Parametric Subharmonic Instability (PSI)

• Parametric Subharmonic Instability (PSI) is a resonant three-wave interaction in which a primary wave transfers energy to two lower-frequency daughter waves that satisfy specific phase-matching conditions.
• Laboratory experiments and numerical simulations validate PSI by demonstrating measurable triadic decay with exponential growth rates that align closely with theoretical predictions.
• PSI is pivotal in diverse applications—from ocean mixing and tidal dissipation to optical and boundary-layer phenomena—highlighting its significance in energy cascade and nonlinear saturation processes.

Searching arXiv for recent and foundational papers on parametric subharmonic instability across internal-wave, Faraday, and optomechanical contexts. Parametric subharmonic instability (PSI) is a resonant nonlinear instability in which a finite-amplitude primary wave transfers energy to two lower-frequency secondary waves that satisfy temporal and spatial phase-matching conditions. In the internal-wave literature, PSI is the canonical triadic decay of a primary internal gravity wave into two daughter waves; in adjacent literatures, closely related subharmonic or parametric instabilities are often formulated through Floquet, three-mode, or signal–idler language. Across these formulations, the common structure is selective energy transfer away from a coherent pump or parent state toward new modes with altered frequency, wavenumber, and propagation geometry, often opening a route to smaller scales, dissipation, mixing, or nonlinear saturation (Bourget et al., 2013, Joubaud et al., 2012, Bianchini et al., 14 Jul 2025).

1. Resonant structure and defining conditions

In its classical stratified-fluid form, PSI is a triadic resonance. A primary internal wave with frequency $\omega_0$ and wavevector $\mathbf{k}_0$ excites two secondary waves with $(\omega_1,\mathbf{k}_1)$ and $(\omega_2,\mathbf{k}_2)$ such that

$$\omega_0=\omega_1+\omega_2,\qquad \mathbf{k}_0=\mathbf{k}_1+\mathbf{k}_2.$$

Each member of the triad must also satisfy the internal-wave dispersion relation. In the two-dimensional Boussinesq formulation this may be written as

$\omega^2=N^2\frac{k^2}{k^2+m^2},$

while an equivalent form used in beam geometry is

$$\omega_j=s_j N \frac{|\ell_j|}{\sqrt{\ell_j^2+m_j^2}}.$$

Because the secondary frequencies are lower than the primary frequency and are often near, but not exactly equal to, $\omega_0/2$, the instability is termed “subharmonic” (Joubaud et al., 2012, Bourget et al., 2013).

The linearized growth problem is not determined by resonance alone. Viscous damping, interaction geometry, and parent-wave amplitude enter explicitly. For resonant internal-wave triads, one representative early-stage growth-rate formula is

$$\sigma_\pm = -\frac{\nu}{4}(\kappa_1^2+\kappa_2^2) \pm \sqrt{\frac{\nu^2}{16}(\kappa_1^2-\kappa_2^2)^2+I_1I_2|\Psi_0|^2},$$

with $\kappa_j^2=\ell_j^2+m_j^2$ and $\mathbf{k}_0$0 the interaction coefficients. This formulation makes clear that PSI is strongest only when nonlinear coupling overcomes damping (Bourget et al., 2013).

A common simplification is that the dominant resonance must occur at exact half frequency. That view is explicitly challenged in one nonlinear analysis, which shows that neither the initial growth rate nor the maximum long-term amplification need occur at $\mathbf{k}_0$1. In the examples treated there, the strongest resonances occurred for daughter-wave frequencies around $\mathbf{k}_0$2 to $\mathbf{k}_0$3, with much longer wavelengths than the classical half-frequency picture would suggest (Liang et al., 2017).

