Kramers-Kronig Relations: Theory & Applications

• Kramers-Kronig relations are integral transforms that connect the real and imaginary parts of causal, linear response functions, ensuring consistency between dispersion and absorption.
• They rely on the principles of causality and analyticity, serving as a key tool for extracting optical constants and validating experimental data in various scientific fields.
• Advanced numerical methods, including FFT-based Hilbert transforms and subtractive techniques, enable stable reconstruction of physical parameters from broadband absorption spectra.

The Kramers-Kronig (KK) relations are a set of integral transforms that express a fundamental connection between the real and imaginary parts of complex response functions in linear, causal, time-invariant systems. Emerging from the principle of causality and the associated analyticity of response functions in the complex frequency domain, the KK relations constitute the mathematical backbone linking dispersion (frequency-dependent phase velocity) and absorption (attenuation) across physics, chemistry, engineering, and materials science.

1. Mathematical Foundations and Physical Prerequisites

The KK relations originate from the condition that any physically admissible, linear response function—such as electric susceptibility, conductivity, dielectric function, or refractive index—must be analytic in the upper half of the complex frequency plane, a direct consequence of causality. Given a causal impulse response $G(t)$ (with $G(t) = 0$ for $t < 0$), its Fourier transform $H(\omega) = \int_0^{\infty} e^{i\omega t} G(t)\,dt$ is analytic for $\Im \omega > 0$ (Prevedelli et al., 2024). Under $L^1$-integrability, the Laplace-transform-based proof shows that

$$\mathrm{Re}\ H(\omega) = \frac{1}{\pi}\ \mathcal{P}\!\! \int_{-\infty}^{\infty} \frac{\mathrm{Im}\ H(\nu)}{\nu-\omega}\,d\nu,\qquad \mathrm{Im}\ H(\omega) = -\frac{1}{\pi}\ \mathcal{P}\!\! \int_{-\infty}^{\infty} \frac{\mathrm{Re}\ H(\nu)}{\nu-\omega}\,d\nu.$$

Here, $\mathcal{P}$ denotes the Cauchy principal value. Titchmarsh's classical proof requires only $L^2$-integrability (finite energy) but is technically more involved (Prevedelli et al., 2024, Bechhoefer, 2011).

For response functions with symmetry properties, these can be recast as one-sided integrals: $$n(\omega)-1 = \frac{2}{\pi}\,\mathcal{P}\!\int_{0}^{\infty}\frac{\omega' k(\omega')}{\omega'^{2} - \omega^{2}}\,d\omega',\qquad k(\omega) = -\frac{2\omega}{\pi}\,\mathcal{P}\!\int_{0}^{\infty} \frac{n(\omega') - 1}{\omega'^{2} - \omega^{2}}\,d\omega'$$ for the complex refractive index $G(t) = 0$0 (Cundin et al., 2010).

2. Theoretical Implications: Causality, Analyticity, and Dispersion

Causality ensures that dissipation and dispersion are necessarily coupled. The analyticity (in the frequency upper half-plane) required by causality imposes that the real and imaginary parts of any linear response function—or susceptibility $G(t) = 0$1, dielectric function $G(t) = 0$2, magnetic permeability $G(t) = 0$3, conductivity $G(t) = 0$4, or impedance $G(t) = 0$5—form a Hilbert transform pair (Bedard et al., 2018, Hickey et al., 2010, Carcione et al., 2022, Margo et al., 2014).

For instance, for the dielectric function: $G(t) = 0$6 (Margo et al., 2014, Cundin et al., 2010, Bedard et al., 2018). Analogous expressions exist for frequency-dependent conductivity, mechanical compliance, and admittance.

These relations enforce strict consistency: if any model or measured dataset fails the KK checks (e.g., a frequency-dependent real conductivity with zero imaginary part), it violates causality and is nonphysical (Bedard et al., 2018).

3. Symmetry Properties, Generalizations, and Modified Forms

The explicit form of the KK integrals is adapted depending on symmetries and physical context:

• Parity: For functions with $G(t) = 0$7 (even real, odd imaginary), the KK integrals reduce to one-sided forms over positive frequencies (Cundin et al., 2010, Pain et al., 17 Jun 2025).
• Singularities and Static Conductivity: If response functions contain a pole at $G(t) = 0$8 (e.g., finite DC conductivity), the standard relations must be modified. For dielectric functions with nonzero static conductivity $G(t) = 0$9, the imaginary part develops a term $t < 0$0, and the modified KK relations are (Bobrov et al., 2010, Klimchitskaya et al., 2018, Margo et al., 2014):

$t < 0$1

• Finite Geometries and Diffusion Systems: For Poisson-Nernst-Planck-type impedances diverging as $t < 0$2 at low frequency (capacitive branch), an extra term reflecting this singularity must be included (Evangelista et al., 2013).

