Asymptotically good bosonic Fock state codes

Introduction

Bosonic modes are a natural platform for quantum information, but their susceptibility to amplitude‑damping (photon loss) limits the scalability of error‑corrected operations. Traditional multi‑mode Fock‑state codes protect only a sublinear number of losses relative to the total excitation, which means that as the code size grows the loss rate eventually overwhelms the code. A recent construction overcomes this limitation by leveraging random classical codes defined on the discrete simplex. The resulting asymptotically good bosonic Fock‑state codes correct a number of photon losses that scales linearly with the total photon number, while also guaranteeing a bounded occupancy per mode.

---

Randomness from Classical Indexing

The key insight is to separate the quantum code’s structure from the randomness that typically appears in Haar‑random constructions. Instead of randomizing the quantum states directly, the authors randomize the classical indexing that maps each logical basis vector to a specific multi‑mode excitation pattern. By drawing these index vectors from a random classical code in the simplex (the set of non‑negative integer vectors with fixed sum), they obtain a family of Fock‑state superpositions whose statistical properties mimic those of fully random quantum codes. This approach simplifies analysis: concentration inequalities for sums of bounded random variables (e.g., Hoeffding’s inequality and McDiarmid’s bounded‑difference method) can be applied directly to the classical index distribution, yielding high‑probability guarantees on the code’s distance and error‑correction capability.

---

Linear Loss Protection and Bounded Occupancy

In the asymptotic regime where the total excitation \(N\) grows without bound, the construction ensures that the code can correct up to \(\alpha N\) photon losses for some constant \(0<\alpha<1\). This linear scaling contrasts with earlier families that only tolerated \(o(N)\) losses. Moreover, because each index vector is constrained to have entries bounded by a constant \(b\), no single mode ever contains more than \(b\) photons. This bounded per‑mode occupancy is crucial for experimental platforms such as superconducting resonators or trapped‑ion phonon modes, where higher photon numbers lead to faster decoherence and nonlinearities. By keeping each mode’s occupation low, the codes extend the effective coherence lifetime of the logical qubit without sacrificing the asymptotic error‑correction rate.

---

Connections to Permutation‑Invariant and Qudit Codes

A notable by‑product of the construction is a bridge to permutation‑invariant (PI) codes. By symmetrizing the randomly indexed Fock states over mode permutations, one obtains PI codes that inherit the same linear loss protection. This yields asymptotically good families of qudit PI codes, expanding the toolbox for error correction in systems where symmetry is a natural resource (e.g., atomic ensembles or spin‑chain architectures). Additionally, the simplex‑based indexing can be interpreted as coding in a monolithic nuclear state space, where the total excitation plays the role of a conserved quantum number. The same random‑code arguments then provide distance guarantees for these more abstract settings.

---

Practical Outlook

The construction is conceptually simple and amenable to implementation. Generating a random classical code on the simplex can be done efficiently with standard combinatorial algorithms, and the resulting Fock‑state superpositions can be prepared using existing techniques for deterministic photon‑addition or controlled displacement operations. Because the per‑mode photon number is bounded, experimental overhead remains modest even as the total excitation grows. Future work may explore tailored deterministic index sets that achieve similar performance with reduced randomness, or hybrid schemes that combine the simplex approach with bosonic cat or binomial encodings to exploit additional hardware‑specific advantages.

---

Conclusion

By injecting randomness at the level of classical index selection, the authors have produced a family of bosonic Fock‑state codes that are asymptotically good: they protect a linear fraction of photon losses while keeping each mode’s occupation low. This resolves a long‑standing gap in bosonic error correction, opens a pathway to robust, scalable quantum memories in photonic and microwave platforms, and links bosonic codes to broader families of permutation‑invariant and qudit codes. The result is a versatile, experimentally friendly framework for protecting quantum information against the dominant amplitude‑damping noise in bosonic systems.