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

arXiv:2609.09663

Finite Realizations and Effective Memory in Nonlinear Quantum Filtering

Jacob Emerson

quant-phSubmitted Sep 9, 2026• Announced Sep 10, 2026• 1 citation• Revised Oct 10, 2026 (v3)
v1 · Sep 9, 2026v2 · Sep 26, 2026v3 · Oct 10, 2026
arXivHTMLPDF

Abstract

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.

Comments
25 pages, 5 figures

Related papers

Exact Conditional Momentum Moments for Nonlinear Homodyne Trajectories

Jacob Emerson

Simulation of nonlinear stochastic master equations generally requires trajectory-level evolution of large Hilbert-space density matrices, making strongly nonlinear continuously monitored systems computationally challenging. Here we identify an exactly solvable class centered on polynomial Hamiltonians of commuting quadratures with linear damping and homodyne measurement of those same quadratures. For initial states with a Gaussian measured-quadrature marginal and polynomial conditional-momentum sections, we derive an exact finite stochastic representation through any fixed momentum order. The result extends to interacting multimode systems, arbitrary measurement efficiency, and finite thermal occupation. Its computational cost is linear in trajectory length and independent of Hilbert-space dimension, requiring no Fock-space truncation, Gaussian approximation, or approximate moment closure. As an application, we derive an exact causal filter for monitored cubic-phase gates that uses the homodyne record to determine the variance-minimizing momentum displacement. Numerical simulations validate the representation against direct stochastic-master-equation evolution, demonstrate substantial speedups over converged Fock-space propagation, and show that this record-conditioned recentering can efficiently recover nonlinear squeezing otherwise lost through trajectory averaging.

quant-pharXiv:2607.26093•Jul 29, 2026•1 citation

Lie-Algebraic Classical Simulation of Bosonic Systems Beyond Gaussian Dynamics

Adelina Bärligea, Timothy Heightman, Jakob S. Kottmann +1 more

Classical simulability is ultimately determined by both the dynamics of a quantum system and the observables being evaluated. Lie-algebraic simulation exploits the latter to make exact polynomial-time classical simulations by propagating observables through low-dimensional invariant operator spaces. However, its conventional formulation in terms of polynomial-dimensional dynamical Lie algebras does not directly accommodate bosonic systems as their algebras are neither compact nor semisimple. In this contribution, we overcome this limitation, making bosonic systems accessible to the Lie-algebraic programme of exact polynomial-time classical simulation. We prove that expectation values, fixed-order correlation functions, including multi-time correlators and out-of-time-ordered correlators, and gradients are efficiently computable whenever their operator modules have polynomial dimension. This recovers Gaussian quantum optics and extends it to non-Gaussian input states, while identifying exact polynomial regimes of interacting non-Gaussian dynamics including bounded-photon Kerr and pair-hopping Hamiltonians and nilpotent polynomial phase dynamics. We show that unlike in the finite-dimensional spin and fermionic setting treated previously, a finite-dimensional bosonic generator algebra alone does not guarantee finite observable dynamics. We further derive a controlled perturbative hierarchy for squeezing beyond exact sector confinement and confirm the predicted error orders numerically. We also evaluate operator spreading on interacting chains of up to $400$ modes and connect a topological doublon band with flux-reversed edge motion. These results provide a unified formalism for classifying, discovering, and systematically approximating tractable bosonic quantum dynamics with classical polynomial-time simulation.

quant-pharXiv:2608.17094•Aug 19, 2026•1 citation

The State Equation: Causal Cone Rank and Minimal Physical Memory

Максим Чурилов

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.

•Nov 3, 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

Finite-Resolution Position Measurements and Their Sharp Limit Detector POVMs, Instruments, and the MTT Completion Boundary

Peter Nero

Measurement is an ordinary physical interaction. Its mathematical description nevertheless has several layers: an effect-valued measure gives outcome probabilities, an instrument also gives conditional post-interaction states, a realized record requires an outcome-completion rule, and later dynamics may stabilize that record. This paper constructs the first two layers exactly for finite-resolution position detection and keeps the latter two distinct. On the square-integrable Hilbert space over Euclidean position space, a normalized detector response defines a commutative positive operator-valued measure that is precisely a classical smearing of the sharp position projection-valued measure. Its outcome density is the convolution of the detector response with the sharp Born density, and converges in total variation to that density. For densities with one integrable weak derivative, an explicit error bound separates detector resolution, detector shape, and spatial variation of the input. A square-root instrument realizes the same effects and produces a finite post-measurement filter. For a Gaussian response, detector blur adds exactly epsilon squared times the identity to the outcome covariance. A two-outcome qubit example shows that finite unsharp effects converge in operator norm to a discrete projective measurement, so not every projective measurement is a distributional singularity. In Modal Triplet Theory, these results supply a rigorous downstream detector interface. They do not derive the instrument from selected geometry, select one realized outcome, or identify detector width with a coherence or Ornstein-Uhlenbeck floor.

