System Model H2 Emulators¶
Default Noise Model Parameters¶
Default Settings |
H2-1E |
General |
|
Qubits (Statevector) |
32 |
Qubits (Stabilizer) |
56 |
Connectivity |
all-to-all |
Parallel 2-qubit operations |
4 |
Wasm Enabled |
True |
Physical Noise |
|
1-Qubit Fault Probability (p1) |
2.9e-05 |
2-Qubit Fault Probability (p2) |
0.00128 |
Bit-Flip Measurement Probability (0 outcome) (p_meas) |
0.0005 |
Bit-Flip Measurement Probability (1 outcome) (p_meas) |
0.0025 |
Crosstalk Measurement Fault Probability (p_crosstalk_meas) |
7.4e-06 |
Initialization Fault Probability (p_init) |
4e-05 |
Crosstalk Initialization Probability (p_crosstalk_init) |
9.6e-06 |
Ratio of 1-Qubit Spontaneous Emission to p1 (p1_emission_ratio) |
0.32 |
Ratio of 1-Qubit Spontaneous Emission in 2-Qubit Gate to p2 (p2_emission_ratio) |
0.48 |
Dephasing Noise |
|
Quadratic Dephasing Rate (quadratic_dephasing_rate) |
0.043 |
Linear Dephasing Rate (linear_dephasing_rate) |
0.0028 |
Coherent to Incoherent Factor (coherent_to_incoherent_factor) |
2.0 |
Arbitrary Angle Noise Scaling |
|
Fit Parameter 1 (przz_a) |
1.518 |
Fit Parameter 2 (przz_b) |
0.241 |
Fit Parameter 3 (przz_c) |
1.518 |
Fit Parameter 4 (przz_d) |
0.241 |
Polynomial (przz_power) |
1.0 |
Performance¶
The performance of the System Model H2 Emulator is measured in comparison to the H2 hardware. With the inclusion of an accurate and up-to-date noise model, the System Model H2 Emulator provides a detailed representation of System Model H2 output, operating on a GPU backend. Because of the memory requirements, a slowdown in the number of H-System Quantum Credits (HQCs) processed in an hour is observed for state vector emulations using 25 qubits or more. The difference in processing time is due to the higher number of qubits combined with the realistic physical noise model of System Model H2 performing a state vector emulation. To estimate the amount of time the emulation will take for state vector emulation, users can use the guideline in Figure 1 in conjunction with the total HQCs for their quantum program. It is generally recommended that users perform state vector emulations with less than 28 qubits. Though it is dependent on the submitted circuit, the execution speed of the System Model H2 hardware is generally higher than the emulator with qubit counts at 28 qubits or more.
Accuracy is verified at Quantinuum by comparing emulator and hardware outputs. Users should be aware that System Model H2 has a more complex trapping geometry than System Model H1. In comparing the H2 emulation to H2 quantum computer results, users may experience a larger discrepancy than experienced comparing System Model H1 emulators to H1 quantum computers. System Model H2 error models will mature over time. In the case of exceptional or unexplained variance, users should contact Quantinuum technical support at QCsupport@quantinuum.com to discuss the circuit and results.
Job Execution¶
Jobs submitted with large shot counts are automatically divided into appropriately sized chunks of smaller shot counts in a method called chunking. Chunking allows for a more incremental distribution of emulation resources to users and an increased flexibility in the management of long running jobs. The number of shots in a chunk is dynamically chosen by the runtime and varies with the complexity of the circuit.