Sufficiency of non-Gaussian resource states for universal continuous-variable quantum computation with Gaussian operations and measurements

In continuous-variable (CV) quantum computing with light or microwave resonators, Gaussian operations and Gaussian measurements alone are classically simulable, so universality requires non-Gaussian resources. Marek and Fiurášek (2009) proved that any set of pure single-mode non-Gaussian states with finite expansion in the Fock (photon-number) basis suffices, as off-line ancillas, for implementing arbitrary multimode gates using only Gaussian operations and homodyne detection, and posed the open problem: which further non-Gaussian states or finite-parametric classes — beginning with the cubic phase state — are sufficient? Later work established one side and pieces of the other: Wigner negativity is necessary for quantum advantage (Mari and Eisert, 2012) but not sufficient (García-Álvarez et al., 2020); Gottesman-Kitaev-Preskill (GKP) grid states suffice universally and fault-tolerantly (Baragiola et al., 2019; Hosseinynejad et al., 2026); and a first general sufficient condition was given, but only within GKP-based circuit families (Calcluth et al., 2024). A necessary-and-sufficient characterization of non-Gaussian resources under Gaussian processing remains open.

Background

In continuous-variable (CV) quantum information, information is carried by bosonic modes of light or microwave resonators described by quadrature operators $\hat q$ and $\hat p$ (position-like and momentum-like observables) rather than by two-level systems. The experimentally accessible toolbox consists of Gaussian operations — displacements, passive linear optics (beam splitters and phase shifters), and squeezers (unitaries generated by Hamiltonians at most quadratic in the quadratures, which reshape noise between conjugate quadratures) — together with homodyne detection, the measurement of a general quadrature. Gaussian unitaries map Gaussian states (coherent and squeezed states) to Gaussian states, and Bartlett, Sanders, Braunstein, and Nemoto (2002) proved that computation using only Gaussian states, operations, and measurements can be simulated efficiently classically. Universality therefore requires going beyond the Gaussian regime: Lloyd and Braunstein (1999) showed that any Hamiltonian of degree higher than two, combined with Gaussian operations, yields universal CV quantum computation, the cubic phase gate $e^{i\gamma \hat q^{3}/3}$ being the canonical sufficient non-Gaussian ingredient.

Because strong optical nonlinearities are extremely hard to realize directly, Gottesman, Kitaev, and Preskill (2001) introduced the off-line resource approach: instead of applying a nonlinear Hamiltonian to the data mode, one prepares a non-Gaussian resource state off-line and consumes it via Gaussian gate teleportation — a teleportation in which the resource state, rather than the operations, supplies the non-Gaussian gate. The cubic phase state, an idealized (non-normalizable) state introduced in that work, enables the cubic phase gate in this way. This strategy, the CV analogue of magic states in qubit computation, moves all non-Gaussianity from gates into state preparation.

Marek and Fiurášek (2009) asked precisely which non-Gaussian resource states suffice in this model. They proved that any set of pure single-mode non-Gaussian states with finite expansion in the Fock basis (the photon-number basis, whose states are eigenstates of the mode's number operator) — for example, single-photon states — is sufficient: with Gaussian operations and homodyne measurements, arbitrary gates on multimode states can be implemented probabilistically, which they illustrated with an explicit scheme for the nonlinear sign gate (the operation that flips the sign of the single-photon component of a state). Their conclusion explicitly leaves open which further classes of states are sufficient, suggesting finite-parametric families such as the cubic phase state, while judging it unlikely that sufficiency could be decided for completely generic states of the infinite-dimensional Hilbert space.

The necessary side of the classification is now understood. The Wigner function is a quasiprobability distribution over phase space whose negative values signal nonclassicality; Mari and Eisert (2012) proved that any circuit whose initial states and subsequent operations admit positive Wigner representations can be simulated efficiently classically, so Wigner negativity is necessary for quantum advantage under Gaussian processing. Since a classical result identifies exactly the Gaussian states as the pure states with non-negative Wigner function, every pure non-Gaussian ancilla is Wigner-negative. Negativity is not sufficient, however: García-Álvarez, Calcluth, Ferraro, and Ferrini (2020) constructed CV circuits with large Wigner negativity that remain efficiently simulable, and state explicitly that the conditions under which Wigner-negative CV circuits offer advantage remain open.

