Qubit Number Optimization for Restriction Terms of QUBO Hamiltonians

In usual restriction terms of the Quadratic Unconstrained Binary Optimization (QUBO) hamiltonians, a integer number of logical qubits R, called the Integer Restriction Coefficient (IRC), are forced to stay active. In this paper we gather the well-known methods of implementing these restrictions, as well as some novel methods that show to be more efficient in some frequently implemented cases. Moreover, it is mathematically allowed to ask for fractional values of R. For these Fractionary Restriction Coefficients (FRC) we show how they can reduce the number of qubits needed to implement the restriction hamiltonian even further. Lastly, we characterize the response of DWave's Advantage_system4.1 Quantum Annealer (QA) when faced with the implementation of FRCs, and offer a summary guide of the presented methods and the situations each of them is to be used.