Statistical regimes of electromagnetic wave propagation in randomly time-varying media

Basic Information

• Paper ID: 2507.11062
• Title: Statistical regimes of electromagnetic wave propagation in randomly time-varying media
• Authors: Seulong Kim (Ajou University), Kihong Kim (Ajou University & Korea Institute for Advanced Study)
• Classification: physics.optics, cond-mat.dis-nn
• Publication Date: September 3, 2025
• Paper Link: https://arxiv.org/abs/2507.11062

Abstract

This paper investigates the statistical characteristics of electromagnetic wave propagation in randomly time-varying media. Time-varying media break energy conservation through temporal modulation, enabling unique energy transport control. The study identifies three distinct statistical regimes for unidirectional input waves: early-time gamma distribution, mid-time exponential distribution, and long-time quasi-lognormal distribution. Symmetric bidirectional input produces genuine lognormal statistics across all timescales. These findings are verified through two complementary disorder models, demonstrating that the observed statistical properties are robust and governed by input symmetry.

Research Background and Motivation

1. Problems to be Addressed:
   • The statistical behavior of electromagnetic wave propagation in time-varying media remains incompletely understood
   • Existing research suggests energy distribution evolves from exponential to lognormal forms, but critical questions persist: What controls transitions between these regimes? Is lognormal behavior truly universal? How do initial wave conditions affect statistical evolution?
2. Problem Significance:
   • Time-varying systems are increasingly important in modern optics, materials science, and communication technologies
   • Temporal modulation enables dynamic functionalities beyond the capabilities of static systems
   • Temporal disorder can induce strong localization effects with potential applications in ultrafast switching, energy harvesting, and time-gated information processing
3. Limitations of Existing Approaches:
   • Previous research focused primarily on exponential and lognormal statistics but lacked deep understanding of transition mechanisms
   • Neglected the influence of initial wave configuration on statistical evolution
   • Lacked a unified theoretical framework for understanding statistical wave dynamics in time-varying systems
4. Research Motivation: Establish a comprehensive theoretical framework for statistical behavior of wave propagation in time-varying media and provide design principles for dynamically tunable photonic and electromagnetic devices.

Core Contributions

1. Identification of Three Distinct Statistical Regimes: For unidirectional input, early-time gamma distribution, mid-time exponential distribution, and long-time quasi-lognormal distribution are identified
2. Revelation of Initial Conditions' Critical Role: Demonstrates that symmetric bidirectional input produces genuine lognormal statistics, while unidirectional input produces quasi-lognormal distribution
3. Establishment of Unified Theoretical Framework: Uses invariant embedding method to derive exact moment equations, providing a complete analytical framework
4. Verification of Statistical Behavior Universality: Validates result robustness through two different disorder models
5. Connection Between Momentum Conservation and Statistical Behavior: Establishes linkage between momentum conservation and long-time statistical behavior

Methodology Details

Task Definition

Study the statistical characteristics of electromagnetic wave propagation in isotropic, spatially homogeneous media with randomly time-varying dielectric constants. Input consists of initial electromagnetic waves (unidirectional or bidirectional), with output being statistical distributions of reflectance, transmittance, and total energy density.

Model Architecture

1. Fundamental Wave Equation: The displacement field D satisfies the wave equation: $\frac{d}{dt}\left(\mu\frac{dD}{dt}\right) + \frac{\omega_0^2}{\epsilon}D = 0$

where $\omega_0 = ck$, and $\epsilon$ and $\mu$ are the dielectric constant and permeability, respectively.

2. Invariant Embedding Method (IEM): Transforms boundary value problems into initial value problems, yielding coupled differential equations for reflection and transmission amplitudes: $\frac{1}{\omega_0}\frac{dr}{d\tau} = i\beta r + i\alpha s$$\frac{1}{\omega_0}\frac{ds}{d\tau} = -i\alpha r - i\beta s$

where: $\alpha = \frac{1}{2n^2}\left(\frac{\epsilon_2}{\epsilon} - \frac{\mu_2}{\mu}\right), \quad \beta = \frac{1}{2n^2}\left(\frac{\epsilon_2}{\epsilon} + \frac{\mu_2}{\mu}\right)$

3. Moment Equation Derivation: Define moment functions $Z_{abcd} = \langle r^a(r^*)^b s^c(s^*)^d \rangle$ and apply the Furutsu-Novikov formula to derive their evolution equations: $\frac{1}{\omega_0}\frac{d}{d\tau}Z_{abcd} = \sum_{i=1}^{13} C_i Z_{a'b'c'd'}$

Technical Innovations

1. Extension of Invariant Embedding Method: Successfully extends IEM, originally developed for spatially inhomogeneous systems, to temporally disordered systems
2. Exact Moment Equations: Derives coupled systems of equations with deterministic coefficients using stochastic calculus tools
3. Multi-Timescale Analysis: Identifies and analyzes three distinct temporal regions with corresponding statistical distributions
4. Initial Condition Dependence: Discovers and explains sensitive dependence of statistical behavior on initial wave symmetry

Experimental Setup

Datasets

Model 1: Delta-correlated Gaussian Noise$\epsilon(t) = \bar{\epsilon} + \delta\epsilon(t)$$\langle\delta\epsilon(t)\delta\epsilon(t')\rangle = g_0\delta(t-t'), \quad \langle\delta\epsilon(t)\rangle = 0$

Model 2: Piecewise Constant Disorder$\delta\epsilon$ remains constant within each time interval of duration $\Lambda = 1/\omega_0$, independently sampled from uniform distribution $[-a_0, a_0]$.