On the sufficient side, universality has been proved for specific classes beyond finite Fock expansion: GKP grid states — ideal states that are simultaneous eigenstates of two commuting displacement operators, forming a comb in position and momentum space and approximated experimentally by finite-squeezing states — give universal and fault-tolerant computation with Gaussian operations alone (Baragiola et al., 2019), including realistic noisy GKP stabilizer states (Hosseinynejad et al., 2026). Resource theories of non-Gaussianity and Wigner negativity (Takagi and Zhuang, 2018; Albarelli et al., 2018) quantify these resources but do not decide computational sufficiency; the stellar-rank hierarchy of Chabaud, Markham, and Grosshans (2020), which classifies pure non-Gaussian states by the minimal number of photon-creation operators needed to build them from Gaussian states, provides a structural language for such questions; and Calcluth, Reichel, Ferraro, and Ferrini (2024) gave the first general sufficient condition for universality, but only within the restricted circuit family of GKP states, Gaussian operations, and homodyne measurements. No necessary-and-sufficient characterization is known.

Problem Statement

Consider the ancilla-assisted model of continuous-variable quantum state engineering: an experimenter may perform arbitrary multimode Gaussian unitary operations (passive linear optics, squeezing, displacements), homodyne (quadrature) measurements with arbitrary classical feed-forward, and may prepare arbitrary Gaussian states; in addition, a supply of specified non-Gaussian resource states can be prepared off-line. Marek and Fiurášek (2009) proved sufficiency for every set of pure single-mode states of finite Fock expansion and posed the question of which further classes suffice. The open problem is to characterize the non-Gaussian states that are sufficient resources in this model:

(1) Is there a necessary-and-sufficient condition — structural, phase-space, or resource-theoretic — for a state, or class of states, to render this Gaussian toolbox universal for CV quantum state engineering, i.e. for the implementation of arbitrary multimode unitary gates on bosonic states (equivalently, universal CV quantum computation)? Gate implementation is understood in the sense of the source model: probabilistic (heralded) and approximate — arbitrary multimode gates, with finite but arbitrary mode number, are implemented on arbitrarily large finite Fock-space truncations with nonzero success probability, the fidelity being traded against the success probability via the widths of the homodyne acceptance windows — rather than as exact deterministic gates on the full infinite-dimensional space.

(2) Concretely, for which further finite-parametric families of non-Gaussian states — such as cat states (superpositions of coherent states), binomial-type bosonic code states, photon-subtracted or photon-added squeezed states of infinite Fock rank, and finite-energy approximations to the ideal cubic phase state — can sufficiency be proved, and for which can insufficiency — the failure to render the Gaussian toolbox universal — be proved, for instance by exhibiting efficient classical simulability of the resulting circuits under Gaussian processing? Sufficiency of GKP-type grid states (including realistic noisy stabilizer versions) and of CV hypergraph states is already established; the question concerns the remaining families, and the families listed in this item constitute the classification target, each to receive a definite sufficiency-or-insufficiency verdict.

The question is posed at the level of ideal state classes together with their well-defined finite-energy approximations; no finite-size, energy, or parameter restriction is part of the problem beyond the class definitions themselves — the mode number and truncation size entering the model's notion of gate implementation are arbitrary, not fixed. The source authors conjectured that sufficiency is unlikely to be decidable for completely generic states of the infinite-dimensional Hilbert space; accordingly the task is resolved by either of two answer forms, and by nothing weaker. The first is a necessary-and-sufficient criterion as formulated in item (1) above, whether by exhibiting such a criterion or by proving that no such criterion exists in the stated generality. The second is a definite verdict for every family enumerated in item (2) above, each verdict proving sufficiency or insufficiency for that family. Verdicts on only some of the enumerated families, or a criterion restricted to a sub-class of circuit models such as the GKP-based circuits cited in the Background, do not resolve the task.

