A Quantum Walk for Proton Therapy That Works Perfectly — and Is Still 160× Too Expensive to Use

First unitary quantum-walk oracle for charged-particle transport in radiotherapy. Grover amplification delivers near-quadratic speedup — but a brutal resource analysis shows clinical crossover is nowhere close.

Imagine you're playing a pinball machine where the ball loses speed predictably as it rolls through bumpers, but also wobbles randomly left and right. You could simulate that ball by flipping coins at every bumper — that's classical Monte Carlo. Or you could replace each coin flip with a quantum superposition, letting the ball explore all paths at once and then measure only the statistic you care about. That's the core mechanism here: a discrete-time quantum walk where a proton's energy register ticks down through a continuous-slowing-down model, with stochastic range straggling emerging from quantum coin dynamics rather than classical sampling. The committed claim: this paper constructs the first unitary transport oracle for proton energy loss that can be plugged directly into quantum amplitude estimation, and then does the honest arithmetic on whether it's actually useful. The oracle encodes energy-dependent stopping power into a multiplexed coin operator, with one fresh ancilla qubit per step and gate count linear in the number of energy levels. On the estimation side, maximum-likelihood QAE (MLQAE) targets a scalar statistic — the probability a proton penetrates past a distal depth threshold — rather than trying to reconstruct a full 3D dose map. The physics check is clean. The quantum-walk oracle reproduces a Bragg-peak depth-deposit profile that agrees with a classical MC simulation of identical rules to within 0.27%. Grover amplification matches the theoretical sin²((2k+1)θ) curve to machine precision up to k=32. At matched computational budget, MLQAE error scales as N^{-0.93} versus N^{-0.49} for classical sampling — a 4.8× error advantage at the largest budget tested. These are not approximate results; the oracle is exact within its own physics model. But then the paper does something unusual and admirable: it runs a three-gate resource analysis asking whether any of this matters at clinical scale. The answer is no, and the paper says so plainly. The oracle costs ~1.0×10⁶ logical gates per call (rising to ~3.5×10⁷ with rotation synthesis), dominated by the multiplexed coin operator. Against the gate budget required for wall-clock competitiveness with classical MC at clinical resolution, this is 160× to 5,400× too expensive. Full dose maps — the thing clinicians actually need — are "structurally unfavourable irrespective of hardware," meaning no amount of future error correction alone fixes the architectural mismatch between QAE's single-scalar output and per-voxel dose grids. The ladder positioning is honest. Classical Monte Carlo for proton therapy (codes like TOPAS, FLUKA, Geant4) runs billions of histories on commodity hardware in minutes. This paper doesn't claim to beat them. It claims to demonstrate that the quantum amplitude estimation framework can be correctly applied to charged-particle transport physics, and then to measure exactly how far the gap is. The 0.27% agreement with classical MC of identical rules confirms the oracle's physics fidelity, but the comparison is self-referential by design — the oracle IS the classical model, made unitary. The architecture belongs to the discrete-time quantum walk family — non-variational, gate-based, deterministic circuit structure. The key structural choice is encoding a continuous energy variable into a qubit register driven by a position-dependent multiplexed coin, which is elegant but expensive: the coin's controlled rotations dominate the gate count. This is a fundamentally different approach from variational quantum eigensolvers or quantum phase estimation for molecular energies; it's closer to quantum lattice Boltzmann methods but for particle transport rather than fluid dynamics. The milestone math is the most useful part of the paper for tracking purposes. The multiplexed coin must be compressed by roughly 2-3 orders of magnitude in gate cost to reach wall-clock crossover for the scalar statistic alone. Full dose maps require a qualitatively different output strategy, not just cheaper gates. The paper identifies the specific term responsible and reports the reduction factor needed, which gives future work a concrete target rather than a hand-wave.