4. Computational and Numerical Implementation

KK analysis is central for reconstructing dispersive properties from absorption spectra or impedance data:

FFT Approaches: The discrete Hilbert transform via FFT is a standard, efficient route. For a regularly sampled absorption $t < 0$3, compute FFT, multiply each bin by $t < 0$4, apply inverse FFT, and add unity to recover $t < 0$5 (Cundin et al., 2010, Fitzgerald, 2020, Carcione et al., 2022). Windows, scaling to known refractive index values, and interpolation (e.g., Neville's, Richardson's methods) compensate for experimental bandwidth limitations and missing spectral data (Cundin et al., 2010, Gienger et al., 2017).

Singly/Subtractively-Subtracted Forms: Anchor point subtraction mitigates truncation errors in finite-bandwidth datasets. The kernel is adjusted, and the Hilbert transform becomes more stable numerically (Fitzgerald, 2020).

Principal Value Singularities: Care must be taken handling numerical singularities at the poles. Specialized segmentations (e.g., Newton–Cotes for smooth windows, Lagrange interpolation near poles) enforce correct principal value treatment (Fitzgerald, 2020).

Physical Example: For complex index data spanning from DC to x-ray, empirical and theoretical absorption data are merged optimally, interpolated/extrapolated to fill gaps, then input to a numerically stable Hilbert transform process (Cundin et al., 2010).

5. Extensions: Spatial, Time-Varying, and Non-Standard Domains

Spatial Kramers-Kronig Relations: For a permittivity profile $t < 0$6 analytic in $t < 0$7, the real and imaginary parts satisfy

$t < 0$8

enabling the construction of unidirectional reflectionless media and spatial analogues to frequency causality (Horsley et al., 2015, Zhang et al., 2020). Such structures have been implemented in cold-atom systems and gradient-index photonic devices.

Time-Varying Media: In spatiotemporally modulated or pulsed systems, a generalized KK framework arises by treating the time-delay variable as the key parameter and deriving Hilbert-transform relations for the Doppler/conjugate frequencies (Solís et al., 2020). Even for rapid modulation, as long as causality in input delay is preserved, generalized KK dispersion constraints hold.

Acoustics, Mechanics, and Quantum Systems: The same analytic structure underlies viscoelastic compliance, acoustic propagation (including in waveguides and leaky modes (Krylov, 2019)), and even quantum field theoretical scattering amplitudes (Klimchitskaya et al., 2018). The KK relations thus impose universal dispersive-absorptive constraints in both classical and quantum domains.

6. Applications and Physical Consequences

KK relations are indispensable in experimental analysis and theoretical model validation:

• Extraction of Optical Constants: From broadband absorption or reflectance, use KK to reconstruct frequency-dependent refractive index and phase velocity—essential in biological tissue optics, plasma physics, and semiconductor metrology (Cundin et al., 2010, Gienger et al., 2017, Pain et al., 17 Jun 2025).
• Separation of Conduction and Polarization Contributions: Permittivity measurements often conflate dipolar relaxation and conductivity. Proper KK analysis enables unique separation, yielding static conductivity estimates and true dielectric functions (Margo et al., 2014).
• Verification of Data Consistency: Any measurement purporting to show frequency-dependent $t < 0$9 must pass the KK reconstruction for $H(\omega) = \int_0^{\infty} e^{i\omega t} G(t)\,dt$0, and vice versa. Significant discrepancies indicate violation of causality or presence of experimental artifacts (Bedard et al., 2018).
• Constraints on Negative Refraction: KK relations derived from causality fundamentally restrict the possible frequency windows for negative refractive index and impose the Depine–Lakhtakia criterion without manual branch selection (Hickey et al., 2010).
• Gravitational Wave and Lensing Analysis: KK relations, adapted to amplification factors in lensing, impose constraints on the extraction of lensing signals from observed waveforms and provide practical consistency checks on template accuracy (Tanaka et al., 2023).

7. Broader Context, Limitations, and Consistency Checks

KK relations are necessary (but not by themselves sufficient) for physical admissibility; they assume complete causality and linearity. Non-minimum-phase systems or those with upper-half-plane zeros (e.g., pure delay, certain all-pass systems) respect KK for real/imaginary parts but not for magnitude–phase pairs (cf. Bode relations) (Bechhoefer, 2011). Incomplete experimental bandwidth or improper extrapolation require careful treatment (windowing, modeling, or anchor-point selection) to avoid artifacts (Fitzgerald, 2020, Margo et al., 2014, Gienger et al., 2017). In the presence of non-trivial singularities or static conduction, modified KK formulas explicitly incorporate the associated terms (Bobrov et al., 2010, Evangelista et al., 2013).

The universality of the KK framework, extending from classical electromagnetism to quantum fields, from optical materials to plasma and biosystems, underpins its centrality in the physical sciences. The rigorous imposition of causality via analyticity and the Hilbert transform is the essential link between absorption and dispersion, measurement and model, experiment and theory.