Quantum state transfer performance of Heisenberg spin chains with site-dependent interactions designed using a generic genetic algorithm

Abstract Designing a good transfer channel for arbitrary quantum states in spin chains implies optimising a cost function, usually the averaged fidelity of transmission. The fidelity of transmission measures how much the transferred state resembles the state prepared at the beginning of the transfer protocol. When averaged over all the possible initial states, the figure of merit quantifies the quality of the protocol. There are proposals for optimising a given Hamiltonian to accomplish a particular task. The transfer of quantum states is one of them. In particular, we consider the design of Heisenberg spin chains using a genetic algorithm. This very efficient algorithm allows us to study different properties of Hamiltonians with good to excellent transfer ability. Using an evolutionary population method results in exchange coefficient strengths that change abruptly from site to site, which could hinder implementing an actual physical system with such exchange coefficients. By modifying the cost function, we obtain Hamiltonians with exchange coefficients varying smoothly along the chain length without compromising their transfer ability concerning the rough ones. Besides, both kinds of chain Hamiltonians show similar robustness against static disorder. By studying the statistical properties of the eigenvalues of Hamiltonians with varying transfer abilities, we determine the ensemble of random matrices to which the spectra belong.