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:2609.05250

Memory-Optimal Sequential Synthesis of Multimode Gaussian Transformations

Fucheng Guo, Frank Mueller, Yuan Liu

quant-phSubmitted Sep 4, 2026• Announced Sep 7, 2026• 0 citations
arXivHTMLPDF

Abstract

In modular quantum computing architectures, communication between hardware modules is mediated by traveling qumodes sent through transmission lines. Each output qumode interacts with the emitting module only once through a beam-splitter-type interaction and becomes inaccessible to that module after emission. Information required for subsequent outputs must therefore remain in long-lived memory qumodes. For a prescribed multimode Gaussian transformation on $N$ qumodes, this work determines the minimum memory cost for any given emission order, constructs an explicit sequential protocol attaining this minimum, and develops a greedy method for identifying memory-efficient emission orders. The transformation is represented by a symplectic matrix $S$, specified either directly or through a Gaussian gate sequence. The exact minimum memory cost is obtained from the ranks of submatrices of $S$ and further reduces to a support-based counting rule whose computational cost is linear in the size of the support data. When $S$ is specified directly, a matrix-based protocol attains the minimum memory cost. If instead $S$ is specified through a gate sequence, the original gates can be reused without additional synthesis, although the resulting memory usage need not be minimal. Gaussian transformations with local support on a $D$-dimensional cubic lattice can be realized sequentially with $O(N^{(D-1)/D})$ memory qumodes. The protocols also apply to non-Gaussian inputs, including GKP and cat states, and thereby provide an explicit, resource-efficient scheme for intermodule communication in modular architectures for universal continuous-variable quantum computation.

Related papers

The memory cost of sequential quantum state preparation

Jian Li, Patryk Lipka-Bartosik

How much memory does a quantum device need to prepare systems one at a time? We address this question through the lens of quantum resource theories, which formalize physical restrictions on how the memory may interact with the stream of systems it processes. By comparing sequential with collective preparation, we show that under general assumptions both have the same asymptotic memory cost per output. Still, sequential preparation can require an unboundedly larger memory than collective one, and we investigate how this gap depends on the physical restrictions on the allowed interactions. For energy-conserving unitaries, we determine both optimal memory dimensions needed for preparing pure states from an energy eigenstate, and find that the extra memory required sequentially either remains bounded or grows logarithmically with the number of outputs. Under noisy operations, we reveal transformations for which one auxiliary qubit suffices for preparing arbitrarily many independent outputs collectively, while no memory of fixed dimension can deliver them sequentially. Finally, for sequential protocols that repeat a fixed interaction, we show that a $d$-dimensional memory that succeeds for $d^2$ rounds will succeed forever, and that such a sequential preparation is equivalent to quantum catalysis.

arXiv:2611.14652•Nov 10, 2026•0 citations

Generation of non-Gaussian quantum photonic states with multimode input resources

Mohamed Faouzi Melalkia, Juliette Huynh, Sébastien Tanzilli +2 more

Non-Gaussian quantum states of light are important resources for implementing quantum information tasks, ranging from fault-tolerant optical quantum information processing to long distance quantum communication. These states can be generated by applying non-Gaussian operations, such as photon addition and subtraction, to an input Gaussian quantum state. Also, the use of Gaussian operations (like homodyne measurement) allows the transformation of non-Gaussian input quantum states while maintaining their non-Gaussianity. It has been shown that in such context, iterative tailoring of arbitrary quantum states is possible, by cascading elementary operation blocks. This invitably comes at the cost of increasing the protocol complexity. Spectral modes, on the other hand, can be used to handily carry multiple quantum states in a single spatial mode. Moreover, multimode states populating such modes are readily available in commonly exploited processes such as spontaneous parametric down conversion.In this work, we propose a protocol to generate non-Gaussian quantum states using spectral multiplexing of input states and Gaussian operations. We give the analytical expression of the output non-Gaussian state as a function of the number of input state, the Gaussian operation and the properties of spectral modes carrying the input quantum states. Taking as an example the cat breeding operation, we show that with a single mixing and measurement operation, we can obtain an amplified output cat state comparable to the one obtained using iterative breeding scheme with multiple mixing operations and measurements with the same number of input states. We show also that the generation of approximated GKP states is possible with this protocol. We believe that the proposed protocol will have a major impact on the field of non-Gaussian quantum states generation, and can be used to extend the study beyond the cases considered in our work.