2. Laboratory validation in stratified fluids

Controlled laboratory studies established PSI as an experimentally measurable triadic instability rather than a purely formal resonance condition. In a linearly stratified salt-water tank, monochromatic mode-1 or plane-wave forcing produces a clean primary internal wave, after which secondary waves emerge after a finite number of forcing periods. The diagnostics are based on synthetic schlieren, pointwise time-frequency analysis, and Hilbert-transform filtering, which together resolve the onset time, secondary frequencies, propagation directions, and wavevector components (Joubaud et al., 2012, Bourget et al., 2013).

One experiment on propagating mode-1 internal waves reported the first direct measurement of the growth rate of resonant triad instability in this setting. A representative case gave

$\mathbf{k}_0$4

with $\mathbf{k}_0$5, and the measured wavevectors satisfied $\mathbf{k}_0$6 within experimental uncertainty. The secondary-wave amplitude displayed a linear rise on semilog plots, indicating exponential growth, and the measured growth rates agreed quantitatively with theory once the actual primary-wave amplitude was inferred from the measured density field rather than from wavemaker displacement alone (Joubaud et al., 2012).

A complementary plane-wave experiment used a wavelength $\mathbf{k}_0$7 mm, a representative run with $\mathbf{k}_0$8, $\mathbf{k}_0$9, and $(\omega_1,\mathbf{k}_1)$0 cm, and observed secondary peaks after about 10 periods at $(\omega_1,\mathbf{k}_1)$1 and $(\omega_1,\mathbf{k}_1)$2. The measured resonance conditions were satisfied within uncertainty,

$(\omega_1,\mathbf{k}_1)$3

and the predicted growth rate $(\omega_1,\mathbf{k}_1)$4 matched the measured $(\omega_1,\mathbf{k}_1)$5. These studies also showed that laboratory confinement can bias triad selection: daughter waves permitted by infinite-domain theory may not appear if the finite beam cannot support them (Bourget et al., 2013).

3. Geometric focusing, finite width, and attractor destruction

PSI is strongly conditioned by geometry because the primary wave must remain coherent and intense long enough for the daughter pair to amplify. In a trapezoidal, uniformly stratified tank, the internal-wave beam angle obeys

$(\omega_1,\mathbf{k}_1)$6

so repeated reflections focus the wave field onto a closed limit cycle known as an internal wave attractor. Under stronger forcing, the attractor becomes unstable first in its most energetic branch and is then destroyed by PSI through a triadic resonance. A reported case had $(\omega_1,\mathbf{k}_1)$7, $(\omega_1,\mathbf{k}_1)$8, and $(\omega_1,\mathbf{k}_1)$9, with measured wavenumber magnitudes $(\omega_2,\mathbf{k}_2)$0, $(\omega_2,\mathbf{k}_2)$1, and $(\omega_2,\mathbf{k}_2)$2. The associated transfer from about $(\omega_2,\mathbf{k}_2)$3 at injection to about $(\omega_2,\mathbf{k}_2)$4 in the daughter field corresponds to secondary wavelengths roughly 25 times shorter than the forcing scale, and the attractor degrades into small-scale patches and layers (Scolan et al., 2013).

Finite beam width introduces another non-ideal constraint. A simple energy-balance theory for a control region inside the primary beam adds an advection-loss term for the secondary waves,

$(\omega_2,\mathbf{k}_2)$5

so that the daughter waves can exit the interaction region before being strongly amplified. In this picture, narrow beams act as effective stabilizers, not by added dissipation but by reduced residence time of the resonant daughters. Experiments and simulations showed that for $(\omega_2,\mathbf{k}_2)$6 no clear triadic resonance developed, whereas PSI appeared clearly for $(\omega_2,\mathbf{k}_2)$7, and one parameter set exhibited a transition between competing triads around $(\omega_2,\mathbf{k}_2)$8. This modifies both threshold behavior and mode selection relative to the infinite-plane-wave theory (Bourget et al., 2014).

These results collectively shift the interpretation of PSI from a purely local resonance to a resonance-plus-transport problem. A plausible implication is that geometric focusing and geometric escape are dual controls: focusing can create the high-amplitude parent state needed for instability, while finite interaction length can suppress or redirect the resulting triad.

