Response Functions and Correlation Functions

In the study of statistical mechanics and many-body physics, understanding how a system behaves—both in isolation and under external influence—is fundamental. This behavior is captured by two primary mathematical constructs: response functions and correlation functions.

While they appear to describe different phenomena, they are two sides of the same coin. A response function quantifies how a system reacts to an external perturbation (such as an electric field or a mechanical force), characterizing the "dissipative" or "reactive" nature of the system. Conversely, a correlation function describes the internal, spontaneous fluctuations that occur even in equilibrium, capturing how a physical quantity at one time is related to itself or another quantity at a later time. The profound link between these two concepts is encapsulated in the Fluctuation-Dissipation Theorem (FDT), which provides a unified framework connecting microscopic fluctuations to macroscopic transport properties.

1. Defining the Core Quantities

Response Functions

When a system is subjected to a weak external field $h_B(t)$ that couples to an observable $B$, the expectation value of another observable $A$ will change. In the regime of linear response theory, this change $\delta\langle A(t)\rangle$ is assumed to be proportional to the history of the perturbation. This relationship is expressed as:

[
\delta\langle A(t)\rangle = \int_{-\infty}^{t} \chi_{AB}(t-t') , h_B(t') , \mathrm{d}t'
]

Here, $\chi_{AB}(t)$ is the response function. For a stationary system, the response depends only on the time difference $t-t'$. Crucially, the response function must satisfy the principle of causality: a system cannot respond to a perturbation before it occurs, meaning $\chi_{AB}(t) = 0$ for $t < 0$.

Correlation Functions

In thermal equilibrium, a system is never truly "still"; it undergoes continuous, spontaneous fluctuations. The correlation function (or covariance) measures the statistical relationship between the fluctuations of two observables, $A$ and $B$, at different times:

[
C_{AB}(t) = \langle \delta A(t) \delta B(0) \rangle
]

where $\delta A(t) = A(t) - \langle A \rangle$ represents the deviation from the equilibrium mean. For stationary processes, these functions exhibit specific properties:

• Symmetry: $C_{AB}(t) = C_{BA}(-t)$.
• Positive Definiteness: The auto-correlation at $t=0$ must be non-negative, $C_{AA}(0) = \langle (\delta A)^2 \rangle \ge 0$.
• Spectral Representation: By applying a Fourier transform, we obtain the power spectrum $S_{AB}(\omega)$, which describes the intensity of fluctuations at a specific frequency $\omega$:
  [
  S_{AB}(\omega) = \int_{-\infty}^{\infty} C_{AB}(t) e^{i\omega t} , \mathrm{d}t
  ]

2. Linear Response Theory and the Kubo Formula

To derive the relationship between the Hamiltonian and the response, we consider a system perturbed by a term $-h_B(t)B$. Using the time-evolution operator and the Dyson expansion, we arrive at the celebrated Kubo formula. In the quantum mechanical regime, the response function is given by:

[
\chi_{AB}(t) = \frac{i}{\hbar} \theta(t) \langle [A(t), B(0)] \rangle_{0}
]

Where:

• $\theta(t)$ is the Heaviside step function, enforcing causality.
• $[A(t), B(0)]$ is the commutator of the operators, indicating that the quantum response is fundamentally driven by the non-commutative nature of the system's observables.
• $\langle \dots \rangle_0$ denotes the expectation value taken over the equilibrium ensemble.

In the classical limit ($\hbar \to 0$), the commutator term is replaced by the Poisson bracket, and the prefactor $i/\hbar$ transitions to $\beta = 1/k_{\mathrm{B}}T$, yielding the classical version of the Kubo formula.

3. The Fluctuation-Dissipation Theorem (FDT)

The most significant result in this field is the Fluctuation-Dissipation Theorem, which bridges the gap between the response (dissipation) and the correlation (fluctuations). It states that the dissipative part of the response—specifically the imaginary part of the frequency-dependent susceptibility $\chi''(\omega)$—is directly proportional to the power spectrum of the fluctuations.

In a general quantum equilibrium state, the relationship is:
[
\operatorname{Im} \chi_{AB}(\omega) = \frac{\pi}{2\hbar} \left[ 1 - e^{-\beta\hbar\omega} \right] S_{AB}(\omega)
]

In the classical limit, where thermal energy dominates quantum effects ($\hbar\omega \ll k_{\mathrm{B}}T$), this simplifies to a much more intuitive form:
[
\chi_{AB}''(\omega) = \frac{\omega}{2k_{\mathrm{B}}T} S_{AB}(\omega)
]

This identity tells us something profound: the way a system dissipates energy when pushed by an external force is identical to the way it relaxes from its own spontaneous thermal fluctuations. This allows scientists to predict macroscopic transport coefficients (like electrical conductivity or viscosity) simply by observing the noise in an equilibrium system.

4. Illustrative Example: The Damped Harmonic Oscillator

To see these concepts in action, consider a classical damped harmonic oscillator with mass $m$, spring constant $k$, and damping coefficient $\gamma$, subject to an external force $F(t)$. The equation of motion is:
[
m\ddot{x} + \gamma\dot{x} + kx = F(t)
]

The Response Function:
If we apply an impulsive force (a delta function $\delta(t)$), the system's response (the Green's function) is:
[
\chi(t) = \frac{1}{m\omega_{d}} e^{-\frac{\gamma}{2m}t} \sin(\omega_{d}t) \theta(t)
]
where $\omega_{d} = \sqrt{\omega_{0}^{2} - (\gamma/2m)^2}$ is the damped natural frequency. This function describes how the system "rings" and eventually decays after being struck.

The Correlation Function:
At thermal equilibrium at temperature $T$, the position auto-correlation function is:
[
C_{xx}(t) = \frac{k_{\mathrm{B}}T}{k} e^{-\frac{\gamma}{2m}|t|} \cos(\omega_{d}t)
]
By taking the Fourier transform of $C_{xx}(t)$ and applying the FDT, one can directly recover the imaginary part of the oscillator's susceptibility, confirming the theoretical link.

5. Computational Methodologies

In modern research, calculating these functions requires various numerical and analytical strategies depending on the complexity of the system:

In practice, a common workflow in computational physics involves using MD + Kubo. Researchers use Molecular Dynamics to simulate a system in equilibrium, calculate the time-correlation functions of specific properties (like the stress tensor or dipole moment), and then use the FDT to derive macroscopic transport coefficients such as viscosity or conductivity.

Summary

The interplay between response functions and correlation functions is a cornerstone of statistical physics. While the former describes the system's reaction to the "outside world," the latter describes its "inner life" of fluctuations. The Fluctuation-Dissipation Theorem provides the essential mathematical bridge, proving that dissipation and fluctuation are fundamentally inseparable. Whether through analytical derivation or advanced molecular simulations, mastering these tools is indispensable for characterizing the dynamics of matter across all scales.