Figure 10
From: Practical overview of image classification with tensor-network quantum circuits
Left: The simulation time of an MPS circuit increases exponentially with the number of bond qubits, regardless of the total number of qubits. For this data, we used 16-qubit blocks. Note that the total number of qubits can vary slightly as the number of bond qubits changes. The MPS shape dictates the variation in qubit numbers, e.g., an MPS circuit with 16-qubit blocks and two bond qubits per block can only result in circuits with \(16+14n\) qubits, where n is a positive integer. Middle: At a constant number of bond qubits and five block qubits, the simulation time increases linearly with the total number of qubits. Right: For a constant total circuit size of 100 qubits, increasing the size of the tensor blocks initially reduces the simulation time and then increases it. This is an artifact of how the tensor blocks are defined. Initially, increasing the number of block qubits reduces the total number of circuits to simulate during circuit cutting. However, as the size of the blocks increases, the time gained by having larger circuits surpasses the time saved by having fewer circuits. All simulations are performed on a personal laptop computer with 16 GB of RAM and a four-core i7-1185G7 processor operating at 3.00GHz.
