arXiv:2610.35828
Interference-induced dynamical crossovers in quantum walks on disordered square lattices
Ramnarayan Bera, Arijit Roy, Soumyadip Ghosh, Soumen Roy
Abstract
We study discrete-time quantum walks with Fourier, Grover, and Hadamard coin operators on disordered square lattices, alongside classical random walks. Edges in the lattice are absent with a probability, $p$. In this case, both classical and quantum walks exhibit an essentially identical degradation of transport. We then study a general version of this model, in which the existing edges are retained. However, the absent edges are no longer absent, but now become partially transmitting, with a transmission coefficient $f$. We find that the response is now quite sensitive to the walk protocol and coin symmetry. The classical walk exhibits asymptotic stability of the anomalous diffusion exponent, $β(t)$. In comparison, quantum walks exhibit a monotonically decaying $β(t)$ beyond the initial transient regime, leading to the absence of quantum diffusion in two dimensions. We map the transport regimes in the $(f,p)$ plane and identify the contour for $β=1$, which separates the subdiffusive and superdiffusive regimes. We also examine the effective diffusion coefficient, the inverse participation ratio, and the first two moments of edge transmission. We restrict ourselves to analyzing bulk propagation prior to finite-size boundary reflections. The results indicate that quantum transport depends significantly on coin-specific interference and cannot be fully described by transmission statistics alone.
AI Summary
The paper investigates how static bond disorder affects transport in two-dimensional discrete-time quantum walks, and whether partial transmission exposes interference effects obscured by complete bond removal. It compares Fourier, Grover, and tensor-Hadamard coins with a classical random walk whose directional transition probabilities are $1/4$. Each square-lattice bond is defective independently with probability $p$. Pristine bonds transmit perfectly; defective bonds have transmission probability $f$ and reflection probability $1-f$, implemented quantum mechanically by a unitary beam splitter with relative reflection phase $i$. Numerical diagnostics include the mean-squared-displacement exponent $\beta(t)=d\ln\sigma^2/d\ln t$, an effective diffusion coefficient, and the spatial inverse participation ratio (IPR), evaluated before boundary reflections. Complete bond removal ($f=0$) is reported to degrade classical and quantum transport similarly. With partial transmission, transport becomes coin dependent: the classical exponent approximately plateaus, whereas quantum exponents generally decrease over the observed interval. Decreasing exponents together with IPR plateaus indicate suppressed spreading. Maps in $(f,p)$ identify finite-time $\beta=1$ crossovers between subdiffusive and superdiffusive behavior; the extracted transmission threshold peaks near $p\approx0.50$. Mean and variance of bond transmission do not fully describe the quantum response. The paper explicitly distinguishes these crossovers from thermodynamic phase transitions and states that finite-time diagnostics do not establish asymptotic Anderson localization. Its finite-size check compares $L=64,128,256$ at $f=0.50$, fitting exponents over $t=15$–$30$; longer simulations and additional fitting windows are needed to assess asymptotic behavior.
Central claim may not hold: The abstract infers absence of quantum diffusion from finite-time decreasing transport exponents, although those observations do not exclude a later approach to ordinary diffusion. The abstract claims asymptotic stability of the classical transport exponent, while the reported evidence establishes only approximate stability within the simulated interval. The conclusion treats finite-time exponent decay and IPR saturation as demonstrating exclusion of a stable quantum-diffusive phase and progression toward localization, contradicting the explicit limitations acknowledged in the results.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 10 pages, 7 figures