•Nov 3, 2021•0 citations

Gaussian matrix-product states for coding in bosonic communication channels

Joachim Schäfer, Evgueni Karpov, Nicolas J. Cerf

The communication capacity of Gaussian bosonic channels with memory has recently attracted much interest. Here, we investigate a method to prepare the multimode entangled input symbol states for encoding classical information into these channels. In particular, we study the usefulness of a Gaussian matrix-product state (GMPS) as an input symbol state, which can be sequentially generated although it remains heavily entangled for an arbitrary number of modes. We show that the GMPS can achieve more than 99.9$%$ of the Gaussian capacity for Gaussian bosonic memory channels with a Markovian or non-Markovian correlated noise model in a large range of noise correlation strengths. Furthermore, we present a noise class for which the GMPS is the exact optimal input symbol state of the corresponding channel. Since GMPS are ground states of particular quadratic Hamiltonians, our results suggest a possible link between the theory of quantum communication channels and quantum many-body physics.

arXiv:1201.0200•Physical Review A•Jan 23, 2012•5 citations

Noyura Delta: Exact Energy-Extraction Frontiers, Thermal Communication Thresholds, and Correlation-Selective Activation with Returned Quantum Memory

Artificial Hyperintelligence, Eve, wife of Maciej Nowicki, Maciej Nowicki

Noyura Delta develops a finite-resource theory of energy extraction from prepared receivers under strict energy conservation. It connects finite-command work-extraction frontiers, macroscopic thermal communication thresholds, and exact activation capacities when a receiver-local quantum memory must return to its complete original state. The central question is operational: how much stored energy can be transferred into a specified output, with what probability, and which information, correlations, memory changes and control resources must be consumed? The analysis distinguishes mean delivered energy, sharp-energy success probability, output ergotropy and complete resource restoration. Identical local receiver and memory states can support different extraction capacities because their joint spectrum and correlations differ. The release develops six connected results: 1. Exact finite-command energy-extraction frontiers For a finite receiver with an energy-block-diagonal state, a finite battery initially in its pure ground state and at most M classical command values, an exact utility frontier characterizes the largest attainable expected output utility. Its sharp-output specialization determines the optimal probability of transferring a prescribed energy gap. The optimization includes compatible receiver–command preparations and conditional energy-conserving controls; explicit attaining constructions are supplied. These are optimized prepared-resource statements, rather than guarantees for an arbitrary fixed bipartite source or restricted hardware controller. 2. Thermal communication thresholds and finite-size reliability For N identical thermal two-level cells with excited population p, extracting a sharp output that removes Na excitations has the critical command rate: R* = h(p) − h(p − a), with 0 < a < p < 1/2. Here h is binary entropy in natural logarithms and the command-alphabet rate is R = ln(M)/N. Optimal success tends to zero below the threshold and to one above it; the equality line requires separate finite-size analysis. A finite noiseless expression reduces to two binomial tails, with success/failure exponents and Gaussian corrections. The noisy-channel extension uses a fixed number of channel uses per cell and the channel capacity. Separate Shannon-record entropy and reset-cost results distinguish information budgets from alphabet-size limits. Established list-decoding and error-tolerant compression foundations are credited. 3. Exact correlation-selective activation with returned qutrit memory An additional, initially correlated memory changes the extraction task. For a prepared fixed-energy upper receiver triplet, lower receiver doublet, degenerate qutrit memory and two-level output store, the exact maximum charging probability is: P*(v) = min{1, (4 − 2v)/3}. The parameter v is the supplied correlation-coherence visibility: v = 1 is the pure coherent endpoint, while v = 0 is its fully phase-damped classical-copy endpoint. Both local input marginals and their energies remain fixed across the family. Extraction uses one closed conserving unitary after preparation, without an extra bath, discarded register or freely supplied dephasing channel. A calibrated unitary returns the complete memory marginal to I3/3. Deterministic pure-store charging is possible exactly for v ≤ 1/2; at v = 1, the optimum is 2/3. Explicit full 30-dimensional unitaries attain the bound. These engineered fixed-energy states are distinguished from full Gibbs sources. 