Selective Phase Resonance

• Selective Phase Resonance is a phenomenon where the phase variable, rather than amplitude, identifies and tunes resonant interactions in complex systems.
• It leverages rapid phase evolution and locking in specific channels to selectively isolate resonant modes in mechanical, optical, and quantum frameworks.
• Applications include nonlinear modal testing, optical phase shifters, and qubit-resonator coupling, demonstrating its impact on signal processing and device programmability.

Taken together, the cited works indicate that selective phase resonance can denote resonance phenomena in which phase is the variable through which a resonant interaction is identified, selected, tuned, or read out, rather than treated only as a secondary consequence of an amplitude maximum. In this usage, phase may be the phase of a complex reflection coefficient, the phase lag of a selected harmonic relative to forcing, the phase of a multiphoton transition amplitude, the phase of a guided or spin-wave mode, or a transient optical phase delay induced by resonant absorption (Novitsky, 2011, Zhou et al., 2024, Zhang et al., 2018). A common feature is that resonance is resolved in the complex response itself: the relevant observable is not only where a spectrum is large, but where the phase of a selected channel evolves rapidly, locks to a prescribed value, or acquires a controllable shift.

1. Conceptual structure of selective phase resonance

In nonlinear mechanical testing, the most explicit formulation is the use of a resonant phase lag as the control objective. For a $\kappa:\upsilon$ resonance, the relevant oscillatory content is the component at frequency

$$\omega_{\kappa,\upsilon}=\frac{\kappa \omega}{\upsilon},$$

and the phase lag is defined between the selected harmonic component of the response and the fundamental forcing, not between total displacement and total force. In that framework, phase is both a resonance indicator and a continuation coordinate, so resonance is selected by locking to the phase of the component that is actually resonating (Zhou et al., 2024).

A closely related formulation appears in nonlinear modal theory. For forced Duffing-type systems, phase resonance is defined through the phase lag $\phi_k$ between the forcing and the $k$-th harmonic of the response. The phase condition is not universal across resonance families: phase resonance occurs at

$\phi_k=\frac{\pi}{2}$

when $k$ and $\nu$ are odd, and at

$\phi_k=\frac{3\pi}{4\nu}$

when either $k$ or $\nu$ is even (Volvert et al., 2020, Volvert et al., 2022).

In wave and optical systems, the same idea appears in a different language. In selective reflection from a resonant two-level medium in a dielectric host, the complex reflection coefficient

$$\omega_{\kappa,\upsilon}=\frac{\kappa \omega}{\upsilon},$$0

contains the full phase information, and the resonance becomes asymmetric because the resonant atomic channel interferes with a nonresonant dielectric-interface background. The phase of $$\omega_{\kappa,\upsilon}=\frac{\kappa \omega}{\upsilon},$$1 then shows an asymmetric dispersive evolution rather than a simple symmetric $$\omega_{\kappa,\upsilon}=\frac{\kappa \omega}{\upsilon},$$2-jump (Novitsky, 2011).

A further generalization appears in phase-based resonance diagnostics. For an arbitrary complex response function $$\omega_{\kappa,\upsilon}=\frac{\kappa \omega}{\upsilon},$$3, the dimensionless quality function

$$\omega_{\kappa,\upsilon}=\frac{\kappa \omega}{\upsilon},$$4

extracts resonance characteristics from the phase spectrum itself. This formulation treats resonance as a local property of spectral phase variation and explicitly separates pole and zero contributions (Soltane et al., 2024).

2. Interference, dispersive phase, and antiresonance in wave and optical systems

Selective reflection provides a direct phase-sensitive example. The physical system is a monochromatic plane wave incident normally from an outer medium of refractive index $$\omega_{\kappa,\upsilon}=\frac{\kappa \omega}{\upsilon},$$5 onto two-level atoms embedded in a dielectric host of refractive index $$\omega_{\kappa,\upsilon}=\frac{\kappa \omega}{\upsilon},$$6. In the local-interface limit, the reflection amplitude is

$$\omega_{\kappa,\upsilon}=\frac{\kappa \omega}{\upsilon},$$7

with normalized detuning $$\omega_{\kappa,\upsilon}=\frac{\kappa \omega}{\upsilon},$$8. The dielectric function contains both a nonresonant background and a resonant term,

$$\omega_{\kappa,\upsilon}=\frac{\kappa \omega}{\upsilon},$$9

so the reflected field is the coherent sum of a broad background channel and a narrow resonant channel. The refractive-index difference

$\phi_k$0

acts as the asymmetry parameter: when $\phi_k$1, the resonance is symmetric; when $\phi_k$2, the modulus $\phi_k$3 becomes asymmetric and the phase shows a gain followed by a drop across resonance (Novitsky, 2011).