4. Oceanographic, tidal, and boundary-layer realizations

In geophysical settings, PSI is a candidate route from organized internal-wave energy to small-scale turbulence and mixing. One local model of homogeneous tidal deformation treats a stratified Lagrangian fluid parcel subject to periodic stretching and shear, with the tide acting as a parametric pump. Floquet analysis and direct numerical simulation show instability tongues clustering around

$(\omega_2,\mathbf{k}_2)$9

so the excited internal waves resonate at roughly half the tidal forcing frequency. The measured growth rates agreed with Floquet predictions to within about $$\omega_0=\omega_1+\omega_2,\qquad \mathbf{k}_0=\mathbf{k}_1+\mathbf{k}_2.$$0, the maximum growth rate scaled approximately linearly with ellipticity as $$\omega_0=\omega_1+\omega_2,\qquad \mathbf{k}_0=\mathbf{k}_1+\mathbf{k}_2.$$1, and the saturated state was a weak internal-wave turbulence with spectra close to

$$\omega_0=\omega_1+\omega_2,\qquad \mathbf{k}_0=\mathbf{k}_1+\mathbf{k}_2.$$2

and mixing efficiency $$\omega_0=\omega_1+\omega_2,\qquad \mathbf{k}_0=\mathbf{k}_1+\mathbf{k}_2.$$3 for a superposition of weak linear internal waves (Reun et al., 2017).

For coherent internal tides, PSI can be diagnosed spectrally. Bispectral analysis recasts the nonlinear transfer terms in the Boussinesq energy equations as third-order spectral moments and therefore provides both a detector and a transfer-rate estimator. For an $$\omega_0=\omega_1+\omega_2,\qquad \mathbf{k}_0=\mathbf{k}_1+\mathbf{k}_2.$$4 parent tide, the resonance geometry implies a latitude constraint: PSI can occur only equatorward of about $$\omega_0=\omega_1+\omega_2,\qquad \mathbf{k}_0=\mathbf{k}_1+\mathbf{k}_2.$$5, where $$\omega_0=\omega_1+\omega_2,\qquad \mathbf{k}_0=\mathbf{k}_1+\mathbf{k}_2.$$6. In an idealized non-hydrostatic Boussinesq model, the time-integrated transfer bispectrum matched the growth of daughter-wave energy well and identified $$\omega_0=\omega_1+\omega_2,\qquad \mathbf{k}_0=\mathbf{k}_1+\mathbf{k}_2.$$7 as the dominant nonlinear term in that configuration, whereas application to HOME velocity profiles remained inconclusive because the bispectra were noisy and the realizations limited (Frajka-Williams et al., 2014).

PSI can also enable nonlinear penetration across stratification interfaces that are opaque in linear theory. In a two-layer experiment with $$\omega_0=\omega_1+\omega_2,\qquad \mathbf{k}_0=\mathbf{k}_1+\mathbf{k}_2.$$8, the primary high-frequency wave was trapped in the upper layer, but PSI generated daughter waves at approximately $$\omega_0=\omega_1+\omega_2,\qquad \mathbf{k}_0=\mathbf{k}_1+\mathbf{k}_2.$$9 and $\omega^2=N^2\frac{k^2}{k^2+m^2},$0 from a parent $\omega^2=N^2\frac{k^2}{k^2+m^2},$1. Because $\omega^2=N^2\frac{k^2}{k^2+m^2},$2, the daughters could propagate into the weakly stratified lower layer. The measured transmission was about $\omega^2=N^2\frac{k^2}{k^2+m^2},$3 of the primary-wave power via the $\omega^2=N^2\frac{k^2}{k^2+m^2},$4 daughter and about $\omega^2=N^2\frac{k^2}{k^2+m^2},$5 via $\omega^2=N^2\frac{k^2}{k^2+m^2},$6, in contrast with the near-zero linear prediction for the primary wave (Ghaemsaidi et al., 2016).

