Negative-Mass-Like Oscillator

• Negative-mass-like oscillators are effective harmonic modes whose reversed energetic response distinguishes them from conventional oscillators.
• They are realized via methods such as inverted Hamiltonians, inverted susceptibility, and band curvature engineering in spin, circuit, and matter-wave systems.
• Engineered implementations enable enhanced mode hybridization, quantum noise cancellation, and controlled amplification in both photonic and metamaterial setups.

Searching arXiv for papers on negative-mass-like oscillators and effective negative-mass oscillator realizations. A negative-mass-like oscillator is an effective harmonic mode whose dynamical response has the opposite energetic or susceptibility sign from that of an ordinary positive-mass oscillator. In the literature surveyed here, this designation most often refers not to a literal object with negative inertial mass, but to one of several mathematically distinct constructions: a mode whose excitations lower the total energy, a mode with an inverted linear susceptibility, a collective coordinate with negative band-curvature effective mass, or a normal-mode sector whose Hamiltonian enters with the opposite sign (Kohler et al., 2017). Across atomic spin systems, cavity and circuit platforms, optical-lattice condensates, phononic metamaterials, and formal non-Hermitian or higher-derivative models, the common feature is a sign-reversed oscillator description that changes amplification, mode hybridization, back-action, transport, or spectral stability relative to conventional oscillators (Møller et al., 2016).

1. Definitional framework

The modern usage of “negative-mass-like oscillator” is predominantly effective. In the atomic-spin formulation, the sign reversal is explicit at the Hamiltonian level: for a spin prepared near its highest-energy Zeeman state, the bosonized fluctuation mode obeys $\mathcal H_s=-\hbar \omega_s \hat b^\dagger \hat b$, so adding a bosonic excitation lowers the energy (Kohler et al., 2017). In optical-lattice condensates, the sign reversal is encoded in band curvature through $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$, and for $m^*<0$ the collective excitation is described as a negative-frequency optomechanical mode (Zhang et al., 2013). In driven superconducting circuits, the operational definition is susceptibility-based: $\chi_-(\omega)=-\chi_+(\omega)$, or more generally $$\chi_{\mathcal G}(\omega)=\mathcal G/(\kappa/2+i(\omega-\omega_0))$$ with $\mathcal G<0$ (Rodrigues et al., 2022). In a non-Hermitian oscillator construction, the sign reversal appears as a negative discrete ladder $E_n=-(n+\tfrac12)$ together with real-axis Gaussian decay (Rath, 2015).

Regime	Defining sign reversal	Representative papers
Spin or collective-mode realization	$\mathcal H=-\hbar \omega \hat b^\dagger \hat b$	(Kohler et al., 2017, Møller et al., 2016)
Band-engineered matter-wave mode	$m^*=\hbar^2/\mathcal E''(q_0)<0$	(Zhang et al., 2013)
Driven photonic or circuit mode	$\chi_-=-\chi_+$ or $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$0	(Rodrigues et al., 2022, Johny et al., 11 Nov 2025)
Formal negative-energy oscillator sector	$m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$1 or $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$2	(Rath, 2015, Masterov, 2015, Pavšič, 2016)

This terminology excludes several nearby but non-equivalent notions. The non-Hermitian oscillator with preserved commutator $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$3 is explicitly described as “not literally a ‘negative-mass oscillator’ in the usual mechanical sense” and as distinct from the standard inverted oscillator (Rath, 2015). Likewise, the spin-based AMO realizations repeatedly use “negative mass” in the standard effective sense, not as a statement about literal negative inertia of the atoms (Kohler et al., 2017).

