Asymmetric Response Function

• Asymmetric Response Function is a response law that changes when operations like input exchange or spatial reversal occur, revealing hidden biases and side dependencies.
• It is characterized by quantifiable shifts and nonreciprocal elements found in systems ranging from optical metasurfaces and silicon detectors to macro-financial models.
• The study of these functions enhances calibration, metamaterial design, and predictive modeling by providing practical insights into nonlinear and asymmetric behaviors.

An asymmetric response function is a response law that is not invariant under a symmetry operation that would otherwise identify two perturbations or two observation channels as equivalent. In the surveyed literature, that broken symmetry may be an exchange of input frequencies, reversal of spatial frequency, interchange of illumination side, reversal of perturbation sign, exchange of two temperatures, or displacement of a detector response center away from a geometrical reference. The result is a broad but technically coherent class of objects: side-dependent optical reflectance differences, antisymmetric nonlinear conductivities, shifted strip-response functions, threshold-like Josephson critical currents, asymmetric generalized impulse responses, and direction-dependent covariance kernels all instantiate the same formal idea, namely that the response map distinguishes between operations that a symmetric model would identify (Kanda et al., 2024, Landi et al., 2014, Deop-Ruano et al., 2024, Guarcello et al., 2018, AlBahar et al., 2021).

1. Conceptual scope and symmetry classes

The surveyed literature suggests that there is no single universal formalism called the asymmetric response function. Instead, the phrase denotes a family of response objects whose asymmetry is defined relative to a context-specific symmetry. In nonlinear macro-financial models, asymmetry means that responses differ by shock sign, shock magnitude, and state, so that $$\bm \gamma_{jh,t}^{(+\varsigma)} \neq - \bm \gamma_{jh,t}^{(-\varsigma)}$$ and $$\bm \gamma_{jh,t}^{(6)} \neq 6\,\bm \gamma_{jh,t}^{(1)}$$ (Huber et al., 2024). In static network games, asymmetry is the mismatch of strategic cross-effects, $$\frac{\partial^2 u_i}{\partial a_i \partial a_j} \neq \frac{\partial^2 u_j}{\partial a_j \partial a_i}$$, or, in linear-quadratic form, the nonreciprocity $G\neq G^T$ (Rokade et al., 8 Aug 2025). In solvable nonequilibrium spin models, asymmetry is encoded by $\mathcal A_{ij}(t,t')$, the difference between two time-ordered correlational insertions, and the analytically tractable class is precisely the one for which that asymmetry vanishes (Corberi et al., 2024).

These uses are related by a common structural criterion: the response changes when one exchanges arguments, reverses direction, or compares forward and backward propagation. A recurrent clarification in the literature is that asymmetry is not identical to nonreciprocity, irreversibility, or disorder. A planar bilayer can be passive and reciprocal yet still possess a side-dependent optical response (Deop-Ruano et al., 2024). Conversely, a nonequilibrium model can violate detailed balance and still satisfy $\mathcal A_{ij}(t,t')=0$ (Corberi et al., 2024).

Context	Response object	Broken symmetry
Silicon microstrips	$g(x)$, with shift $\delta_g$	strip-axis symmetry (Landi et al., 2014)
Thermal radiation	$\Delta \mathcal R(\omega,\mu)$	left-right side exchange (Deop-Ruano et al., 2024)
Chiral nonlinear optics	$\sigma_{\mathrm A}(\omega_1,\omega_2)$	$$\bm \gamma_{jh,t}^{(6)} \neq 6\,\bm \gamma_{jh,t}^{(1)}$$0 (Kanda et al., 2024)
Oscillator susceptibility	$$\bm \gamma_{jh,t}^{(6)} \neq 6\,\bm \gamma_{jh,t}^{(1)}$$1	phase alignment at criticality (Terada et al., 2018)
Bayesian optimization	$$\bm \gamma_{jh,t}^{(6)} \neq 6\,\bm \gamma_{jh,t}^{(1)}$$2	ordering of inputs (AlBahar et al., 2021)

2. Canonical mathematical forms

One common form is a shifted response centroid. In silicon microstrip detectors, the strip response function $$\bm \gamma_{jh,t}^{(6)} \neq 6\,\bm \gamma_{jh,t}^{(1)}$$3 is asymmetric when its center of gravity does not coincide with the strip’s geometrical reference. The asymmetry parameter is

$$\bm \gamma_{jh,t}^{(6)} \neq 6\,\bm \gamma_{jh,t}^{(1)}$$4

and the key identity

$$\bm \gamma_{jh,t}^{(6)} \neq 6\,\bm \gamma_{jh,t}^{(1)}$$5

shows that asymmetry introduces a constant reconstruction bias into center-of-gravity and $$\bm \gamma_{jh,t}^{(6)} \neq 6\,\bm \gamma_{jh,t}^{(1)}$$6-based position estimators (Landi et al., 2014).

