-relative quantum memory.
Exact physical memory is an unstable target: a process of Hankel rank one can be arbitrarily close to processes requiring larger classical and quantum carriers. We make this obstruction quantitative and then identify the extra structure under which a physical finite-sample upper theorem is possible. First, on a stationary binary Markov family whose mixing gap is uniformly at least one half, every estimator of exact linear rank, minimum classical memory, or minimum quantum Hilbert dimension has worst-case error at least 3/8 at local separation Θ([N(L−1)]−1/2)\Theta([N(L-1)]^{-1/2})Θ([N(L−1)]−1/2) from the rank-one model. The smallest nonzero singular value of an explicit Hankel core is proportional to the same parameter. Thus a spectral or dynamical margin is necessary even under rapid mixing. Second, finite Hankel data determine a minimum linear realization only up to similarity, whereas nonnegativity and complete positivity are not invariant under a generic similarity. We give an exact two-state counterexample and then formulate the missing assumption as a calibrated physical core. For an invertible core HHH, all shifts are recovered in one basis by A~a=H~−1H~a\tilde{A}_a = \tilde{H}^{-1}\tilde{H}_aA~a​=H~−1H~a​. An explicit perturbation theorem, followed by Choi positive-part projection and one common trace-preserving correction, returns a genuine quantum instrument. A direct-query finite-sample corollary controls its error uniformly over adaptive policies and exposes every dependence on core conditioning, physical calibration, horizon, and confidence. The resulting linear-in-horizon coefficient is first-order sharp: a one-state classical process attains the exact discrepancy 1−(1−Δ/2)L1-(1-\Delta/2)^L1−(1−Δ/2)L at one-step diamond radius Δ\DeltaΔ. Third, we prove a compact-state classicalization theorem. Any finite-dimensional physical memory described by a compact base-norm state space has, at horizon LLL and operational error η>0\eta>0η>0, a classical realization whose number of states is the seeded covering number at radius 2η/L2\eta/L2η/L, uniformly over adaptive policies. For a ddd-dimensional quantum instrument this gives, for d≥2d\geq 2d≥2, (1+2(1−d−1)Lη)d2−1 \left(1+\frac{2(1-d^{-1})L}{\eta}\right)^{d^2-1} (1+η2(1−d−1)L​)d2−1 states, with no extra transient state; one state suffices for d=1d=1d=1. The ambient exponent d2−1d^2-1d2−1 is optimal for universal trace-norm discretization, while a compact branch-invariant reachable set replaces it by its own metric entropy. Consequently, an unbounded exact quantum/classical memory separation cannot remain unbounded at fixed ddd, LLL, η\etaη. Robust quantum advantage must instead be quantified as a joint precision–horizon tradeoff. Deterministic verification code audits the mixing lower bound, the similarity-gauge counterexample, common-shift recovery, physical repair, and finite-net simulation.
How much internal quantum memory is needed to reproduce a stationary stochastic process, and how does reducing that memory change the environmental resources required when output records are retained? This paper establishes exact quantum-source constructions and resource bounds under explicitly specified readout and retained-output conditions. A general completion theorem identifies the minimum Hilbert-space dimension for stationary triple data in the orthogonal-current-symbol source class: it equals the sum of the complex positive-semidefinite ranks of the middle-symbol slices. An attaining construction reproduces the entire canonical stationary order-two process. For the delayed trine process, the minimum classical hidden Markov model requires nine states at every non-independent parameter value. Quantum instruments reproduce the same complete output law using four-dimensional memory throughout the parameter range, with four dimensions proved necessary in a certified low-noise regime and three dimensions both sufficient and necessary on an explicit high-noise interval. The paper then separates reusable memory dimension from environmental dilation cost. On the certified high-noise interval, the exact one-step Pareto frontier consists of a three-dimensional source with recorded-channel Choi rank nine and a four-dimensional source with rank three. Here, Choi rank measures the minimum initially pure environmental remainder dimension for the channel retaining both the emitted classical symbol and final quantum memory. A separate mixed-support qutrit construction at noise parameter 0.99 improves complete-history costs and gives upper bounds close to universal lower bounds. These results show why optimizing a single step need not optimize a retained multi-step history. The analysis also distinguishes complete channels, fixed-input states and archive-only outputs, and demonstrates how access to a retained reference can distinguish generators with identical passive output statistics. Full-word proofs, algebraic complete-positivity certificates, a technical companion and reproducibility materials accompany the results. The exact intermediate memory boundary and complete-history optimum remain open. All results use standard quantum probabilities and apply to the stated operational resource models; they do not establish total hardware, energy or thermodynamic minima.
