Quantic has got a new paper out! arXiv:2102:04032. In this work we explore how functions can be stored within a quantum circuit
We want to make a single-qubit quantum circuit to behave as a target function. For doing so, we construct a circuit as a series of single-qubit gates depending on a independent variable x and some parameters. The parameters are defined by the target function and found with an optimizer
Depending on how the gates are defined, this is equivalent to a Fourier series or a Neural Network! And it is possible to fit single- or multi-variable real and complex functions!
We have compared classical models, quantum simulations and an actual experiment to numerically check that our model can fit arbitrary functions
The more layers are added to the model, the better approximations we get
This work proves that single-qubit circuits suffice to represent functions, but what would happen when more qubits are added? We think that the best is yet to come!