A second form is a side-difference kernel in radiative transport. For thermal radiation forces on a planar bilayer, the decisive asymmetric response function is

$$\bm \gamma_{jh,t}^{(6)} \neq 6\,\bm \gamma_{jh,t}^{(1)}$$7

which enters the force law

$$\bm \gamma_{jh,t}^{(6)} \neq 6\,\bm \gamma_{jh,t}^{(1)}$$8

Here the asymmetry is entirely in the difference of side-dependent reflectances; if $$\bm \gamma_{jh,t}^{(6)} \neq 6\,\bm \gamma_{jh,t}^{(1)}$$9, the force vanishes even out of thermal equilibrium (Deop-Ruano et al., 2024).

A third form is frequency antisymmetry in nonlinear conductivity. In chiral optical response,

$$\frac{\partial^2 u_i}{\partial a_i \partial a_j} \neq \frac{\partial^2 u_j}{\partial a_j \partial a_i}$$0

with

$$\frac{\partial^2 u_i}{\partial a_i \partial a_j} \neq \frac{\partial^2 u_j}{\partial a_j \partial a_i}$$1

The antisymmetric sector is the nonlinear frequency-asymmetric response, and in the cubic chiral case it is tied to the electric toroidal monopole $$\frac{\partial^2 u_i}{\partial a_i \partial a_j} \neq \frac{\partial^2 u_j}{\partial a_j \partial a_i}$$2 (Kanda et al., 2024).

A fourth form is the separation of entropic and frenetic contributions on trajectory space. For nonequilibrium response theory,

$$\frac{\partial^2 u_i}{\partial a_i \partial a_j} \neq \frac{\partial^2 u_j}{\partial a_j \partial a_i}$$3

so the response is asymmetric relative to equilibrium fluctuation-dissipation structure whenever the frenetic term survives. In the complementary spin-model formulation, the generalized fluctuation-dissipation relation reads

$$\frac{\partial^2 u_i}{\partial a_i \partial a_j} \neq \frac{\partial^2 u_j}{\partial a_j \partial a_i}$$4

with $$\frac{\partial^2 u_i}{\partial a_i \partial a_j} \neq \frac{\partial^2 u_j}{\partial a_j \partial a_i}$$5 the asymmetry term; when $$\frac{\partial^2 u_i}{\partial a_i \partial a_j} \neq \frac{\partial^2 u_j}{\partial a_j \partial a_i}$$6, the response simplifies but does not thereby become equilibrium-like (Maes, 2020, Corberi et al., 2024).

3. Electromagnetic and photonic realizations

In metamaterials, asymmetry is frequently introduced deliberately to unlock otherwise dark or symmetry-forbidden modes. A toroid-like cluster of six asymmetric double-bar resonators supports an optical toroidal dipolar response under normal incidence only when the bar-length asymmetry is arranged in the specific flipped pattern that breaks rotational symmetry while preserving the toroidal circulation. The mode is identified by a closed $$\frac{\partial^2 u_i}{\partial a_i \partial a_j} \neq \frac{\partial^2 u_j}{\partial a_j \partial a_i}$$7-field vortex, a strong longitudinal electric field at the toroid center, and a multipole decomposition in which $$\frac{\partial^2 u_i}{\partial a_i \partial a_j} \neq \frac{\partial^2 u_j}{\partial a_j \partial a_i}$$8 dominates around $$\frac{\partial^2 u_i}{\partial a_i \partial a_j} \neq \frac{\partial^2 u_j}{\partial a_j \partial a_i}$$9 (Dong et al., 2012).

