Opening the book…
This chapter gives a detailed, step-by-step derivation of the memory function (damping kernel) and the generalized Langevin equation (GLE) for a single "system" harmonic coordinate coupled bilinearly to a bath of harmonic oscillators. The bath is integrated out exactly using the causal Green's function of each oscillator. The resulting equation of motion for the system contains a convolution (memory) term and a fluctuating force.
12.1 Introduction
An overarching goal of statistical physics is to derive macroscopic irreversible equations starting with microscopic reversible equations of motion. A direct model was provided by Magalinski in 1959, solving a mechanical model of an oscillator coupled to \(N\) bath oscillators to derive the generalized Langevin equation (GLE). This stands in contrast to heuristic derivations where stochastic forces are merely postulated; here, they are derived analytically.
12.2 Model and Hamiltonian
Consider a system coordinate \(X\) with mass \(M\) and frequency \(\Omega\), bilinearly coupled to a bath of \(n\) oscillators. We use the Caldeira-Leggett Hamiltonian:
\[ H=\frac{P^{2}}{2 M}+\frac{1}{2} M \Omega^{2} X^{2}+\sum_{n}\left[\frac{p_{n}^{2}}{2 m_{n}}+\frac{1}{2} m_{n} \omega_{n}^{2}\left(q_{n}-\frac{c_{n}}{m_{n} \omega_{n}^{2}} X\right)^{2}\right] . \tag{12.1} \label{12.1} \]
Physical Role of the Counterterm: The expansion of the bath potential includes a term \(\sum_{n} \frac{c_{n}^{2}}{2 m_{n} \omega_{n}^{2}} X^{2}\). This is the "counterterm," which ensures that the coupling does not shift the equilibrium position of the system. Without it, the bath would exert a static pinning force on \(X\), making it impossible to recover free Brownian motion in the limit where the bare potential vanishes.
12.3 The Green's Function Method
The evolution of the bath coordinates \(q_{n}\) is governed by a linear inhomogeneous differential equation representing a driven harmonic oscillator:
\[ \ddot{q}_{n}(t)+\omega_{n}^{2} q_{n}(t)=\frac{c_{n}}{m_{n}} X(t) . \tag{12.2} \label{12.2} \]
The Green's function method is the mathematical engine of this derivation. We define the causal (retarded) Green's function \(g_{n}(t)\), which satisfies:
\[ \left(\frac{d^{2}}{d t^{2}}+\omega_{n}^{2}\right) g_{n}(t)=\delta(t) . \tag{12.3} \label{12.3} \]
Derivation of the Impulse Response
To solve for \(g_{n}(t)\), we apply the following physical and mathematical constraints:
- 1.Causality: As the response to an impulse at \(t=0\), we require \(g_{n}(t)=0\) for \(t<0\). This implies \(g_{n}(0)=0\).
- 2.Homogeneous Solution ( \(t>0\) ): For \(t>0\), the equation reduces to \(\ddot{g}_{n}+\omega_{n}^{2} g_{n}=0\). The general solution is \(g_{n}(t)=A \cos \left(\omega_{n} t\right)+B \sin \left(\omega_{n} t\right)\). Applying \(g_{n}(0)=0\) forces \(A=0\).
- 3.The Jump Condition: Integrating the governing equation over an infinitesimal interval \([-\epsilon,+\epsilon]\) yields:
\[ \int_{-\epsilon}^{+\epsilon}\left(\frac{d^{2} g_{n}}{d t^{2}}+\omega_{n}^{2} g_{n}\right) d t=\int_{-\epsilon}^{+\epsilon} \delta(t) d t \Longrightarrow \dot{g}_{n}\left(0^{+}\right)-\dot{g}_{n}\left(0^{-}\right)=1 . \tag{12.4} \label{12.4} \]
Since \(\dot{g}_{n}(t)=0\) for \(t<0\), we find \(\dot{g}_{n}\left(0^{+}\right)=1\). Differentiating \(g_{n}(t)=B \sin \left(\omega_{n} t\right)\) gives \(\dot{g}_{n}(t)=B \omega_{n} \cos \left(\omega_{n} t\right)\). At \(t=0, B \omega_{n}=1\), so \(B=1 / \omega_{n}\).