Two-photon ionization in helium shows the same principle at the level of transition amplitudes. In the perturbative decomposition of the near-resonant two-photon matrix element, the strictly resonant term is real and the principal-value near-resonant term is imaginary. As the XUV harmonic is tuned across the $\phi_k$4 series, the phase of the two-photon amplitude rotates rapidly: at exact resonance the dominant real term gives

$\phi_k$5

whereas away from resonance the phase approaches

$\phi_k$6

Between neighboring resonances, antiresonances appear where near-resonant pathways cancel. There the amplitude minimum is accompanied by a pronounced phase jump, so antiresonance is a phase-selective destructive-interference condition rather than merely a dip in yield (Drescher et al., 2021).

Phase-centered resonance extraction makes this interference structure explicit. From the singularity-and-zero factorization

$\phi_k$7

the phase derivative is

$\phi_k$8

For an individual singularity $\phi_k$9,

$k$0

This shows that resonance signatures are encoded in the phase derivative as Lorentzian-like contributions of poles and zeros; the paper’s central conclusion is that poles alone do not suffice, because zeros must also be taken into account to retrieve the full quality function (Soltane et al., 2024).

3. Phase-conditioned identification and continuation of nonlinear mechanical resonances

In phase-locked-loop testing of nonlinear systems, phase is used as a selective coordinate for resonance continuation. The excitation is

$k$1

and the PLL updates the forcing frequency by

$k$2

The phase detector estimates the chosen harmonic via adaptive Fourier decomposition,

$k$3

and the estimated phase lag relative to the fundamental forcing is

$k$4

Because the selected phase often evolves monotonically even when amplitude-frequency curves fold, the method can track primary, superharmonic, and subharmonic resonances, including branch switching toward detached subharmonic isolas (Zhou et al., 2024).

The modal formulation of the same idea is given by phase resonance nonlinear modes. For single-point, single-harmonic forcing, PRNMs are periodic solutions of a feedback-modified system in which the forcing is replaced by a $k$5-periodic velocity feedback built from the selected harmonic. For resonances with quadrature points,

$k$6

whereas for resonance families without quadrature points the feedback is delayed,

$k$7

with

$k$8

This construction yields periodic solutions that lie on the nonlinear frequency-response curves of the forced system and characterizes primary, superharmonic, subharmonic, and ultra-subharmonic resonances through the appropriate phase lag of the relevant harmonic (Volvert et al., 2020).

The Duffing analysis supplies the explicit resonance conditions. For the forced oscillator

$k$9

the phase lag of the $\phi_k=\frac{\pi}{2}$0-th response harmonic is the selective variable. The paper’s summary statement is that phase resonance occurs at

$\phi_k=\frac{\pi}{2}$1

for odd-odd resonance families and at

$\phi_k=\frac{\pi}{2}$2

when either $\phi_k=\frac{\pi}{2}$3 or $\phi_k=\frac{\pi}{2}$4 is even. This makes phase selectivity harmonic-specific: for a $\phi_k=\frac{\pi}{2}$5 resonance one tracks $\phi_k=\frac{\pi}{2}$6, for a $\phi_k=\frac{\pi}{2}$7 resonance one tracks $\phi_k=\frac{\pi}{2}$8, and so on (Volvert et al., 2022).

The same logic extends from single oscillators to weakly coupled nonlinear arrays. In a parametrically excited van der Pol-Duffing chain driven by a uniform high-frequency signal, fast-scale elimination yields renormalized onsite coefficients

$\phi_k=\frac{\pi}{2}$9

with $k$0. The resulting dispersion shift controls which standing-wave modes satisfy the parametric resonance condition

$k$1

so a uniform high-frequency drive selectively activates particular bulk modes through phase-sensitive subharmonic parametric locking at $k$2 (Roy et al., 4 Dec 2025).

4. Phase-selective transduction, modal control, and programmable devices

Magnonic Fabry–Pérot resonators realize selective phase resonance as a programmable phase shifter. In the reported YIG/CoFeB device, dynamic dipolar coupling converts an incident YIG spin wave into a short-wavelength bilayer mode inside an $k$3-wide cavity. Near the $k$4 Fabry–Pérot resonance, the transmission phase evolves from about $k$5 to more than $k$6 as frequency increases. Because the cavity dispersion depends on magnetic alignment, the parallel and antiparallel states can be placed on opposite sides of the same dispersive phase response. At about $k$7 and $k$8, the two states have nearly equal transmitted amplitudes, about $k$9 of the reference wave, but phases differing by approximately $\nu$0 (Lutsenko et al., 2024).

Silicon photonics provides an integrated guided-wave analogue through the mode-selective thermo-optic phase shifter. The phase shift of a heated waveguide section is

$\nu$1

and the mode selectivity is defined as

$\nu$2

By adding subwavelength grating side regions to a 220 nm SOI waveguide, the device makes TE$\nu$3 overlap more strongly with the central silicon core and TE$\nu$4 overlap more strongly with the SWG regions. The experimentally demonstrated selectivity is $\nu$5 for duty cycle $\nu$6, meaning the thermo-optic coefficient of TE$\nu$7 is 44% larger than that of TE$\nu$8. Cascading the MS-TOPS with a mode-insensitive TOPS gives

$\nu$9

and therefore sufficient degrees of freedom to manipulate the relative phase of each mode independently (Safaee et al., 2023).