A more recent bottom-boundary-layer formulation shows that PSI can occur for locally near-inertial waves when baroclinic bottom boundary layers reduce Ertel potential vorticity and hence lower the minimum allowable internal-wave frequency,

$\omega^2=N^2\frac{k^2}{k^2+m^2},$7

In the idealized arrested-Ekman-layer model, the modified near-inertial frequency is

$\omega^2=N^2\frac{k^2}{k^2+m^2},$8

and PSI occupies the window

$\omega^2=N^2\frac{k^2}{k^2+m^2},$9

between stable conditions and symmetric instability. Linear stability analysis and nonlinear simulations show that wave shear production is the primary energy source, buoyancy production can contribute positively depending on slope parameters, geostrophic shear production is a partially compensating sink, and PSI growth rates are typically $$\omega_j=s_j N \frac{|\ell_j|}{\sqrt{\ell_j^2+m_j^2}}.$$0 (Knudsen et al., 27 May 2026).

5. Analytical and diagnostic frameworks

The modern analysis of PSI uses several distinct but complementary frameworks. In periodic or parametrically forced free-surface problems, a weakly nonlinear multiple-time-scale expansion yields amplitude equations of normal-form type. For vertically vibrated circular cylinders with pinned contact line, the standing-wave amplitude satisfies

$$\omega_j=s_j N \frac{|\ell_j|}{\sqrt{\ell_j^2+m_j^2}}.$$1

This formulation accommodates viscosity, contact-angle effects, and static meniscus geometry. In the brimful case $$\omega_j=s_j N \frac{|\ell_j|}{\sqrt{\ell_j^2+m_j^2}}.$$2, the base state is flat and there are no harmonic meniscus waves; in the nearly-brimful case $$\omega_j=s_j N \frac{|\ell_j|}{\sqrt{\ell_j^2+m_j^2}}.$$3, harmonic meniscus waves exist below threshold, the Faraday branch bifurcates from an oscillatory base state rather than from rest, and the bifurcation diagram becomes imperfect with a tailing effect. The nonlinear coefficient $$\omega_j=s_j N \frac{|\ell_j|}{\sqrt{\ell_j^2+m_j^2}}.$$4 varies with static contact angle, so the bifurcation can change between supercritical and subcritical behavior, and for some modes decreasing $$\omega_j=s_j N \frac{|\ell_j|}{\sqrt{\ell_j^2+m_j^2}}.$$5 can suppress subcriticality entirely (Bongarzone et al., 2022).

In internal-tide analysis, the bispectrum and bicoherence provide a direct spectral representation of triadic phase locking. The transfer bispectrum links the real part of a third-order spectral quantity to net energy transfer and the imaginary part to phase evolution, so it can quantify not only whether a PSI triad exists but also how strongly it is exchanging energy. This is especially useful when frequency resonance alone is ambiguous, because PSI is fundamentally a joint frequency–wavenumber phenomenon rather than a purely spectral coincidence (Frajka-Williams et al., 2014).

A different line of work has supplied a rigorous spectral justification of PSI for the inviscid two-dimensional Boussinesq system. Linearization about a small-amplitude periodic traveling wave, followed by Floquet–Bloch decomposition on the torus, isolates a double purely imaginary eigenvalue at a resonant Floquet parameter satisfying

$$\omega_j=s_j N \frac{|\ell_j|}{\sqrt{\ell_j^2+m_j^2}}.$$6

Using Riesz projectors and Kato-type similarity ideas, the problem reduces to a $$\omega_j=s_j N \frac{|\ell_j|}{\sqrt{\ell_j^2+m_j^2}}.$$7 spectral block whose eigenvalues take the form

$$\omega_j=s_j N \frac{|\ell_j|}{\sqrt{\ell_j^2+m_j^2}}.$$8

Since $$\omega_j=s_j N \frac{|\ell_j|}{\sqrt{\ell_j^2+m_j^2}}.$$9 on the relevant resonant set away from a discrete exceptional set, one eigenvalue acquires positive real part. This is presented as the first rigorous justification of PSI of inviscid internal waves, and it is particularly relevant where viscous effects are negligible (Bianchini et al., 14 Jul 2025).

