A new quantum similarity estimation algorithm proposes combining discrete and continuous encoding in a single circuit, aiming to compare quantum states without reconstructing them in full. The technique is presented as a tool for precision measurements in atomtronics, the field that uses ultra-cold atomic gases confined in optical traps as analogues of electronic circuits.

The underlying problem is that estimating how similar two quantum states are usually requires full tomography, a process that grows exponentially with the number of qubits and is unworkable for systems of practical size. Earlier work has explored fidelity measures and quantum kernels, but how the data are encoded — classical or quantum — determines both the precision and the cost of the circuit. The hybrid proposal seeks to distribute the information between discrete registers and continuous variables to lighten that load.

The method builds a circuit that prepares a controlled superposition of the two states to be compared and applies phase estimation adapted to the mixed encoding. The discrete part stores state labels, while the continuous part captures analog parameters of the physical system, such as amplitudes or displacements in phase space. The final readout returns an estimate of the similarity with an error that depends on the number of circuit repetitions and on the resolution of the continuous encoding.

The reference application is atomtronic metrology: distinguishing configurations of cold atoms in optical traps or lattices with enough precision to detect small external perturbations. The paper places the algorithm within the framework of quantum sensors, where efficient comparison between prepared and reference states is a step prior to calibration. The practical advantage would lie in avoiding full tomography and in exploiting both digital and analog degrees of freedom of the available hardware.

Testing performance on real cold-atom devices and benchmarking it against established fidelity estimation protocols remains pending. The scale at which the hybrid encoding keeps its edge over tomography is still an open question, as is its behaviour against decoherence and control noise in experimental implementations.