References

1. Seth Lloyd and Samuel L. Braunstein, 'Quantum Computation over Continuous Variables', Phys. Rev. Lett. 82, 1784 (1999). DOI: 10.1103/PhysRevLett.82.1784. arXiv:quant-ph/9810082. https://arxiv.org/abs/quant-ph/9810082
2. Daniel Gottesman, Alexei Kitaev, and John Preskill, 'Encoding a qubit in an oscillator', Phys. Rev. A 64, 012310 (2001). DOI: 10.1103/PhysRevA.64.012310. arXiv:quant-ph/0008040. https://arxiv.org/abs/quant-ph/0008040
3. Stephen D. Bartlett, Barry C. Sanders, Samuel L. Braunstein, and Kae Nemoto, 'Efficient Classical Simulation of Continuous Variable Quantum Information Processes', Phys. Rev. Lett. 88, 097904 (2002). DOI: 10.1103/PhysRevLett.88.097904. arXiv:quant-ph/0109047. https://arxiv.org/abs/quant-ph/0109047
4. Shohini Ghose and Barry C. Sanders, 'Non-Gaussian ancilla states for continuous variable quantum computation via Gaussian maps', Journal of Modern Optics 54(6), 855-869 (2007). DOI: 10.1080/09500340601101575. arXiv:quant-ph/0606026. https://arxiv.org/abs/quant-ph/0606026
5. Petr Marek and Jaromír Fiurášek, 'Resources for universal quantum state manipulation and engineering', Phys. Rev. A 79, 062321 (2009). DOI: 10.1103/PhysRevA.79.062321. arXiv:0908.1476. https://arxiv.org/abs/0908.1476
6. A. Mari and J. Eisert, 'Positive Wigner functions render classical simulation of quantum computation efficient', Phys. Rev. Lett. 109, 230503 (2012). DOI: 10.1103/PhysRevLett.109.230503. arXiv:1208.3660. https://arxiv.org/abs/1208.3660
7. Ryuji Takagi and Quntao Zhuang, 'Convex resource theory of non-Gaussianity', Phys. Rev. A 97, 062337 (2018). DOI: 10.1103/PhysRevA.97.062337. arXiv:1804.04669. https://arxiv.org/abs/1804.04669
8. Francesco Albarelli, Marco G. Genoni, Matteo G. A. Paris, and Alessandro Ferraro, 'Resource theory of quantum non-Gaussianity and Wigner negativity', Phys. Rev. A 98, 052350 (2018). DOI: 10.1103/PhysRevA.98.052350. arXiv:1804.05763. https://arxiv.org/abs/1804.05763
9. Ben Q. Baragiola, Giacomo Pantaleoni, Rafael N. Alexander, Angela Karanjai, and Nicolas C. Menicucci, 'All-Gaussian Universality and Fault Tolerance with the Gottesman-Kitaev-Preskill Code', Phys. Rev. Lett. 123, 200502 (2019). DOI: 10.1103/PhysRevLett.123.200502. arXiv:1903.00012. https://arxiv.org/abs/1903.00012
10. Darren W. Moore, 'Quantum Hypergraph States in Continuous Variables', Phys. Rev. A 100, 062301 (2019). DOI: 10.1103/PhysRevA.100.062301. arXiv:1909.03871. https://arxiv.org/abs/1909.03871
11. Laura García-Álvarez, Cameron Calcluth, Alessandro Ferraro, and Giulia Ferrini, 'Efficient simulatability of continuous-variable circuits with large Wigner negativity', Phys. Rev. Research 2, 043322 (2020). DOI: 10.1103/PhysRevResearch.2.043322. arXiv:2005.12026. https://arxiv.org/abs/2005.12026
12. Ulysse Chabaud, Damian Markham, and Frédéric Grosshans, 'Stellar Representation of Non-Gaussian Quantum States', Phys. Rev. Lett. 124, 063605 (2020). DOI: 10.1103/PhysRevLett.124.063605. arXiv:1907.11009. https://arxiv.org/abs/1907.11009
13. Timo Hillmann, Fernando Quijandría, Göran Johansson, Alessandro Ferraro, Simone Gasparinetti, and Giulia Ferrini, 'Universal Gate Set for Continuous-Variable Quantum Computation with Microwave Circuits', Phys. Rev. Lett. 125, 160501 (2020). DOI: 10.1103/PhysRevLett.125.160501. arXiv:2002.01402. https://arxiv.org/abs/2002.01402
14. Cameron Calcluth, Nicolas Reichel, Alessandro Ferraro, and Giulia Ferrini, 'Sufficient condition for universal quantum computation using bosonic circuits', PRX Quantum 5, 020337 (2024). DOI: 10.1103/PRXQuantum.5.020337. arXiv:2309.07820. https://arxiv.org/abs/2309.07820
15. Lina Vandré, Boxuan Jing, Yu Xiang, Otfried Gühne, and Qiongyi He, 'Graphical Framework for Non-Gaussian Quantum States', Quantum 9, 1809 (2025). DOI: 10.22331/q-2025-07-23-1809. arXiv:2409.07300. https://arxiv.org/abs/2409.07300
16. Boxuan Jing, Feng-Xiao Sun, and Qiongyi He, 'Task-Oriented Gaussian Optimization for Non-Gaussian Resources in Continuous-Variable Quantum Computation', arXiv:2509.15747 (2025). https://arxiv.org/abs/2509.15747
17. Fariba Hosseinynejad, Pavithran Iyer, Guillaume Dauphinais, and David L. Feder, 'Realistic Gottesman-Kitaev-Preskill Stabilizer States Enable Universal Quantum Computation', Phys. Rev. Lett. 136 (2026). DOI: 10.1103/ffln-vd4x. arXiv:2511.03874. https://arxiv.org/abs/2511.03874
18. Sharon David, Jack Davis, Ulysse Chabaud, and Francesco Arzani, 'Improved GKP magic states from error-corrected non-Gaussian quantum states', arXiv:2607.07833 (2026). https://arxiv.org/abs/2607.07833