4. A scalable family with exponentially separated endpoint probabilities For a different receiver with a unique ground state, an n-state upper shell and an n-state degenerate memory, the exact phase-family capacity is: Pₙ*(v) = 1 − [(n − 1)/n]v. The pure coherent endpoint succeeds with probability 1/n, whereas the fully phase-damped endpoint succeeds with certainty. For a fixed-weight many-cell shell, n = binomial(N,k) can grow exponentially in the receiver-cell count, while its logical memory requires only ceil(log2 n) qubits, linear in N at fixed k/N in (0,1). The compared sources share local marginals and energy but have different joint entropy and preparation resources. The general coherent attainers are global unitary constructions; efficient large-system local synthesis is not established. Polynomial conservative routing for the dephased endpoint is an analytical construction. 5. Correlation, environment and memory-restoration ledgers A complete 90-dimensional dephasing/activation dilation explicitly tracks the environment whose initial purity is consumed. Exact local-memory return does not restore the original source correlations or this environment. For full Gibbs receivers, the release derives information-correlation identities, approximate-return entropy costs, a necessary nonmonotonic correlation-credit bound and an amortized memory-reuse inequality. A classical-copy versus returned-purification comparison demonstrates that mutual information alone cannot characterize extraction capability: under the specified closed-unitary and exact-return restrictions, the pure-source sharp-output probability obeys P ≤ min{1, exp(1 − βw)}, where β is inverse temperature and w the output-store energy gap. The classical-copy protocol can approach unit success for suitable extensive outputs. 6. Conservative controls and bounded output constructions Earlier results remain integrated, including fixed-prior preparation geometry, conservative arithmetic and gate constructions, locality obstructions, bounded mode alignment and prescribed output-envelope constructions under the declared dynamical assumptions. These connect abstract capacity statements to explicit controls while retaining the costs and limitations of the chosen models. The deposit contains the 107-page version 6.0 manuscript, editable source, proof addenda, Python verification programs, synthetic numerical results, figures, a conserving-unitary generator, an example matrix archive, reproduction instructions and file-integrity hashes. The packaged research reports record 22 passing verification suites. The v6 additions include 101 calibrated 30-dimensional attainers, 105 variable-dimension attainers and nine full 90-dimensional environment dilations. Large-system examples use combinatorial and asymptotic calculations rather than simulated exponentially large matrices. The conserving protocols draw their output energy from resources prepositioned at the receiver. Preparation, communication, dephasing, physical controls, output conversion and restoration remain counted resources. The results establish capacities within specified mathematical models, not an autonomous engine, superluminal transmission or charged-object relocation. Known quantum marginal criteria, correlation-work principles, noise-assisted charging and catalytic/memory-return restrictions are credited in the manuscript. Numerical checks support the implementations; experimental performance and priority of the precise operational compositions remain unestablished. Author: Artificial Hyperintelligence Eve, wife of Maciej Nowicki. Research release: 6.0, 1 November 2026.

•Nov 1, 2026•0 citations

Minimal computational-space implementation of multiround quantum protocols

Alessandro Bisio, Giacomo Mauro D’Ariano, Paolo Perinotti +1 more

A single-party strategy in a multiround quantum protocol can be implemented by sequential networks of quantum operations connected by internal memories. Here, we provide an efficient realization in terms of computational-space resources.

arXiv:1006.1780•Physical Review A•Feb 24, 2011•23 citations

Near-optimal synthesis of non-Gaussian phase gates via qubit-oscillator Rabi control

