Skip to content

Benchmarks

The following benchmarks demonstrate the performance of the pytket-custatevec backend compared to other statevector simulators within the pytket ecosystem.

Comparison Scope

These tests compare pytket-custatevec against pytket-qiskit and pytket-qulacs.

All simulations are performed via the abstract pytket Backend interface to demonstrate the speedup available to users simply by switching their backend instance.

Methodology

The Circuit

We use a Random Line Circuit to ensure a fair stress test. This structure creates a highly entangled state that is difficult to simplify or optimize trivially.

  • Structure: Linear connectivity (1D chain).
  • Depth: Fixed at 10 layers.
  • Gates: Random TK1 (single qubit) gates followed by CX (two-qubit) gates between neighbors.
def random_line_circuit(n_qubits: int, layers: int) -> Circuit:
    c = Circuit(n_qubits)
    for i in range(layers):
        # 1. Random Single Qubit Rotations
        for q in range(n_qubits):
            c.TK1(np.random.rand(), np.random.rand(), np.random.rand(), q)

        # 2. Linear Entanglement (Alternating Even/Odd links)
        offset = np.mod(i, 2)
        qubit_pairs = [[c.qubits[i], c.qubits[i+1]] for i in range(offset, n_qubits-1, 2)]

        for pair in qubit_pairs:
            c.CX(pair[0], pair[1])
    return c

1. Statevector Simulation

This benchmark measures the time required to calculate the full statevector (\(2^n\) complex amplitudes).

2. Shot-based Sampling

This benchmark measures the time to simulate measurement outcomes (1000 shots) for the same random circuits.

Environment

The results above were auto-generated by our CI pipeline using the following environment:

Component Specification / Version
GPU Tesla T4
CPU Intel(R) Xeon(R) Platinum 8259CL CPU @ 2.50GHz
Python 3.12.11
OS Linux 6.8.0-1024-aws
pytket 2.11.0
pytket-custatevec 0.0.1
pytket-qulacs 0.41.0
pytket-qiskit 0.74.0
cuquantum-python N/A

Reproduce this Benchmark

You can find the source code for these benchmarks in our repository:

View Benchmark Script