Nonlinear dynamics generate non-Gaussian states and operations, but they also make continuously monitored quantum systems difficult to track. A quantum filter performs this task by compressing a noisy measurement record into a set of evolving variables that predict future observables. For Gaussian dynamics, this compression is exact and finite. In nonlinear systems, conventional moment equations generally expand without bound, but the failure of one representation does not prove that every finite description is impossible. Here we formulate quantum filtering as a realization problem for general quantum input--output maps. We introduce the linearized observable Hankel operator KOK_OKO​, which maps changes in the past measurement record to changes in a future conditional observable. Any finite-dimensional smooth filter forces KOK_OKO​ to have finite rank, so infinite rank provides a coordinate-independent proof that no exact finite-dimensional filter exists. Applied to single-mode polynomial bosonic systems under continuous quadrature monitoring, this yields a sharp boundary: Gaussian dynamics and the measurement-aligned Conditional Momentum Moment class admit finite realizations, whereas dynamics outside these classes produce infinitely many response directions and admit no finite-dimensional smooth exact filter. Exact impossibility, however, need not imply large practical complexity. The singular values of KOK_OKO​ quantify how many response directions matter at a chosen local accuracy, defining the effective memory of an observable. For the Kerr and Duffing oscillators, these singular values decay rapidly despite infinite exact rank, showing that infinite exact memory can coexist with low finite-accuracy complexity. The framework therefore separates exact realizability from effective complexity in nonlinear quantum filtering.
A coarse description of a dynamical system generally fails to evolve autonomously: evolving and coarse-graining in different orders gives different predictions, and the compressed propagator fails the semigroup law. We show that this path dependence is the observable boundary of missing state. For a finite continuous-time Markov chain with generator LLL, coarse channel CCC, and stochastic section RRR satisfying CR=ICR=ICR=I, define P=RCP=RCP=RC, Q=I−PQ=I-PQ=I−P, and K=CLRK=CLRK=CLR. The exact coarse propagator Tt=CetLRT_t=C\mathrm{e}^{tL}RTt​=CetLR obeys a Volterra equation with memory kernel M(t)=CLQ etQLQ QLR\mathcal{M}(t)=CLQ\,\mathrm{e}^{tQLQ}\,QLRM(t)=CLQetQLQQLR. Every memory jet M(k)(0)\mathcal{M}^{(k)}(0)M(k)(0) is recursively recoverable from the short-time derivatives of TtT_tTt​ alone. The rank of the block Hankel operator H∞=(M(i+j)(0))i,j≥0\mathbb{H}_{\infty}=(\mathcal{M}^{(i+j)}(0))_{i,j\geq 0}H∞​=(M(i+j)(0))i,j≥0​ is exactly the smallest number of auxiliary linear state variables required to turn the coarse dynamics into an exact autonomous, time-homogeneous system. A minimal completion is constructible by Ho--Kalman factorization and is unique up to similarity. If d=min⁡{k:M(k)(0)≠0}d=\min\{k:\mathcal{M}^{(k)}(0)\neq 0\}d=min{k:M(k)(0)=0}, then every smooth classical fff-divergence generically scales as t2d+4t^{2d+4}t2d+4. Exact Markov examples realize exponents 4,6,84,6,84,6,8 and closure ranks 1,2,31,2,31,2,3. We also derive finite-sample singular-value certificates for hidden predictive coordinates, including a matrix-Bernstein bound with intrinsic-variance control. Finally, histories are identified when they induce the same future law under every admissible intervention, yielding the minimal counterfactual predictive state. The results turn coarse-graining into a constructive state-completion problem and separate ordinary linear realization from the harder requirement of a positive or Markov realization.
