Asymmetric Quantum LDPC Codes

Recently, quantum error-correcting codes were proposed that capitalize on the fact that many physical error models lead to a significant asymmetry between the probabilities for bit flip and phase flip errors. An example for a channel which exhibits such asymmetry is the combined amplitude damping and dephasing channel, where the probabilities of bit flips and phase flips can be related to relaxation and dephasing time, respectively. We give systematic constructions of asymmetric quantum stabilizer codes that exploit this asymmetry. Our approach is based on a CSS construction that combines BCH and finite geometry LDPC codes.

💡 Research Summary

The paper addresses the problem of quantum error correction in physical systems where the probabilities of bit‑flip (X) and phase‑flip (Z) errors differ dramatically—a situation common in many quantum hardware platforms because the relaxation time (T₁) is typically an order of magnitude larger than the dephasing time (T₂). The authors first quantify this asymmetry by analyzing a combined amplitude‑damping and dephasing channel. By applying the Pauli‑twirl technique they convert the non‑Pauli channel into an equivalent Pauli channel with error probabilities

 pₓ = p_y = (1 – e^{–t/T₁})/4, p_z = ½ – pₓ – ½ e^{–t/T₂}.

The ratio A = p_z/pₓ = 1 + 2(1 – e^{–t/T₁})(1 – T₁/T₂) e^{–t/T₁} – 1 quantifies the asymmetry; for short interaction times (t ≪ T₁) this simplifies to A ≈ 2T₁/T₂ – 1, showing that when T₁ ≫ T₂ the channel is dominated by phase errors.

To exploit this asymmetry the authors adopt the CSS (Calderbank‑Shor‑Steane) construction, which uses two classical linear codes Cₓ and C_z to correct X‑ and Z‑errors respectively. The key idea is to choose C_z with a larger minimum distance d_z (to protect against the frequent phase errors) while allowing Cₓ to have a smaller distance dₓ (since X‑errors are rarer). Two systematic families of asymmetric quantum codes are presented.

Family 1 – Pure EG‑LDPC Codes.
The authors use Euclidean geometry (EG) LDPC codes defined on the vector space 𝔽_{p^s}^m. For integers μ (0 < μ < m) they construct the incidence matrix of μ‑flats versus (μ–1)‑flats; the null space of this matrix yields the LDPC code C^{(1)}{EG}(m, μ, 0, s, p). Its minimum distance is lower‑bounded by A{EG}(m, μ, μ–1, s, p) + 1, where A_{EG} counts how many (μ–1)‑flats lie in a μ‑flat. By selecting parameters μ_z < μₓ (and ensuring m – μ_z + 1 ≤ μₓ < m) they obtain C_z = C^{(1)}{EG}(m, μ_z, …) with a stronger distance and Cₓ = C^{(1)}{EG}(m, μₓ, …) with a weaker distance. Lemma 2 (the CSS construction) guarantees the existence of an asymmetric quantum code