Tunnel or Climb

A quantum walker and a classical hill-climber race across the same convex landscape, but there's a narrow spike planted in the slope between them and the goal. Watch what happens as you crank the barrier up.
Grounded in "Log-concavity and tunneling: adiabatic quantum optimization for convex functions (with a spike)" — Braida, Bermot & Apers, arXiv:2606.23614 (quant-ph). The paper proves that for 1D discrete Schrödinger operators whose ground state is log-concave — a class that includes convex potentials — the spectral gap stays large even when a narrow, tall spike is stuck into an otherwise smooth bowl. That surviving gap is exactly what lets adiabatic quantum search tunnel past the spike quickly, extending Reichardt's 2004 "Hamming weight with a spike" analysis from an exactly-solvable linear bowl to the harder quadratic (non-exactly-solvable) case. This demo turns that spectral-gap advantage into something you can watch: a real finite-difference (Crank–Nicolson) evolution of a wavepacket on a discretized 1D bowl-plus-spike potential, raced against a Metropolis/simulated-annealing hill-climber on the identical landscape.

The Race cycle 1

classical climbers (Metropolis / annealing) quantum probability density |ψ(x)|² goal (potential minimum) the spike

Success Rate vs. Spike Height

Orange = classical success rate (fraction of climbers that reach the goal before giving up). Blue = quantum tunneling probability (peak probability mass that reaches the goal region). The vertical marker tracks your current spike-height slider. Narrow the spike and watch the blue curve resist collapse far longer than the orange one — that gap between the curves is the spectral-gap advantage.

Live Readout

--%
classical success
--%
quantum tunneling

Controls