Thus, the impulse response is \(g_{n}(t)=\frac{\sin \left(\omega_{n} t\right)}{\omega_{n}} \Theta(t)\). By convolving this response with the driving force \(c_{n} X(t)\), we obtain the exact solution:
\[ q_{n}(t)=q_{n}(0) \cos \left(\omega_{n} t\right)+\frac{\dot{q}_{n}(0)}{\omega_{n}} \sin \left(\omega_{n} t\right)+\frac{c_{n}}{m_{n}} \int_{0}^{t} \frac{\sin \left(\omega_{n}(t-s)\right)}{\omega_{n}} X(s) d s . \tag{12.5} \label{12.5} \]
12.4 Generalized Langevin Equation
Substituting the solution for \(q_{n}(t)\) back into the system equation of motion and performing integration by parts, we arrive at the GLE:
\[ M \ddot{X}(t)+M \Omega^{2} X(t)+M \int_{0}^{t} \gamma(t-s) \dot{X}(s) d s=\xi(t) \tag{12.6} \label{12.6} \]
The memory kernel \(\gamma(t)\) represents the dissipative drag exerted by the bath:
\[ \gamma(t):=\frac{1}{M} \sum_{n} \frac{c_{n}^{2}}{m_{n} \omega_{n}^{2}} \cos \left(\omega_{n} t\right) . \tag{12.7} \label{12.7} \]
The fluctuating force \(\xi(t)\) arises from the thermalized initial conditions of the bath oscillators.
12.5 Spectral Density and the Continuum Limit
To describe a macroscopic environment, we take the limit where the number of bath oscillators \(N \rightarrow \infty\). In this continuum limit, the properties of the bath are summarized by the spectral density \(J(\omega)\) :
\[ J(\omega)=\frac{\pi}{2} \sum_{n} \frac{c_{n}^{2}}{m_{n} \omega_{n}} \delta\left(\omega-\omega_{n}\right) . \tag{12.8} \label{12.8} \]
The memory kernel \(\gamma(t)\) can then be expressed as an integral over the spectral density:
\[ \gamma(t)=\frac{2}{M \pi} \int_{0}^{\infty} \frac{J(\omega)}{\omega} \cos (\omega t) d \omega . \tag{12.9} \label{12.9} \]
12.5.1 Physical Models and the Drude Model
The choice of \(J(\omega)\) determines the physical nature of the dissipation.
- \(\eta\) (Friction Coefficient): Represents the magnitude of the coupling between the system and the bath.
- \(\omega_{c}\) (Cut-off Frequency): The frequency scale above which the bath oscillators can no longer follow the system motion efficiently.
The Drude Model is defined by the spectral density:
\[ J(\omega)=\eta \omega \frac{\omega_{c}^{2}}{\omega^{2}+\omega_{c}^{2}} . \tag{12.10} \label{12.10} \]
Physically, the Drude model accounts for the finite response time of a realistic bath. While the simple Ohmic model implies an instantaneous response, the Drude model leads to a memory kernel that decays exponentially:
\[ \gamma(t)=\eta \omega_{c} e^{-\omega_{c}|t|} . \tag{12.11} \label{12.11} \]
This indicates that the influence of the system's past velocity on its current acceleration persists for a characteristic time scale \(\tau=1 / \omega_{c}\).
| Model | \(J(\omega)\) | \(\gamma(t)\) | Physical Context |
|---|---|---|---|
| Ohmic | \(\eta \omega\) | \(2 \eta \delta(t)\) | Markovian friction |
| Ohmic (Cutoff) | \(\eta \omega e^{-\omega / \omega_{c}}\) | \(\frac{2 \eta}{\pi} \frac{\omega_{c}}{1+\left(\omega_{c} t\right)^{2}}\) | Finite response time |
| Drude | \(\eta \omega \frac{\omega_{c}^{2}}{\omega^{2}+\omega_{c}^{2}}\) | \(\eta \omega_{c} e^{-\omega_{c}|t|}\) | Exponential memory |
Bibliography
[1] V. I. Magalinski, JETP 36, 1942-1944 (1959).
[2] H. Mori, Prog. Theor. Phys. 33, 423-455 (1965).
[3] R. Zwanzig, Nonequilibrium Statistical Mechanics (Oxford, 2001).
[4] R. Kubo et al., Statistical Physics II (Springer, 2012).
[5] S. Dattagupta and A. Ghosh, Physics of Fluids 37, 027199 (2025).
This page titled 12: Magalinski's Derivation of the Memory Function and the Generalized Langevin Equation for a System Coupled to a Bath of Harmonic Oscillators was last modified on Mon, 02 Nov 2026 06:09:26 GMT and is shared under a CC BY-NC-SA 4.0 license and was authored, remixed, and/or curated by Pabitra N. Sen.
The book could not be loaded just now.