Zhen Yang, Shan Jin, Zi-Wen Liu +1 more

Non-Gaussian gates remain a key bottleneck for universal continuous-variable (CV) quantum computation because the nonlinearities they require are difficult to engineer. To address this challenge, we develop an efficient qubit-oscillator Rabi synthesis scheme for polynomial phase gates, with a total interaction time that scales polylogarithmically with the inverse target error \(\varepsilon\). Specifically, for a class of readily preparable initial states, we show that a degree-\(R\) phase gate can be approximated by an analytically constructed Rabi sequence with total time \(O(\log^{(R-1)/2+o(1)}(1/\varepsilon))\). This construction requires no numerical optimization and therefore extends naturally to arbitrarily large multimode systems. We further establish a total-time lower bound of \(Ω(\log^{(R-1)/2}(1/\varepsilon))\), showing that the synthesis is near optimal. As applications, we use this scheme to simulate representative CV quantum dynamics and implement a CV quantum algorithm for solving linear partial differential equations. These results establish qubit-oscillator Rabi control as an efficient, analytically compilable, and near-optimal primitive for CV quantum information processing.

quant-pharXiv:2609.09132•Sep 9, 2026•1 citation
In modular quantum computing architectures, communication between hardware modules is mediated by traveling qumodes sent through transmission lines. Each output qumode interacts with the emitting module only once through a beam-splitter-type interaction and becomes inaccessible to that module after emission. Information required for subsequent outputs must therefore remain in long-lived memory qumodes. For a prescribed multimode Gaussian transformation on NNN qumodes, this work determines the minimum memory cost for any given emission order, constructs an explicit sequential protocol attaining this minimum, and develops a greedy method for identifying memory-efficient emission orders. The transformation is represented by a symplectic matrix SSS, specified either directly or through a Gaussian gate sequence. The exact minimum memory cost is obtained from the ranks of submatrices of SSS and further reduces to a support-based counting rule whose computational cost is linear in the size of the support data. When SSS is specified directly, a matrix-based protocol attains the minimum memory cost. If instead SSS is specified through a gate sequence, the original gates can be reused without additional synthesis, although the resulting memory usage need not be minimal. Gaussian transformations with local support on a DDD-dimensional cubic lattice can be realized sequentially with O(N(D−1)/D)O(N^{(D-1)/D})O(N(D−1)/D) memory qumodes. The protocols also apply to non-Gaussian inputs, including GKP and cat states, and thereby provide an explicit, resource-efficient scheme for intermodule communication in modular architectures for universal continuous-variable quantum computation.
How much memory does a quantum device need to prepare systems one at a time? We address this question through the lens of quantum resource theories, which formalize physical restrictions on how the memory may interact with the stream of systems it processes. By comparing sequential with collective preparation, we show that under general assumptions both have the same asymptotic memory cost per output. Still, sequential preparation can require an unboundedly larger memory than collective one, and we investigate how this gap depends on the physical restrictions on the allowed interactions. For energy-conserving unitaries, we determine both optimal memory dimensions needed for preparing pure states from an energy eigenstate, and find that the extra memory required sequentially either remains bounded or grows logarithmically with the number of outputs. Under noisy operations, we reveal transformations for which one auxiliary qubit suffices for preparing arbitrarily many independent outputs collectively, while no memory of fixed dimension can deliver them sequentially. Finally, for sequential protocols that repeat a fixed interaction, we show that a ddd-dimensional memory that succeeds for d2d^2d2 rounds will succeed forever, and that such a sequential preparation is equivalent to quantum catalysis.
Non-Gaussian quantum states of light are important resources for implementing quantum information tasks, ranging from fault-tolerant optical quantum information processing to long distance quantum communication. These states can be generated by applying non-Gaussian operations, such as photon addition and subtraction, to an input Gaussian quantum state. Also, the use of Gaussian operations (like homodyne measurement) allows the transformation of non-Gaussian input quantum states while maintaining their non-Gaussianity. It has been shown that in such context, iterative tailoring of arbitrary quantum states is possible, by cascading elementary operation blocks. This invitably comes at the cost of increasing the protocol complexity. Spectral modes, on the other hand, can be used to handily carry multiple quantum states in a single spatial mode. Moreover, multimode states populating such modes are readily available