Quantum Noise, Dissipation, and Baths

Quantum noise is the language used when uncontrolled degrees of freedom influence a quantum system. Sometimes the noise is well modeled as a classical random process. More often in open-system theory it is a bath operator with noncommuting fluctuations, asymmetric spectra, dissipation, and backaction. The purpose of this chapter is to connect those descriptions without blurring their assumptions.

The chapter has three jobs. First, it defines the correlation functions and spectra that quantify fluctuations. Second, it explains how dissipation is tied to response, especially near equilibrium. Third, it gives canonical models, such as oscillator baths and two-level systems coupled to bosonic environments, that reappear throughout atomic physics, condensed matter, quantum optics, superconducting devices, and quantum information.

Organizing Picture

A standard microscopic starting point is a system–bath Hamiltonian

H = H_S+H_B+H_I, \qquad H_I = \sum_\alpha A_\alpha\otimes B_\alpha.

The system operators $A_\alpha$ are retained, while the bath operators $B_\alpha$ are unobserved, thermally prepared, driven, monitored, or otherwise modeled. If the bath state $\rho_B$ is stationary, the basic time-domain object is

C_{\alpha\beta}(t) = \operatorname{Tr}_B \left[ B_\alpha(t)B_\beta(0)\rho_B \right], \qquad B_\alpha(t) = e^{iH_Bt/\hbar}B_\alpha e^{-iH_Bt/\hbar}.

The corresponding spectrum, with the convention used in the chapter pages, is

S_{\alpha\beta}(\omega) = \int_{-\infty}^{\infty} dt\, e^{i\omega t} C_{\alpha\beta}(t).

Weak-coupling transition rates, dephasing rates, Lamb shifts, fluctuation–dissipation formulas, and filter-function estimates all sample some version of these correlations. The details depend on operator ordering, Fourier convention, temperature, and which degrees of freedom have been traced out or monitored.

Three Languages

Noise and dissipation are best learned in three complementary languages.

Language	Central object	What it explains
Time domain	$C_{\alpha\beta}(t)$	memory time, Markov approximation, colored noise
Frequency domain	$S_{\alpha\beta}(\omega)$	relaxation, dephasing, filter functions, detailed balance
Response theory	susceptibilities and damping kernels	fluctuation–dissipation relations, friction, impedance, admittance

These languages agree only after the conventions are matched. A classical real stochastic process has a symmetric two-sided spectrum. A quantum bath generally has an unsymmetrized spectrum that distinguishes positive and negative frequencies. At equilibrium the asymmetry is constrained by the Kubo–Martin–Schwinger relation; for one common convention,

S_{BB}(-\omega) = e^{-\beta\hbar\omega} S_{BB}(\omega).

For $\omega \gt 0$ , the zero-temperature limit suppresses the negative-frequency part in this convention. That is one compact way to see why a zero-temperature bath can accept energy from a system but cannot thermally excite it.

Reading Path

Start with Quantum Noise for the distinction between classical stochastic noise, quantum bath noise, vacuum noise, thermal noise, and technical noise. Then read Correlation Functions and Noise Spectra together: the first page explains what is being correlated, and the second explains what frequency information those correlations contain.

Fluctuation–Dissipation Relation is the conceptual bridge from fluctuations to friction, damping, impedance, and linear response. It is the place to learn what equilibrium does and does not guarantee.

For Heisenberg-picture dynamics, use Quantum Langevin Equations. For traveling fields and measurement ports, use Input–Output Theory. These pages are especially important in quantum optics, optomechanics, microwave circuits, and continuous measurement.

For model Hamiltonians, read Caldeira–Leggett Model before Quantum Brownian Motion if the system coordinate is continuous. Read Spin–Boson Model before Pure Dephasing Model if the retained system is a two-level degree of freedom. Thermal and Vacuum Noise and One-Over-F Noise then describe two common noise regimes that appear across platforms.

What This Chapter Owns

This chapter owns the canonical homes for:

quantum noise as bath-operator fluctuations;
symmetrized and unsymmetrized bath correlations;
noise spectra and spectral asymmetry;
fluctuation–dissipation relations in open-system language;
quantum Langevin equations and input–output relations;
standard oscillator-bath, spin–boson, pure-dephasing, and Brownian-motion models;
thermal, vacuum, and one-over-f noise as open-system noise regimes.

It does not own every use of those ideas. General system–environment setup belongs to Baths, Reservoirs, and Environments and System–Bath Hamiltonians. Lindblad derivations and thermal generators belong to Markovian Master Equations. Failure of short-memory approximations belongs to Non-Markovian Dynamics. Thermodynamic heat, work, and entropy production belong to Quantum Thermodynamics.

