Intelligent unitary synthesis for quantum computing

Systems and techniques that facilitate intelligent unitary synthesis for quantum computing are provided. For example, a system can access a unitary matrix of a quantum payload circuit that fails to satisfy design constraints of a quantum computer’s architecture, and synthesize the unitary matrix into a transpiled version that conforms to those architectural requirements. The synthesis leverages deep learning initialization of adjustable parameters of quantum circuit templates: rather than randomly selecting circuit templates or randomly initializing their parameters, the system uses neural networks to intelligently choose templates and set starting parameter values, reducing processing time and helping avoid local minima entrapment during gradient descent optimization.