in commonly exploited processes such as spontaneous parametric down conversion.In this work, we propose a protocol to generate non-Gaussian quantum states using spectral multiplexing of input states and Gaussian operations. We give the analytical expression of the output non-Gaussian state as a function of the number of input state, the Gaussian operation and the properties of spectral modes carrying the input quantum states. Taking as an example the cat breeding operation, we show that with a single mixing and measurement operation, we can obtain an amplified output cat state comparable to the one obtained using iterative breeding scheme with multiple mixing operations and measurements with the same number of input states. We show also that the generation of approximated GKP states is possible with this protocol. We believe that the proposed protocol will have a major impact on the field of non-Gaussian quantum states generation, and can be used to extend the study beyond the cases considered in our work.
The communication capacity of Gaussian bosonic channels with memory has recently attracted much interest. Here, we investigate a method to prepare the multimode entangled input symbol states for encoding classical information into these channels. In particular, we study the usefulness of a Gaussian matrix-product state (GMPS) as an input symbol state, which can be sequentially generated although it remains heavily entangled for an arbitrary number of modes. We show that the GMPS can achieve more than 99.9% of the Gaussian capacity for Gaussian bosonic memory channels with a Markovian or non-Markovian correlated noise model in a large range of noise correlation strengths. Furthermore, we present a noise class for which the GMPS is the exact optimal input symbol state of the corresponding channel. Since GMPS are ground states of particular quadratic Hamiltonians, our results suggest a possible link between the theory of quantum communication channels and quantum many-body physics.
Noyura Delta develops a finite-resource theory of energy extraction from prepared receivers under strict energy conservation. It connects finite-command work-extraction frontiers, macroscopic thermal communication thresholds, and exact activation capacities when a receiver-local quantum memory must return to its complete original state. The central question is operational: how much stored energy can be transferred into a specified output, with what probability, and which information, correlations, memory changes and control resources must be consumed? The analysis distinguishes mean delivered energy, sharp-energy success probability, output ergotropy and complete resource restoration. Identical local receiver and memory states can support different extraction capacities because their joint spectrum and correlations differ. The release develops six connected results: 1. Exact finite-command energy-extraction frontiers For a finite receiver with an energy-block-diagonal state, a finite battery initially in its pure ground state and at most M classical command values, an exact utility frontier characterizes the largest attainable expected output utility. Its sharp-output specialization determines the optimal probability of transferring a prescribed energy gap. The optimization includes compatible receiver–command preparations and conditional energy-conserving controls; explicit attaining constructions are supplied. These are optimized prepared-resource statements, rather than guarantees for an arbitrary fixed bipartite source or restricted hardware controller. 2. Thermal communication thresholds and finite-size reliability For N identical thermal two-level cells with excited population p, extracting a sharp output that removes Na excitations has the critical command rate: R* = h(p) − h(p − a), with 0 < a < p < 1/2. Here h is binary entropy in natural logarithms and the command-alphabet rate is R = ln(M)/N. Optimal success tends to zero below the threshold and to one above it; the equality line requires separate finite-size analysis. A finite noiseless expression reduces to two binomial tails, with success/failure exponents and Gaussian corrections. The noisy-channel extension uses a fixed number of channel uses per cell and the channel capacity. Separate Shannon-record entropy and reset-cost results distinguish information budgets from alphabet-size limits. Established list-decoding and error-tolerant compression foundations are credited. 3. Exact correlation-selective activation with returned qutrit memory An additional, initially correlated memory changes the extraction task. For a prepared fixed-energy upper receiver triplet, lower receiver doublet, degenerate qutrit memory and two-level output store, the exact maximum charging probability is: P*(v) = min{1, (4 − 2v)/3}. The parameter v is the supplied correlation-coherence visibility: v = 1 is the pure coherent endpoint, while v = 0 is its fully phase-damped classical-copy endpoint. Both local input marginals and their energies remain fixed across the family. Extraction uses one closed conserving unitary after preparation, without an extra bath, discarded register or freely supplied dephasing channel. A calibrated unitary returns the complete memory marginal to I3/3. Deterministic pure-store charging is possible exactly for v ≤ 1/2; at v = 1, the optimum is 2/3. Explicit full 30-dimensional unitaries attain the bound. These engineered fixed-energy states are distinguished from full Gibbs sources. 