Variational Computing: The Energetic Origin of Boolean Logic

We define computation as (i) descent on an energy landscape and (ii) symbolic readout by basins or thresholds. We show that a robust NAND gate can be realized within a variational computing system $(\mathcal{X}, E, \Phi, \mathcal{R})$via a robustness margin that guarantees threshold-invariant readout under $\ell_\infty$ perturbations. The NAND construction is purely existential: it demonstrates that suitable energy landscapes can encode Boolean logic. We briefly note Hopfield recall and PINN-style constraint training as further instances,demonstrating that energy specifies, descent executes, and readout symbolizes.