When a later page needs a spectrum or bath model only as input, it should link here or to the specific canonical page rather than rederiving the same definitions.

Modeling Decisions

Before using a noise model, decide the following.

Question	Why it matters
Is the noise classical or quantum?	Quantum spectra have operator ordering and detailed-balance information.
Is the bath stationary?	Stationarity lets correlations depend only on time differences.
Is the bath thermal?	Thermal equilibrium fixes spectral asymmetry through KMS relations.
Is the noise Gaussian?	Gaussian noise is fully specified by first and second moments; non-Gaussian noise is not.
Is the spectrum broad and smooth near Bohr frequencies?	This is a key condition behind Markovian rates.
Is the coupling weak?	Weak-coupling formulas can fail under strong dressing or slow bath response.
Is the output field monitored?	Monitored degrees of freedom become records, not merely discarded bath modes.

The Approximation Checklist is the practical place to record these decisions before applying a master equation, Langevin equation, or filter-function formula.

Common Mistakes

Treating vacuum noise as absence of fluctuations.
Using a classical symmetric spectrum when the calculation requires an unsymmetrized quantum spectrum.
Mixing Fourier sign conventions without changing the interpretation of positive and negative frequencies.
Calling any frequency-dependent spectrum non-Markovian without checking the relevant system time scales.
Assuming a Lindblad equation is thermodynamically consistent just because its rates are positive.
Forgetting counterterms, renormalization, or cutoff dependence in oscillator-bath models.
Identifying one-over-f noise with a single microscopic mechanism.
Treating a measured output field as if it were an unobserved bath after conditioning on the measurement record.

Exercises

White noise from a short correlation time

Suppose a stationary classical noise process has an idealized correlation function $C(t)=D\delta(t)$ . What is its two-sided spectrum with the convention $S(\omega)=\int dt\,e^{i\omega t}C(t)$ ? What Markovian intuition does this support?

Solution

The spectrum is

S(\omega) = D \int_{-\infty}^{\infty} dt\, e^{i\omega t}\delta(t) = D.

It is flat, so the noise has no preferred frequency scale in the range where the idealization is used. In open-system derivations this supports a memoryless approximation: the bath correlation time is taken to be much shorter than the system time scales. Real spectra always have cutoffs, so the white-noise model is an approximation over a finite band.

Detailed balance from spectral asymmetry

Assume an equilibrium bath satisfies $S(-\omega)=e^{-\beta\hbar\omega}S(\omega)$ for $\omega \gt 0$ . What happens to $S(-\omega)$ as $T\to0$ ? How should this be interpreted?

Solution

As $T\to0$ , $\beta\to\infty$ , and therefore

e^{-\beta\hbar\omega} \to 0 \qquad (\omega \gt 0).

Thus $S(-\omega)\to0$ if $S(\omega)$ remains finite. With this convention, a zero-temperature bath has no thermal population available to supply energy $\hbar\omega$ to the system, although it can still accept energy from a decaying system. This is spectral detailed balance, not the statement that vacuum fluctuations vanish.

Choosing the canonical page

A qubit suffers slow random frequency drift, but no observed energy relaxation, over the duration of a Ramsey experiment. Which pages in this chapter are the most natural starting points, and which page owns the corresponding Markovian master-equation limit?

Solution

The natural starting points are Quantum Noise, Noise Spectra, Pure Dephasing Model, and, if the noise is low frequency over many decades, One-Over-F Noise. The corresponding Markovian generator belongs to Pure Dephasing Master Equation.

Cross-Links

References

C. W. Gardiner and P. Zoller, Quantum Noise, 3rd ed., Springer (2004).
H.-P. Breuer and F. Petruccione, The Theory of Open Quantum Systems, Oxford University Press (2002).
A. A. Clerk, M. H. Devoret, S. M. Girvin, F. Marquardt, and R. J. Schoelkopf, “Introduction to quantum noise, measurement, and amplification,” Reviews of Modern Physics 82, 1155–1208 (2010).
U. Weiss, Quantum Dissipative Systems, 4th ed., World Scientific (2012).
A. O. Caldeira and A. J. Leggett, “Path integral approach to quantum Brownian motion,” Physica A 121, 587–616 (1983).
A. J. Leggett, S. Chakravarty, A. T. Dorsey, M. P. A. Fisher, A. Garg, and W. Zwerger, “Dynamics of the dissipative two-state system,” Reviews of Modern Physics 59, 1–85 (1987).
R. Kubo, “The fluctuation-dissipation theorem,” Reports on Progress in Physics 29, 255–284 (1966).