In bond-selective transient phase microscopy, the resonant interaction is vibrational absorption and the measured quantity is a transient visible-light phase delay. The optical phase of a transparent object is

$\phi_k=\frac{3\pi}{4\nu}$0

while the pump-induced transient phase is

$\phi_k=\frac{3\pi}{4\nu}$1

Using

$\phi_k=\frac{3\pi}{4\nu}$2

the phase signal becomes

$\phi_k=\frac{3\pi}{4\nu}$3

Hence the transient phase is directly proportional to the IR absorption spectrum, and bond selectivity is achieved because only molecular vibrations resonant with the pump wavenumber produce the corresponding phase response (Zhang et al., 2018).

5. Quantum, vibronic, and Hamiltonian formulations

In slow-fast Hamiltonian systems with one fast angle, the relevant selective quantity is the phase at the instant of arrival at a resonant surface $\phi_k=\frac{3\pi}{4\nu}$4. The paper introduces the pseudophase

$\phi_k=\frac{3\pi}{4\nu}$5

and proves the asymptotic formula

$\phi_k=\frac{3\pi}{4\nu}$6

with $\phi_k=\frac{3\pi}{4\nu}$7 accuracy when the resonant effective potential has no critical points. The resonance outcome is therefore phase-selective in the sense that the phase at resonance is asymptotically determined by the incoming slow trajectory (Gao et al., 2022).

A quantum-search analogue appears in energy-selective search with Ising Hamiltonian phase oracles. The oracle is the physical evolution

$\phi_k=\frac{3\pi}{4\nu}$8

so computational basis states are marked continuously by phase rather than by a Boolean predicate. Alternating $\phi_k=\frac{3\pi}{4\nu}$9 with Grover diffusion produces a resonance band centered at

$k$0

or, with detuning,

$k$1

The generated amplification peak is therefore simultaneously selective in oracle phase and in energy (Plyashechnik et al., 2 Jun 2026).

Circuit QED implements selective phase operations through selective resonance. In the quasi-dispersive regime, the resonator frequency becomes qubit-state dependent,

$k$2

and this amplified qubit-state-dependent resonator transition frequency allows a second qubit to be resonant with the resonator only in chosen computational subspaces. By selecting the interaction time so that the wanted Rabi oscillation completes the appropriate cycle while unwanted oscillations complete an integer number of periods, the protocol realizes c-phase and cc-phase gates without drive fields or direct qubit-qubit interaction (Hua et al., 2013).

Vibronic resonance in excitonically coupled aggregates provides a molecular analogue. After transformation to effective normal modes, donor-to-acceptor transfer is promoted only along specific tuning coordinates, and constructive or destructive interference of vibronic couplings determines which effective modes are active promoter modes. The selectively resonant states are donor-acceptor vibronic superpositions such as

$k$3

so selectivity arises from mode-resolved vibronic resonance together with sign-sensitive interference in the projected couplings (Patra et al., 2022).

6. System dependence, caveats, and broader significance

The cited works also delimit the concept. Phase criteria are often system-dependent rather than universal. In PLL-based experimental continuation, the beam’s $k$4 subharmonic resonance used $k$5 in acceleration phase, not the Duffing-like theoretical value expected from the simple cubic model, and the authors state explicitly that resonant phase lag values depend on the oscillator class and measurement convention (Zhou et al., 2024). Likewise, selective reflection exhibits its cleanest Fano-like interpretation only in the thin-layer local-interface limit; for $k$6 or larger, propagation introduces additional minima and broadens transmission dips (Novitsky, 2011).

Several works also show that phase selectivity is not equivalent to amplitude prominence. In helium ionization, antiresonances can produce sharp phase jumps at amplitude minima (Drescher et al., 2021). In the phase-spectrum framework, the quality function can reveal resonances even when amplitude is nearly constant, and the full characterization requires zeros as well as poles (Soltane et al., 2024). In bond-selective transient phase microscopy, the phase response is proportional to the absorption coefficient $k$7, but the transduction efficiency still depends on thermo-optic and thermal-expansion coefficients, so strong absorption need not imply a large phase signal (Zhang et al., 2018).

A plausible implication is that selective phase resonance is most useful when resonance is distributed across multiple channels or harmonics and amplitude alone is insufficient to identify the operative one. In that setting, phase supplies the selective coordinate: it can distinguish resonant from antiresonant interference, choose the harmonic that is actually resonating, determine arrival conditions at a resonant surface, or program modal and spin-wave phase shifts without suppressing transmission. Across the cited literature, the unifying theme is therefore not a single formalism, but a recurrent strategy: resonance is resolved in the complex response, and the decisive variable is the phase of the specific channel, mode, or harmonic that carries the interaction.