Dimension-dependent continuum limits in tissue mechanics

Technical Summary: Dimension-dependent continuum limits in tissue mechanics

Problem Statement
Discrete, cell-based models effectively link microscopic forces to tissue-scale evolution but are often computationally intractable for large cell numbers and difficult to parameterize. Continuum models offer a practical alternative for population-level inference and revealing emergent transport links. However, while 1D continuum limits for epithelial tissue mechanics are well-studied, the corresponding higher-dimensional problems remain largely unexplored. A critical gap exists in understanding how spatial dimensionality influences the link between cell-scale forces and tissue-scale transport, particularly regarding nonlinear diffusivities of the porous-media type, $D(\rho) \propto \rho^\gamma$. Most existing phenomenological applications in fluid mechanics and population biology are limited to $\gamma > 0$, whereas epithelial mechanics motivates both $\gamma > 0$ and $\gamma < 0$. The authors investigate whether the macroscopic transport exponent $\gamma$ is determined solely by microscopic mechanics or if it is fundamentally altered by spatial dimensionality.

Methodology
The authors employ a strategy of matching long-time exponential relaxation rates between discrete mechanical models and macroscopic partial differential equation (PDE) models near a steady state.

1. Discrete Models:

   • 1D: The tissue is modeled as a chain of $N$ mechanical links (cells) with fixed boundaries. Nodes evolve according to an overdamped model driven by an interaction force law $f(\ell)$. The authors utilize a parameterized family of force laws, $f(\ell) = k(\ell^\nu - a^\nu)/(\nu a^{\nu-1})$ (with a logarithmic limit for $\nu=0$), which linearizes to Hooke's law near the resting length $a$.
   • 2D: Cells are modeled as a triangulation network where nodes interact via pairwise mechanical forces. The dynamics are governed by a similar overdamped equation summing forces over adjacent nodes.
   • Analysis: The authors linearize the discrete systems about their equally-spaced steady states. The late-time relaxation is governed by the leading eigenvalue of the system's Jacobian (or Laplacian matrix in 2D), yielding an exponential decay rate.

2. Macroscopic Models:

   • The tissue density $\rho(\mathbf{r}, t)$ is described by a nonlinear diffusion equation: $\partial_t \rho = \nabla \cdot (D(\rho) \nabla \rho)$.
   • The authors assume a general form for the nonlinear diffusivity: $D(\rho) = a^\beta \rho^\gamma / \tau$, where $a$ is the microscopic length scale, $\rho$ is density, and $\tau$ is a characteristic time scale.
   • By linearizing the PDE near the steady state $\bar{\rho}$, the authors derive that the relaxation rate depends on $D(\bar{\rho})$ and the domain geometry (specifically the eigenvalues of the Laplacian).

3. Coarse-Graining Procedure:

   • The core method involves equating the relaxation rates derived from the discrete simulations (dependent on microscopic parameters $k, \eta, a, \nu$ and dimension $d$) with the relaxation rates predicted by the macroscopic PDE.
   • By varying the steady-state density $\bar{\rho}$ (via changing $N$ or domain size $L$) while keeping microscopic parameters constant, the authors infer the functional form of $D(\rho)$ and the exponent $\gamma$.

Key Results

• Dimensional Dependence of the Exponent: The study establishes that the exponent $\gamma$ in the nonlinear diffusivity $D(\rho) \propto \rho^\gamma$ is determined jointly by the microscopic force law exponent $\nu$ and the spatial dimension $d$.
• Scaling Relationship: Dimensional analysis and simulation data reveal a universal scaling relationship: $\gamma = \beta - 2/d$, where $\beta$ is a scaling exponent related to the microscopic force law.
  • In 1D ($d=1$), for a force law with $\nu=1$ (Hooke's law), the resulting diffusivity is $D(\rho) \propto \rho^{-2}$.
  • In 2D ($d=2$), the same microscopic force law ($\nu=1$) yields $D(\rho) \propto \rho^{-1.11}$.
  • Generally, as dimension $d$ increases, the magnitude of the nonlinearity $|\gamma|$ decreases.
• Emergence of Linear Diffusion: As $d \to \infty$, $\gamma \to 0$, implying that linear diffusion becomes increasingly relevant in higher dimensions for the same microscopic mechanics.
• Validation: Numerical simulations in both 1D and 2D confirm that the inferred $D(\rho)$ accurately reproduces the discrete relaxation dynamics. The results hold across various initial conditions and domain sizes, confirming that macroscopic transport depends on local length scales ($a$ and $\rho$) rather than global domain scales.

Significance and Claims
The paper claims to reveal a novel physical mechanism for emergent macroscopic transport phenomena in tissue mechanics. The primary significance lies in demonstrating that dimensionality is not merely a geometric detail but fundamentally alters the macroscopic law obtained after coarse-graining.

• Limitations of 1D Models: Insights derived from 1D models cannot be assumed to hold unchanged in higher dimensions. Specifically, the use of linear models (Hooke's law) in 1D masks critical details about how microscopic length scales map to emergent tissue-scale transport.
• Novel Coarse-Graining Method: The authors propose a method to coarse-grain microscopic dynamics into nonlinear diffusion functions based on comparing relaxation rates. This approach does not rely on traditional continuum limits, making it particularly suitable for finite systems.
• Physical Mechanism: The work provides a physical explanation for why porous-medium-type nonlinear diffusivities arise in tissue mechanics and how the specific exponent is fixed by the interplay of microscopic mechanics and spatial dimension.

The authors conclude that the same microscopic interaction law does not generate a dimension-independent continuum limit; instead, dimension reshapes the effective nonlinear diffusivity, changing the exponent in $D(\rho) \propto \rho^\gamma$. This phenomenon is analogous to dimension-dependent behaviors in random-walk theory and renormalization-group methods.