From Discrete Chains to the Continuum Limit

The Intuition of the Limit

Imagine a heavy rope strung with beads. If you have only ten beads, the motion is jerky; energy hops from one bead to the next in distinct steps. If you have a million beads, the motion appears perfectly smooth. The "continuum limit" is the mathematical process of removing the beads and replacing them with a continuous string. In this limit, the discrete coordinates \( q_i \) become a continuous function \( y(x, t) \). The discrete mass \( m \) becomes a linear mass density \( \mu \), and the discrete spring constant \( k \) becomes a tension \( T \). The key insight is that the equations of motion do not change their fundamental nature; they simply change their notation from difference equations to differential equations. The physics remains the same: local interactions propagate disturbances through the system.

From Discrete Coordinates to a Field

Consider a linear chain of \( N \) identical masses \( m \), each connected to its neighbours by springs of constant \( k \). The displacement of the \( i \)-th mass from its equilibrium position is \( q_i(t) \). The spacing between masses is \( a \). To take the continuum limit, we introduce a continuous spatial coordinate \( x = i a \). The discrete displacement \( q_i \) is replaced by the field variable \( y(x, t) \), such that \( y(i a, t) = q_i(t) \).

The kinetic energy of the discrete system is \( T = \frac{1}{2} \sum_{i=1}^{N} m \dot{q}_i^2 \). In the continuum limit, the sum becomes an integral. Since \( dx = a \), we have \( \sum_i \dots \approx \frac{1}{a} \int \dots dx \). The mass \( m \) is related to the linear density \( \mu \) by \( m = \mu a \). Substituting these, the kinetic energy density becomes \( \frac{1}{2} \mu \dot{y}^2 \), and the total kinetic energy is:

Similarly, the potential energy stored in the springs depends on the relative displacement \( q_{i+1} - q_i \). Using a Taylor expansion, \( q_{i+1} - q_i \approx a \frac{\partial y}{\partial x} \). The potential energy of one spring is \( \frac{1}{2} k (q_{i+1} - q_i)^2 \approx \frac{1}{2} k a^2 \left( \frac{\partial y}{\partial x} \right)^2 \). Summing over all springs and converting to an integral, we identify the tension \( T_{\text{tension}} = k a \). The potential energy becomes:

The Lagrangian is the difference \( L = T - V \), now expressed as an integral of a Lagrangian density \( \mathcal{L} \).

The Euler-Lagrange Equation for Fields

In particle mechanics, the Euler-Lagrange equation is \( \frac{d}{dt} \frac{\partial L}{\partial \dot{q}} - \frac{\partial L}{\partial q} = 0 \). For a field \( y(x,t) \), the Lagrangian is an integral over space. The principle of least action requires that the variation \( \delta S = 0 \) for arbitrary variations \( \delta y(x,t) \) that vanish at the boundaries.

Applying the calculus of variations to the functional \( S = \int dt \int dx \, \mathcal{L}(y, \dot{y}, y') \), we obtain the field Euler-Lagrange equation. The time derivative term involves \( \frac{\partial \mathcal{L}}{\partial \dot{y}} \), and the spatial derivative term involves integration by parts, which shifts the derivative onto the variation. The resulting equation is:

For our string, \( \mathcal{L} = \frac{1}{2} \mu \dot{y}^2 - \frac{1}{2} T_{\text{tension}} (y')^2 \). Note that \( \mathcal{L} \) does not depend explicitly on \( y \), so \( \frac{\partial \mathcal{L}}{\partial y} = 0 \). Calculating the other terms:

Substituting these into the field Euler-Lagrange equation:

Assuming \( \mu \) and \( T_{\text{tension}} \) are constant, this simplifies to the standard wave equation:

The wave speed is \( v = \sqrt{T_{\text{tension}}/\mu} \). This matches the result from the discrete chain in the long-wavelength limit, confirming the consistency of the continuum approximation.

Connection to Normal Modes

In the discrete system, we found that the normal modes are standing waves with specific frequencies \( \omega_r \). In the continuum limit, the number of modes becomes infinite, and the discrete set of frequencies becomes a continuous spectrum. The normal coordinate \( \xi_r \) in the discrete case corresponds to the amplitude of a specific Fourier mode in the continuum.

The decoupling result from D-132, where \( \xi_r(t) = C_r \cos(\omega_r t + \phi_r) \), generalizes to a superposition of continuous modes. The total displacement \( y(x,t) \) can be written as an integral over wave numbers \( k \):

Here, the dispersion relation is \( \omega(k) = v k \), which is linear for a string. This linear dispersion is a direct consequence of the quadratic form of the Lagrangian density. If the potential energy had higher-order terms in \( y' \), the dispersion would be non-linear, leading to phenomena like solitons. The continuum limit thus provides a bridge between the discrete algebraic eigenvalue problem and the continuous spectral analysis of differential operators.