Evaluation Metrics

• Higher-order moments of reflectance and transmittance: $\langle R^n \rangle$, $\langle S^n \rangle$
• Total energy density moments: $\langle U^n \rangle$
• Distribution shape parameters: $F_3 = \langle R^3 \rangle \langle R \rangle^3 / \langle R^2 \rangle^3$
• Logarithmic moments: $\langle \ln U \rangle$, $\text{Var}(\ln U)$

Implementation Details

• Weak disorder approximation: $g = 0.003$
• Time range: $0 < \omega_0 t \lesssim 10^6$
• Model 2 parameters: $\Lambda = 1/\omega_0$, $a_0 = 0.1$
• Statistical samples: $10^6$ independent realizations

Experimental Results

Main Results

1. Three Statistical Regimes:

• Early-time ($\omega_0 t \lesssim 1$): Gamma distribution $\langle R^n \rangle = (2n-1)!! \left(\frac{g\omega_0 t}{4}\right)^n$
• Mid-time ($1 \lesssim \omega_0 t \lesssim g^{-1}$): Exponential distribution $\langle R^n \rangle = n! \left(\frac{g\omega_0 t}{4}\right)^n$
• Long-time ($g^{-1} \lesssim \omega_0 t$): Quasi-lognormal distribution $\langle U^n \rangle = \frac{n!}{(2n-1)!!} \exp\left[n(n+1)\frac{g\omega_0 t}{4}\right]$

2. Influence of Initial Conditions: Symmetric bidirectional input ($R(0) = S(0) = 1/2$) produces genuine lognormal statistics: $\langle U^n \rangle = \exp\left[n(n+1)\frac{g\omega_0 t}{4}\right]$

Ablation Studies

The evolution of distribution shape parameter $F_3$ verifies the existence of three regimes:

• Gamma distribution: $F_3 = 5/9 \approx 0.556$
• Exponential distribution: $F_3 = 0.75$
• Quasi-lognormal: $F_3 = 1.35$
• True lognormal: $F_3 = 1$

Case Studies

Numerical simulations of Model 2 completely validate theoretical predictions, demonstrating that statistical behavior is independent of specific disorder models and depends only on short-range correlations and weak disorder conditions.

Experimental Findings

1. Quasi-lognormal Distribution: Long-time behavior is not genuine lognormal but quasi-lognormal with modified prefactors
2. Momentum Conservation Constraint: The relationship $S - R = 1$ directly reflects momentum conservation and constrains long-time statistical behavior
3. Universality: Statistical behavior exhibits universality across different disorder models

Related Work

Major Research Directions

1. Wave Propagation in Time-Varying Media: Temporal photonic crystals, momentum band gaps, temporal Brewster effect
2. Temporal Disorder and Localization: Analogous studies of temporal Anderson localization
3. Statistical Wave Dynamics: Wave statistics theory in spatially disordered systems

Advantages of This Work

• Provides complete three-regime statistical theory, surpassing previous two-regime descriptions
• Reveals the decisive influence of initial conditions on statistical behavior
• Establishes exact analytical framework, avoiding computational burden of numerical averaging

Conclusions and Discussion

Main Conclusions

1. Wave propagation in time-varying disordered media exhibits three distinct statistical regimes
2. Initial wave symmetry is the key factor determining long-time statistical behavior
3. Momentum conservation constrains statistical evolution through the relationship $S - R = 1$
4. Quasi-lognormal distribution is mathematically distinct from genuine lognormal distribution

Limitations

1. Infinite Medium Assumption: Neglects reflection effects from spatial boundaries
2. Weak Disorder Restriction: Qualitatively different statistical behavior may emerge in strong disorder regimes
3. Short-Range Correlation Assumption: Long-range correlated disorder remains unconsidered

Future Directions

1. Study finite-size effects, particularly when medium length is comparable to wavelength
2. Extend to strong disorder regimes using bounded binary disorder models
3. Consider effects of long-range correlated disorder
4. Experimental verification using metamaterials, temporally modulated dielectrics, or plasmas

In-Depth Evaluation

Strengths

1. Theoretical Innovation: First complete description of three-regime statistical behavior in time-varying disordered media
2. Advanced Methodology: Clever application of invariant embedding method, transforming stochastic boundary value problems into deterministic initial value problems
3. Profound Results: Discovery of quasi-lognormal distribution as a new statistical form, correcting previous understanding
4. Sufficient Verification: Consistent results from two different disorder models demonstrate theoretical universality

Weaknesses

1. Limited Applicability: Restricted to weak disorder and short-range correlations
2. Absence of Experimental Verification: Theoretical results lack experimental validation
3. Incomplete Physical Insight: Physical mechanisms generating quasi-lognormal distribution require deeper explanation

Impact

1. Academic Contribution: Provides new theoretical framework for time-varying media physics
2. Application Value: Offers design principles for dynamically tunable photonic devices
3. Methodological Significance: Demonstrates powerful application potential of invariant embedding method in random systems

Applicable Scenarios

1. Ultrafast optical switch design
2. Time-gated information processing systems
3. Dynamically reconfigurable photonic devices
4. Energy harvesting and localization applications

References

The paper contains abundant references covering important works in time-varying media physics, Anderson localization theory, invariant embedding methods, and related fields, providing readers with comprehensive background knowledge and directions for further research.