A more explicitly functional example is the MIMO metasurface processor with asymmetric optical response. There the required first-order differentiation operator is odd in transverse spatial frequency,

$G\neq G^T$0

so the metasurface must distinguish $G\neq G^T$1 from $G\neq G^T$2 at normal incidence. The GSTC-based synthesis shows that a normal susceptibility component $G\neq G^T$3 is mandatory for the odd transmission term, and the resulting asymmetric transfer function enables simultaneous differentiation of an $G\neq G^T$4-varying and a $G\neq G^T$5-varying input signal (Babaee et al., 2020).

Asymmetric spectral response also appears as a natural measurement pointer. In waveguided plasmonic crystals, the Fano line shape produced by interference between a narrow quasi-guided mode and a broad plasmonic background is intrinsically skewed, with characteristic frequencies

$G\neq G^T$6

A tiny polarization-dependent shift of that asymmetric spectral response enables weak-value amplification: near-orthogonal post-selection produces large peak-frequency shifts for the real weak value and linewidth narrowing or broadening for the imaginary weak value (Singh et al., 2017).

The same theme extends into quantum optics. Under blue-detuned phonon-assisted pulsed excitation of a quantum dot, the two photons emitted in a re-excitation event are not equivalent: the first is emitted during the pulse through a dressed-state pathway and is red-shifted according to

$G\neq G^T$7

whereas the second is the ordinary exciton decay photon near the bare transition frequency and delayed by the exciton lifetime. The measured time-resolved autocorrelation and frequency-filtered time traces therefore reveal a distinctly asymmetric two-photon spectral-temporal response (Jehle et al., 9 Jul 2025).

4. Electronic, superconducting, and detector response

In silicon microstrip detectors, asymmetry is a calibration issue with direct reconstruction consequences. A nonzero average of the center-of-gravity estimator at orthogonal incidence signals $G\neq G^T$8, but the paper distinguishes that empirical flag from the actual extraction of $G\neq G^T$9, which is performed by Fourier-phase alignment and minimization of

$\mathcal A_{ij}(t,t')$0

The resulting shift $\mathcal A_{ij}(t,t')$1 yields $\mathcal A_{ij}(t,t')$2, and the asymmetry must then be inserted into the $\mathcal A_{ij}(t,t')$3-algorithm through the integration-constant correction $\mathcal A_{ij}(t,t')$4. The same framework is used to improve Lorentz-angle determination, because simple average-COG observables remain contaminated by $\mathcal A_{ij}(t,t')$5 (Landi et al., 2014).

In graphene field-effect transistors, the transfer curve asymmetry is attributed chiefly to the source and drain access regions rather than only to intrinsic carrier transport asymmetry. The total resistance is decomposed schematically as

$\mathcal A_{ij}(t,t')$6

and the simulations show that outside the immediate vicinity of the Dirac point the access resistances dominate. Their gate- and bias-dependent behavior produces unequal branch slopes, branch-dependent flattening, and current suppression, even when $\mathcal A_{ij}(t,t')$7. The source access region is particularly influential, and the same asymmetry degrades transconductance and therefore cut-off frequency in RF operation (Toral-Lopez et al., 2019).

A superconducting example is the thermally biased SIS Josephson junction with material asymmetry

$\mathcal A_{ij}(t,t')$8

Its critical current

$\mathcal A_{ij}(t,t')$9

ceases to be invariant under $\mathcal A_{ij}(t,t')=0$0 when $\mathcal A_{ij}(t,t')=0$1. The nonlinear response becomes threshold-like because the condition

$\mathcal A_{ij}(t,t')=0$2

aligns the singularities of the anomalous Green’s functions and produces abrupt jumps in $\mathcal A_{ij}(t,t')=0$3. The paper further defines a temperature-switching asymmetry

$\mathcal A_{ij}(t,t')=0$4

and finds that the maximum current jump is non-monotonic in $\mathcal A_{ij}(t,t')=0$5, peaking near $\mathcal A_{ij}(t,t')=0$6 (Guarcello et al., 2018).

5. Statistical physics, many-body media, and astrophysical plasma

In globally coupled oscillator systems, linear susceptibility provides a sharp criterion for the effect of asymmetry. The general response takes the form

$\mathcal A_{ij}(t,t')=0$7

with $\mathcal A_{ij}(t,t')=0$8. The paper’s central distinction is that asymmetry in the natural-frequency distribution does not permit coexistence of a divergent susceptibility and a nonzero phase gap, whereas asymmetry in the coupling function or in the coupling constants does. The difference arises because only in the latter two cases do the phases of the numerator and denominator cease to be locked (Terada et al., 2018).