What information about environmental memory must a quantum controller retain to guarantee a specified computational output? This preprint formulates that question in terms of task-sufficient process representations, intervention-valid predictions, stochastic clock laws, and explicit resource accounting. The central requirement is to preserve the outcomes and costs of the permitted decisions, with a stated accuracy contract and a finite-sample certificate. The paper connects two complementary lines of analysis. An operational construction studies Gaussian dephasing and controlled qubit tasks. A separate stochastic-clock analysis studies direct and inverse time changes, fractional relaxation, and the information that temporal coarse graining must transport. These constructions address related modelling questions but are not identified as a single microscopic noise model. The main elements are: Identical uncontrolled channels, different controlled responses. A continuum of Gaussian phase processes has the same uncontrolled coherence at every time and zero Breuer-Laine-Piilo backflow, yet different spin-echo responses. The construction shows why an uncontrolled channel family alone can fail to determine an intervention-dependent control decision. Task-sufficient equivalence and resource contracts. The manuscript defines a contract functional collecting accepted-output moments, acceptance probabilities, and physical costs over a declared policy family. Its finite linear task quotient identifies distinctions relevant to those predictions. Approximate sufficiency is accompanied by stability bounds that expose the sensitivity of conditional guarantees to low acceptance probabilities and small feasibility margins. Simultaneous finite-policy certification. Conservative confidence bounds certify conditional task bias and expected cost per accepted outcome after selection among a fixed, finite set of candidate policies. The assumptions include independent reset attempts, bounded observations and costs, and a trusted ideal reference. Failure to certify a candidate is explicitly distinguished from proof that no feasible physical policy exists. An explicit coarse-graining framework for stochastic clocks. The proposed decorated phase space records the process, permitted policy family, observable set, and clock law. The clock component transports stable laws under their natural scaling and relates domains of attraction to regular variation of waiting-time tails. Within the stated two-parameter clock-and-coupling family, the relaxation-rate direction scales with eigenvalue c^alpha. These clock results are placed in the established probabilistic renormalization literature; they do not establish hyperbolicity or convergence for the full decorated operator. A reset-sensitive fractional-relaxation calculation. Direct subordination of a quantum dynamical semigroup is distinguished from inverse-stable time change. For the latter, the paper compares uninterrupted evolution with independent renewal of the hidden clock and relevant environment. At half order and dimensionless rate one, the exact erfcx representation gives the short-time defect Delta(h) ~ 0.660989212585294 sqrt(h). The difference concerns the declared reset protocols, rather than a general assertion that every physical gate resets its environment. The reported classical benchmark uses ten candidate policies with explicit pulse depolarisation and separate X- and Z-expectation contracts. At the quasistatic endpoint, midpoint echo increases average identity-gate fidelity from 0.789293 to 0.990050. The fidelity crossover near r = 0.0467883 is distinct from the stricter X-task feasibility threshold r >= 0.9688074. For the selected Z-task input, dephasing is harmless and unnecessary echo pulses reduce the expectation. Conventional calibrated echo matches the best tested control performance; calibration, sampling, rejected attempts, and amortisation remain part of the resource account. Two questions are isolated as conjectures: boundedness of the task-quotient dimension along the coarse-graining orbit, and rigidity of the task quotient as a complete contract invariant. The manuscript specifies falsifiable numerical and experimental tests, including rescaled relaxation-profile collapse and finite tests of quotient-rank growth. It establishes no hardware advantage, scalable task-compression theorem, logical error-rate improvement, or error-correction overhead reduction. This illustrated version 2.0 contains 30 pages, four scientific figures, derivation and reproduction appendices, a methodological comparison with renormalization in one-dimensional dynamics, a glossary, and 58 references. The manuscript is dated 12 November 2026 and supersedes version 1.0 of 10 November 2026. The four figures show the uncontrolled/control-response contrast, task-dependent protection, finite-policy certification alongside dimensionless fractional relaxation, and the exact scalar clock-reset defect.