•Jul 29, 2026•0 citations

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
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.
Simulation of nonlinear stochastic master equations generally requires trajectory-level evolution of large Hilbert-space density matrices, making strongly nonlinear continuously monitored systems computationally challenging. Here we identify an exactly solvable class centered on polynomial Hamiltonians of commuting quadratures with linear damping and homodyne measurement of those same quadratures. For initial states with a Gaussian measured-quadrature marginal and polynomial conditional-momentum sections, we derive an exact finite stochastic representation through any fixed momentum order. The result extends to interacting multimode systems, arbitrary measurement efficiency, and finite thermal occupation. Its computational cost is linear in trajectory length and independent of Hilbert-space dimension, requiring no Fock-space truncation, Gaussian approximation, or approximate moment closure. As an application, we derive an exact causal filter for monitored cubic-phase gates that uses the homodyne record to determine the variance-minimizing momentum displacement. Numerical simulations validate the representation against direct stochastic-master-equation evolution, demonstrate substantial speedups over converged Fock-space propagation, and show that this record-conditioned recentering can efficiently recover nonlinear squeezing otherwise lost through trajectory averaging.
Classical simulability is ultimately determined by both the dynamics of a quantum system and the observables being evaluated. Lie-algebraic simulation exploits the latter to make exact polynomial-time classical simulations by propagating observables through low-dimensional invariant operator spaces. However, its conventional formulation in terms of polynomial-dimensional dynamical Lie algebras does not directly accommodate bosonic systems as their algebras are neither compact nor semisimple. In this contribution, we overcome this limitation, making bosonic systems accessible to the Lie-algebraic programme of exact polynomial-time classical simulation. We prove that expectation values, fixed-order correlation functions, including multi-time correlators and out-of-time-ordered correlators, and gradients are efficiently computable whenever their operator modules have polynomial dimension. This recovers Gaussian quantum optics and extends it to non-Gaussian input states, while identifying exact polynomial regimes of interacting non-Gaussian dynamics including bounded-photon Kerr and pair-hopping Hamiltonians and nilpotent polynomial phase dynamics. We show that unlike in the finite-dimensional spin and fermionic setting treated previously, a finite-dimensional bosonic generator algebra alone does not guarantee finite observable dynamics. We further derive a controlled perturbative hierarchy for squeezing beyond exact sector confinement and confirm the predicted error orders numerically. We also evaluate operator spreading on interacting chains of up to 400400400 modes and connect a topological doublon band with flux-reversed edge motion. These results provide a unified formalism for classifying, discovering, and systematically approximating tractable bosonic quantum dynamics with classical polynomial-time simulation.
We ask which finite classical or quantum memory is sufficient to generate the predictions of a stochastic process. Starting from its Hankel kernel Hp(u,v)=p(uv)H_p(u,v)=p(uv)Hp​(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∣im⁡Hp)\dim\ker(C|_{\operatorname{im} H_p})dimker(C∣imHp​​) 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 rcl≤dS2(dE2−1)r_{\rm cl}\le d_S^2(d_E^2-1)rcl​≤dS2​(dE2​−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.
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.
Measurement is an ordinary physical interaction. Its mathematical description nevertheless has several layers: an effect-valued measure gives outcome probabilities, an instrument also gives conditional post-interaction states, a realized record requires an outcome-completion rule, and later dynamics may stabilize that record. This paper constructs the first two layers exactly for finite-resolution position detection and keeps the latter two distinct. On the square-integrable Hilbert space over Euclidean position space, a normalized detector response defines a commutative positive operator-valued measure that is precisely a classical smearing of the sharp position projection-valued measure. Its outcome density is the convolution of the detector response with the sharp Born density, and converges in total variation to that density. For densities with one integrable weak derivative, an explicit error bound separates detector resolution, detector shape, and spatial variation of the input. A square-root instrument realizes the same effects and produces a finite post-measurement filter. For a Gaussian response, detector blur adds exactly epsilon squared times the identity to the outcome covariance. A two-outcome qubit example shows that finite unsharp effects converge in operator norm to a discrete projective measurement, so not every projective measurement is a distributional singularity. In Modal Triplet Theory, these results supply a rigorous downstream detector interface. They do not derive the instrument from selected geometry, select one realized outcome, or identify detector width with a coherence or Ornstein-Uhlenbeck floor.
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.