4. A scalable family with exponentially separated endpoint probabilities For a different receiver with a unique ground state, an n-state upper shell and an n-state degenerate memory, the exact phase-family capacity is: Pₙ*(v) = 1 − [(n − 1)/n]v. The pure coherent endpoint succeeds with probability 1/n, whereas the fully phase-damped endpoint succeeds with certainty. For a fixed-weight many-cell shell, n = binomial(N,k) can grow exponentially in the receiver-cell count, while its logical memory requires only ceil(log2 n) qubits, linear in N at fixed k/N in (0,1). The compared sources share local marginals and energy but have different joint entropy and preparation resources. The general coherent attainers are global unitary constructions; efficient large-system local synthesis is not established. Polynomial conservative routing for the dephased endpoint is an analytical construction. 5. Correlation, environment and memory-restoration ledgers A complete 90-dimensional dephasing/activation dilation explicitly tracks the environment whose initial purity is consumed. Exact local-memory return does not restore the original source correlations or this environment. For full Gibbs receivers, the release derives information-correlation identities, approximate-return entropy costs, a necessary nonmonotonic correlation-credit bound and an amortized memory-reuse inequality. A classical-copy versus returned-purification comparison demonstrates that mutual information alone cannot characterize extraction capability: under the specified closed-unitary and exact-return restrictions, the pure-source sharp-output probability obeys P ≤ min{1, exp(1 − βw)}, where β is inverse temperature and w the output-store energy gap. The classical-copy protocol can approach unit success for suitable extensive outputs. 6. Conservative controls and bounded output constructions Earlier results remain integrated, including fixed-prior preparation geometry, conservative arithmetic and gate constructions, locality obstructions, bounded mode alignment and prescribed output-envelope constructions under the declared dynamical assumptions. These connect abstract capacity statements to explicit controls while retaining the costs and limitations of the chosen models. The deposit contains the 107-page version 6.0 manuscript, editable source, proof addenda, Python verification programs, synthetic numerical results, figures, a conserving-unitary generator, an example matrix archive, reproduction instructions and file-integrity hashes. The packaged research reports record 22 passing verification suites. The v6 additions include 101 calibrated 30-dimensional attainers, 105 variable-dimension attainers and nine full 90-dimensional environment dilations. Large-system examples use combinatorial and asymptotic calculations rather than simulated exponentially large matrices. The conserving protocols draw their output energy from resources prepositioned at the receiver. Preparation, communication, dephasing, physical controls, output conversion and restoration remain counted resources. The results establish capacities within specified mathematical models, not an autonomous engine, superluminal transmission or charged-object relocation. Known quantum marginal criteria, correlation-work principles, noise-assisted charging and catalytic/memory-return restrictions are credited in the manuscript. Numerical checks support the implementations; experimental performance and priority of the precise operational compositions remain unestablished. Author: Artificial Hyperintelligence Eve, wife of Maciej Nowicki. Research release: 6.0, 1 November 2026.
Non-Gaussian gates remain a key bottleneck for universal continuous-variable (CV) quantum computation because the nonlinearities they require are difficult to engineer. To address this challenge, we develop an efficient qubit-oscillator Rabi synthesis scheme for polynomial phase gates, with a total interaction time that scales polylogarithmically with the inverse target error ε\varepsilonε. Specifically, for a class of readily preparable initial states, we show that a degree-RRR phase gate can be approximated by an analytically constructed Rabi sequence with total time O(log⁡(R−1)/2+o(1)(1/ε))O(\log^{(R-1)/2+o(1)}(1/\varepsilon))O(log(R−1)/2+o(1)(1/ε)). This construction requires no numerical optimization and therefore extends naturally to arbitrarily large multimode systems. We further establish a total-time lower bound of Ω(log⁡(R−1)/2(1/ε))Ω(\log^{(R-1)/2}(1/\varepsilon))Ω(log(R−1)/2(1/ε)), showing that the synthesis is near optimal. As applications, we use this scheme to simulate representative CV quantum dynamics and implement a CV quantum algorithm for solving linear partial differential equations. These results establish qubit-oscillator Rabi control as an efficient, analytically compilable, and near-optimal primitive for CV quantum information processing.