Search papers, labs, and topics across Lattice.
This paper introduces Quantum Port-Hamiltonian Neural Networks (Q-pHNNs), a novel framework that learns classical dynamics while preserving structure through the Isomorphic Hamiltonian Mapping (IHM). By leveraging mid-circuit measurements for Measurement-Induced Nonlinearity (MINL), the Q-pHNNs ensure conservation and passivity are maintained inherently, rather than through penalty terms. Experimental results on systems like the nonlinear pendulum and damped harmonic oscillator reveal impressive performance, including a mere 1.35% relative energy drift and 100% energy monotonicity for the MINL circuit.
Achieving 100% energy monotonicity in quantum neural networks could redefine our approach to learning conservative and dissipative dynamics.
We introduce Quantum Port-Hamiltonian Neural Networks (Q-pHNNs), a family of parameterised quantum circuits that learn classical dynamics in a structure-preserving manner. The framework relies on the Isomorphic Hamiltonian Mapping (IHM): the skew-symmetric interconnection matrix $\mathbf{J}$ corresponds to unitary gate evolution, and the positive-semidefinite dissipation matrix $\mathbf{R}$ corresponds to Measurement-Induced NonLinearity (MINL) realised via mid-circuit measurement and classical feedforward. This ensures conservation and passivity are enforced by construction rather than penalty terms. We instantiate the IHM in four architectures: (1) a Quantum HNN that learns conservative energy manifolds and extracts Hamilton's equations exactly via the Parameter-Shift Rule; (2) a Q-pHNN using Born-rule measurement for dissipation; (3) a Q-pHNN jointly learning the energy ansatz and damping coefficient; and (4) a topology-entangled Quantum Graph Neural Network for $N$-node coupled-phasor networks. Experiments on the nonlinear pendulum and damped harmonic oscillator demonstrate: (i)~$1.35\%$ relative energy drift with a symplectic integrator and scale correction; (ii)~$100\%$ energy monotonicity for the MINL circuit; and (iii)~$12.1\%$ error in damping-coefficient identification from vector-field snapshots with no direct supervision on the damping coefficient.
