H-Series Emulators¶
To support quantum algorithm development and design, emulators are available that model each machine’s specific ion transport and error rates. The emulators operate on a physical model as well as include a detailed error model of each H-Series machine. In addition, options are provided to the user to experiment with the noise parameters of the emulator.
For complete details on the performance and specification of the emulators, see the Emulator Product Data Sheets found on these pages:
Use Cases¶
H-Series emulators provide high-fidelity emulation of H-Series machines. Use cases include:
Debugging of quantum code before running on physical hardware
Optimization of quantum code in the presence of noise mechanisms
Exploring new algorithms and techniques for quantum error correction
Introduction to H-Series’ unique differentiating capabilities such as qubit reuse after mid-circuit measurement, all-to-all connectivity, and high-fidelity gates
Emulator Access and Output¶
Communication with H-Series emulators occurs through an API endpoint based on the OpenQASM 2.0 standard [1]. Interface details are given in the Quantinuum Application Programming Interface (API) Specification.
Users can select a H-Series emulator from the machine list API, designated with the E suffix machine name. The output of a H-Series Emulator is identical to the output format of H-Series quantum computers. Through the Job Submission API, users may select the type of emulator used and turn on or off the application of the error model.
Features¶
Common features across all H-Series emulators:
TKET supported in the stack provides circuit optimization to all submitted circuits.
OPENQASM 2.0 circuits
Quantinuum QASM enhancements, including classical logic, math, and program flow control
Common compound gates from OPENQASM library, e.g., CX, H
User-defined compound gates
High fidelity noise models and parameters closely mimicking H-Series hardware performance. Each emulator uses the same physical noise model, but noise parameters reflect the performance of the device being emulated.
Uses identical API for job submission as H-Series, enabling seamless translation from emulator to hardware
Uses identical compiler as H-Series, containing all the native gates, transport operations and classical operations used in H-Series
Provides identical output format as H-Series quantum computers
Allows usage of H-Series attributes: all-to-all connectivity and qubit reuse after mid-circuit measurement
Available even while H-Series quantum computers are offline to enable maximized productivity and development time
Large quantum circuits with a limit of 10,000 on the number of shots
Identical queuing prioritization as H-Series quantum computers
Emulation Method¶
The H-Series emulators, accessible via the API, receive instructions directly from the same compilers used by the physical hardware. These compilers translate the submitted quantum program into a set of instructions comprising the native gate operations and the transport operations necessary to reconfigure the ion trap at each step of the program. Users can choose between either a state vector or stabilizer emulation method; in both cases results are performed shot-by-shot. The state vector emulation method can run any general quantum circuits, while the stabilizer emulation method is restricted to circuits involving only quantum unitary gates that are Clifford operations.
For System Model H1, state vector emulator is supported up to all 20 qubits. For System Model H2, state vector emulation is supported up to 32 qubits, with stabilizer emulation supported up to 56 qubits.
The error model for the emulation can be turned on or off, allowing noisy or noise-free emulations, respectively. The emulated error model includes:
Asymmetric depolarizing gate noise
Leakage errors
Crosstalk noise
Dephasing noise due to transport and qubit idling
Except for dephasing, errors on physical qubits are modeled as stochastic processes. For the state vector emulation, dephasing is handled as a coherent \(Z\) rotation according to a dephasing rate and the duration the qubit spends in transport or while idling while other qubits are being gated. For the stabilizer emulation, the dephasing noise is treated as a stochastic Pauli-\(Z\) error where the probability of a Pauli-\(Z\) error is equal to the Pauli twirled approximation of the coherent dephasing channel, which is proportional to the square of the dephasing rate multiplied by the duration.