6. Terminological scope and adjacent parametric instabilities

The term “PSI” is not interchangeable with every use of “parametric instability.” In cavity optomechanics, for example, “parametric instability” in one microtoroid COEMS study denotes the standard dynamical-backaction instability in which radiation pressure provides negative damping to a mechanical mode. There the instability occurs when the modified mechanical linewidth becomes negative; the unstable motion saturates, detuning changes, intracavity power drops, and transduction becomes strongly nonlinear with harmonics and broadband mixing noise. Feedback through electrical gradient-force actuation cancels the radiation-pressure modification of the mechanical susceptibility at

$\omega_0/2$0

yielding a final displacement sensitivity of $\omega_0/2$1 and a $\omega_0/2$2-fold improvement. The paper states explicitly that this is not PSI in the usual subharmonic triadic sense (Harris et al., 2011).

By contrast, three-mode parametric instability in long optical cavities is presented as conceptually close to PSI because a pump optical mode, a higher-order transverse optical mode, and a mirror acoustic mode form a resonant three-mode process. The resonance condition

$\omega_0/2$3

and the threshold $\omega_0/2$4 describe exponential acoustic-mode growth, while dynamic transverse-mode frequency modulation due to mirror figure errors and spot motion suppresses the effective gain. The paper argues that suppression factors of $\omega_0/2$5–$\omega_0/2$6 could be achieved for individual unstable modes in Advanced LIGO, and much larger suppression is possible in narrow-linewidth coupled cavities (Zhao et al., 2015).

Several additional systems are best described as PSI-adjacent rather than as classical internal-wave PSI. A strongly pumped SiC polariton medium supports a Floquet parametric instability in which signal and idler satisfy

$\omega_0/2$7

with a predicted instability threshold near $\omega_0/2$8; the authors interpret this as resonant scattering of pump polaritons into pairs of finite-momentum polaritons (Sugiura et al., 2019). A vibrating odd-viscous film over an inclined bed exhibits subharmonic and harmonic resonance tongues in a Floquet analysis of a time-dependent Orr–Sommerfeld problem, with odd viscosity tending to suppress both subharmonic and harmonic resonance (Hossain et al., 2024). A two-dimensional channel-flow study identifies a subharmonic torsional Floquet mode with half-wavelength shift and growth rate $\omega_0/2$9 at $$\sigma_\pm = -\frac{\nu}{4}(\kappa_1^2+\kappa_2^2) \pm \sqrt{\frac{\nu^2}{16}(\kappa_1^2-\kappa_2^2)^2+I_1I_2|\Psi_0|^2},$$0, but it explicitly states that this is not classical PSI in the usual wave–wave interaction sense (Han et al., 22 Apr 2026). The parametrically driven, damped nonlinear Schrödinger equation likewise supports phase-locked subharmonic stationary solitons whose stability is controlled by forcing, damping, and spectral quartets, a construction that is mathematically analogous to PSI without being the same hydrodynamic mechanism (Carreño-Navas et al., 2024).

Taken together, these distinctions matter because “subharmonic,” “parametric,” and “three-wave” describe overlapping but non-identical categories. In the strict stratified-fluid usage, PSI denotes resonant transfer from a primary wave to two lower-frequency daughters satisfying triadic matching. In broader usage, the term may be extended by analogy to Floquet period-doubling, signal–idler generation, or other parametric resonances, but the underlying coupling mechanism, threshold condition, and nonlinear saturation pathway must then be specified explicitly.