A feasible approach for automatically differentiable unitary coupled-cluster on quantum computers.

Area of Science:

• Quantum computing
• Computational chemistry
• Quantum algorithms

Background:

• Unitary coupled-cluster (UCC) operators are crucial for quantum chemistry simulations.
• Evaluating gradients of UCC operators on quantum computers is computationally expensive.
• Current methods often require a high number of expectation value measurements.

Purpose of the Study:

• To develop computationally affordable and encoding-independent gradient evaluation procedures for UCC operators.
• To reduce the cost associated with gradient evaluation in quantum algorithms.
• To facilitate the construction of differentiable objective functions for quantum computations.

Main Methods:

• Developed novel gradient evaluation strategies for parameterized fermionic excitations.
• The proposed framework requires evaluating four expectation values, reduced to two for real wavefunctions.
• Implemented strategies within the open-source Tequila package for ease of use.

Main Results:

• Demonstrated a significant reduction in the number of required quantum measurements compared to standard parameter-shift-rule methods.
• The method is encoding-independent, offering flexibility in quantum circuit design.
• Successfully applied the strategies to adaptive approaches for electronic ground and excited states.

Conclusions:

• The developed gradient evaluation procedures offer a computationally efficient alternative for UCC operators on quantum computers.
• This work lowers the barrier for applying UCC methods in quantum chemistry and related fields.
• The Tequila implementation enables practical application and further development of quantum machine learning techniques.