DualYbSim

Noise channel reference

Every channel in this library is named after the corresponding channel in the noise-model appendix of the dual-Yb surface-code paper. This page is the mapping between the paper’s notation and the Python identifiers.

Naming convention

Paper notation Identifier Example
subscript joined with _ LOSS_g, ZERR_gm, DECAY_rg
superscript (...) appended with _ LOSS_g^(meas) → LOSS_g_meas, DECAY_mg^(gate) → DECAY_mg_gate
probability p_n^(x) p_n_x p_1^(gm) → p_1_gm, p_2^(dual) → p_2_dual
transition probability p_(a→b)^(x) p_a_b_x p_(g→L)^(meas) → p_g_L_meas
rate Γ_ab gamma_ab Γ_rg → gamma_rg, Γ_gL → gamma_gL
time T_n^(x), t_x T_n_x, t_x T_2^(gm) → T_2_gm, t_1Q^(c) → t_1Q_c

Channel names appear as keys of the noise_channels dict on each Kraus class in dualybsim.kraus.yb171 / dualybsim.kraus.yb174, and as the channel argument of YbNoiseChannel.get_kraus_operators and CPTP.

Channels by category

The paper groups channels into six categories. Loss and decay channels are realised in Stim as HERALDED_ERASE; Pauli channels as PAULI_CHANNEL_1, DEPOLARIZE1, DEPOLARIZE2, X_ERROR or Z_ERROR.

Coherent-control

Channel Isotope Meaning
DEP1_c 174 single-qubit depolarising error on the clock transition
DEP1_g 171 single-qubit depolarising error inside the ground manifold
DEP1_m 171 single-qubit depolarising error inside the metastable manifold
DEP1_gm 171 depolarising error during clock-transition excitation
DECAY_mg_gate 171 metastable→ground damping driven by the nuclear-spin control laser
DEP2_c 174 two-qubit depolarising error, clock–clock
DEP2_m 171 two-qubit depolarising error, m–m
DEP2_dual 171, 174 two-qubit depolarising error, 171-m with 174-clock

Measurement

Channel Isotope Meaning
LOSS_g_meas 171, 174 atom loss from the ground manifold during fluorescence imaging
FLIP_g 171 in-manifold bit flip induced by the imaging lasers
MERR 171, 174 classical discrimination error of the loss-aware readout

Reset

Channel Isotope Meaning
LOSS_g_reset 171, 174 loss from the ground manifold during preparation or motional reset
LOSS_m_reset 171 loss from the metastable manifold during spin reset
FLIP_g 171 nuclear-spin flip in the ground manifold during reset
FLIP_m 171 nuclear-spin flip in the metastable manifold during reset

Idling

Probabilities follow p(t) = 1 − exp(−t/T).

Channel Isotope Timescale
ZERR_c 174 T_2^(c)
ZERR_g 171 T_2^(g)
ZERR_m 171 T_2^(m)
ZERR_gm 171 T_2^(gm), clock-transition coherence. Available in the Kraus layer but not injected into circuits — see parameter_reference.md
XERR_g 171 T_1^(g)
XERR_m 171 T_1^(m)

Decay

Probabilities follow p(t) = 1 − exp(−Γt). The Rydberg branches share a total rate Γ_Ryd split 0.51 / 0.42 / 0.07 into LOSS_r / DECAY_rg / DECAY_rm.

Channel Isotope Rate
LOSS_r 171, 174 Γ_rL, Rydberg decay to untrapped or dark states
DECAY_rg 171, 174 Γ_rg, Rydberg decay to the ground manifold
DECAY_rm 171, 174 Γ_rm, Rydberg decay to the metastable manifold
DECAY_mg 171, 174 Γ_mg, metastable→ground decay
LOSS_g 171, 174 Γ_gL, ground-manifold loss from the finite trap lifetime
LOSS_m 171, 174 Γ_mL, metastable-manifold loss from the finite trap lifetime

Transportation

Channel Isotope Meaning
LOSS_g_hand 171, 174 ground-manifold loss during trap handover
LOSS_m_hand 171, 174 metastable-manifold loss during trap handover

Which channels each Kraus class provides

Class Channels
Kraus1Q_174 DEP1_c, ZERR_c, LOSS_g, LOSS_m, DECAY_mg, LOSS_r, DECAY_rg, DECAY_rm
Kraus2Q_174174 DEP2_c, ZERR_c, LOSS_g, LOSS_m, DECAY_mg, LOSS_r, DECAY_rg, DECAY_rm
KrausRESET_174 LOSS_g_reset
KrausMEASURE_DISC_174 MERR
KrausMEASURE_174 LOSS_g_meas, LOSS_m, DECAY_mg, LOSS_r, DECAY_rg, DECAY_rm
Kraus1QClock_171m DEP1_gm, LOSS_g, LOSS_m, DECAY_mg, LOSS_r, DECAY_rg, DECAY_rm
Kraus1Q_171m DEP1_m, DECAY_mg_gate, ZERR_g, XERR_g, ZERR_m, XERR_m, ZERR_gm, LOSS_g, LOSS_m, DECAY_mg, LOSS_r, DECAY_rg, DECAY_rm
Kraus2Q_171m171m DEP2_m, ZERR_g, XERR_g, ZERR_m, XERR_m, ZERR_gm, LOSS_g, LOSS_m, DECAY_mg, LOSS_r, DECAY_rg, DECAY_rm
KrausRESET_171m LOSS_m_reset, FLIP_m
KrausMEASURE_DISC_171m MERR
KrausMEASURE_171m FLIP_g, LOSS_g_meas, ZERR_m, XERR_m, ZERR_gm, LOSS_m, DECAY_mg, LOSS_r, DECAY_rg, DECAY_rm

Mapping from the legacy channel names

For anyone cross-reading against results produced before this library existed. The paper’s conventions section explicitly avoids leakage as a way of classifying errors, which is why every *LEAK name became either LOSS_* or DECAY_*.

Legacy name Here
GLEAK LOSS_g
MLEAK LOSS_m
MGLEAK DECAY_mg
MGLEAK_gate DECAY_mg_gate
RLEAK LOSS_r
RGLEAK DECAY_rg
RMLEAK DECAY_rm
GZERR ZERR_g
MZERR ZERR_m
GMZERR ZERR_gm
GXERR (T1 idling) XERR_g
MXERR (T1 idling) XERR_m
GXERR_read FLIP_g
MXERR_prep FLIP_m
GLOSS_prep LOSS_g_reset
MLOSS_prep LOSS_m_reset
GLOSS_meas, GLOSS_read LOSS_g_meas
DEP1_gm (174 class) DEP1_c
DEP2_gm (174 class) DEP2_c
ZERR_gm (174 class) ZERR_c
XERR_mr removed — no such channel in the paper, and the legacy default was 0

Three legacy keys were ambiguous because the same string named different physical channels in different classes; the paper names disambiguate them:

The twirled measurement helpers on NoiseModelParameters return the loss probability under the key p_loss (legacy: p_leak), matching the paper’s E_loss channel.