We present an efficient algorithm to reduce the number of non-Clifford gates in quantum circuits and the number of parametrized rotations in parametrized quantum circuits. The method consists in finding rotations that can be merged into a single rotation gate. This approach has already been considered before and is used as a pre-processing procedure in many optimization algorithms, notably for optimizing the number of Hadamard gates or the number of T gates in Clifford+T circuits. Our algorithm has a better complexity than similar methods and is particularly efficient for circuits with a low number of internal Hadamard gates. Furthermore, we show that this approach is optimal for parametrized circuits composed of Clifford gates and parametrized rotations with non-repeated parameters. For the same type of parametrized quantum circuits, we also prove that a previous procedure optimizing the number of Hadamard gates and internal Hadamard gates is optimal. This procedure is notably used in our low-complexity algorithm for optimally reducing the number of parametrized rotations.
This paper presents a study of the influence of the quantum phase on the qubit state in circuits with Hadamard gates in series for further minimisation in software or hardware modelling. The study is carried out using the Azure Quantum computing platform to simulate quantum algorithms and run quantum applications on dedicated hardware. The study provides an overview of the qubit as a unit of quantum information, describes in detail the operation of the Hadamard gate, simulates the operation of a circuit of two Hadamard gates in series with initial inputs and , and simulates the operation of a circuit with a sequential shift on and Hadamard gate inputs and . In the course of the study, it was found how the same intermediate state of the qubit can have different outputs after the Hadamard gate, by means of phase. Also, the results of the schemes where the first Hadamard gate is replaced by a sequential shift on . The results obtained can be used to test the Hadamard gate model when creating its digital model. Key words: Hadamard gate, quantum computing, quantum gates, superposition.
Conventionally, the quantum oracular algorithms, such as Deutsch-Jozsa, Bernstein-Vazirani, and Grover, utilize Hadamard gates to create uniform superposition states for the input qubits of an oracle. However, Hadamard gates are non-native (non-supported) gates in all real quantum computers. For this reason, Hadamard gates are considered cost-expensive gates when realizing (transpiling) such algorithms into a real quantum computer. This paper introduces a new methodology for cost-effective transpilation of Grover's algorithm into real quantum computers, by replacing all Hadamard gates with X gates. In quantum computing, the Hadamard and X gates create uniform superposition states of a qubit on the X-axis and Y-axis of the Bloch sphere, respectively. Hence, we utilize a different axis of the Bloch sphere to construct a cost-effective realization for the final transpiled circuit of Grover's algorithm. Theoretically, the final transpiled circuit of Grover's algorithm using X gates has been minimized to approximately 64% than the circuit of Grover's algorithm using Hadamard gates. Experimentally, because of the X is an IBM native gate, it has been proven that the final transpiled circuit of Grover's algorithm using X gates always has a lower quantum cost of 6% than the circuit of Grover's algorithm using Hadamard gates, due to the limited layout (architecture) connectivity of the physical neighboring qubits for IBM quantum computers.