Nonequilibrium spin systems provide a complementary notion of asymmetry. The quantity

$\mathcal A_{ij}(t,t')=0$9

measures the failure of symmetry under exchanging the dynamical insertion $g(x)$0 and the spin variable $g(x)$1 across the two times. A broad solvable class, including the voter model and the one-dimensional Glauber-Ising chain, is characterized by $g(x)$2; in that case the equal-site fluctuation-dissipation ratio becomes

$g(x)$3

and $g(x)$4 for any site pair. The important point is that vanishing asymmetry does not require detailed balance (Corberi et al., 2024).

In asymmetric nuclear matter, asymmetry means unequal neutron and proton densities, $g(x)$5, and the static response becomes channel dependent. The finite-particle DFT study finds that the isoscalar response is comparatively regular, while the isovector response $g(x)$6 exhibits a non-monotonic peak whose position moves to larger $g(x)$7 and whose magnitude grows as $g(x)$8 increases. The analysis traces this to the competition of $g(x)$9 with the mixed neutron-proton term $\delta_g$0, whose sign change at relatively small $\delta_g$1 affects $\delta_g$2 but not $\delta_g$3 in the same way (Li et al., 5 Aug 2025).

In quasar BAL plasma, the asymmetry is both sign-based and timescale-based. The ionic population equation

$\delta_g$4

leads, after a sudden continuum change, to the characteristic timescale

$\delta_g$5

For low-ionization states $\delta_g$6 is recombination-limited, while for high-ionization states it is ionization-limited and shorter. This asymmetry in response time is used to explain why more than $\delta_g$7 of BAL gases exhibit a negative response and why the observed asymmetry is consistent with a typical density upper limit of $\delta_g$8 (He et al., 2023).

6. Surrogate modeling, strategic interaction, and macro-financial dynamics

In Bayesian optimization, asymmetry is introduced directly into the kernel that shapes the Gaussian-process surrogate response surface. The Asymmetric Elastic Net RBF kernel is

$\delta_g$9

with

$\Delta \mathcal R(\omega,\mu)$0

The kernel is valid, reduces to the standard RBF in the symmetric Euclidean special case, and is argued to yield smaller mean squared prediction error than the baseline RBF under mild unequal-noise conditions. Empirically it is reported to be less sensitive to outliers and to converge faster to the global optimum in Bayesian optimization (AlBahar et al., 2021).

In asymmetric network games, the response object is usually a best-response or gradient-response map rather than a single scalar susceptibility. The asymmetry is quantified by the mismatch of cross-partials, and the paper constructs an $\Delta \mathcal R(\omega,\mu)$1-potential

$\Delta \mathcal R(\omega,\mu)$2

with approximation error

$\Delta \mathcal R(\omega,\mu)$3

In the linear-quadratic specialization this becomes

$\Delta \mathcal R(\omega,\mu)$4

The paper’s response-theoretic point is that directional nonreciprocity prevents exact potentiality but still permits convergence of modified best-response and gradient-play dynamics to approximate equilibria (Rokade et al., 8 Aug 2025).

In international financial spillovers, asymmetry is represented by nonlinear generalized impulse responses

$\Delta \mathcal R(\omega,\mu)$5

generated by a nonlinear conditional mean $\Delta \mathcal R(\omega,\mu)$6. The model permits responses to differ by shock sign, shock magnitude, and calendar date through state dependence. The empirical result is that adverse US financial shocks generate stronger and more persistent declines in output, inflation, and especially stock markets than benign shocks generate improvements, while magnitude asymmetry is present but weaker than sign asymmetry (Huber et al., 2024).

Taken together, the surveyed literature suggests that asymmetric response functions are best understood not as a single special function but as a class of response operators defined by broken exchange symmetry. Their practical consequences recur across fields: bias in estimator centering, side-dependent radiative recoil, odd spatial transfer functions, threshold-like nonlinear transport, phase-shifted critical susceptibilities, non-monotonic isovector peaks, cadence-dependent detectability, and direction-sensitive surrogate modeling. The common theme is that once a symmetry that would equate two perturbations is removed, the response itself becomes an additional source of information rather than merely a deformation of a symmetric baseline.