2. Spin realizations, hybridization, and back-action physics

The cleanest experimentally controlled negative-mass-like oscillator is the collective spin mode of an atomic ensemble prepared near the top of a Zeeman ladder. In an ultracold $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$4 gas containing about $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$5 atoms with total spin $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$6, a magnetic field along $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$7 induces Larmor precession at frequency $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$8, and the Holstein–Primakoff reduction gives $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$9 and $m^*<0$0 with $m^*<0$1. For polarization near the highest-energy state, $m^*<0$2, so the spin Hamiltonian becomes $m^*<0$3 (Kohler et al., 2017). Coupling this mode to the cloud’s center-of-mass oscillator, $m^*<0$4, through a driven optical cavity yields a hybrid Hamiltonian

$m^*<0$5

With $m^*<0$6, $m^*<0$7, $m^*<0$8, and $m^*<0$9, the near-resonant hybrid system crosses from ordinary beam-splitter exchange to pair creation (Kohler et al., 2017).

The reduced two-oscillator interaction clarifies the mechanism. For $\chi_-(\omega)=-\chi_+(\omega)$0, the resonant term is $\chi_-(\omega)=-\chi_+(\omega)$1; for $\chi_-(\omega)=-\chi_+(\omega)$2, the resonant term becomes $\chi_-(\omega)=-\chi_+(\omega)$3. The normal-mode frequencies are $\chi_-(\omega)=-\chi_+(\omega)$4, so for the negative-mass case $\chi_-(\omega)=-\chi_+(\omega)$5, and instability occurs when $\chi_-(\omega)=-\chi_+(\omega)$6 (Kohler et al., 2017). Experimentally, the instability was read out through the cavity field, with the correlated mode phase found near $\chi_-(\omega)=-\chi_+(\omega)$7, consistent with the two-mode parametric-amplifier picture (Kohler et al., 2017). The significance is that the sign-reversed spin mode changes hybridization from avoided crossing to coherent gain and correlated excitation.

A related hybrid realization uses a room-temperature cesium spin ensemble as a negative-mass reference frame for a membrane oscillator. The mechanical mode is the $\chi_-(\omega)=-\chi_+(\omega)$8 drum mode of a silicon nitride membrane with $\chi_-(\omega)=-\chi_+(\omega)$9, while the spin quadratures are $$\chi_{\mathcal G}(\omega)=\mathcal G/(\kappa/2+i(\omega-\omega_0))$$0 and $$\chi_{\mathcal G}(\omega)=\mathcal G/(\kappa/2+i(\omega-\omega_0))$$1, obeying $$\chi_{\mathcal G}(\omega)=\mathcal G/(\kappa/2+i(\omega-\omega_0))$$2. For an inverted spin population, the effective Hamiltonian is $$\chi_{\mathcal G}(\omega)=\mathcal G/(\kappa/2+i(\omega-\omega_0))$$3, so the spin susceptibility has the opposite sign to the mechanical one (Møller et al., 2016). The total readout phase quadrature contains the term $$\chi_{\mathcal G}(\omega)=\mathcal G/(\kappa/2+i(\omega-\omega_0))$$4, making quantum back-action cancel when $$\chi_{\mathcal G}(\omega)=\mathcal G/(\kappa/2+i(\omega-\omega_0))$$5 and $$\chi_{\mathcal G}(\omega)=\mathcal G/(\kappa/2+i(\omega-\omega_0))$$6 (Møller et al., 2016). For matched central frequencies, the reported hybrid negative-mass QBA variance was $$\chi_{\mathcal G}(\omega)=\mathcal G/(\kappa/2+i(\omega-\omega_0))$$7, compared with $$\chi_{\mathcal G}(\omega)=\mathcal G/(\kappa/2+i(\omega-\omega_0))$$8 for mechanics alone and $$\chi_{\mathcal G}(\omega)=\mathcal G/(\kappa/2+i(\omega-\omega_0))$$9 for the positive-mass spin configuration; in a detuned setting, the hybrid negative-mass case reached $\mathcal G<0$0 versus $\mathcal G<0$1 for mechanics alone (Møller et al., 2016). These results established the practical metrological meaning of a negative-mass-like oscillator: its response can be used as an anti-noise reference.

3. Engineered photonic and superconducting implementations

Negative-mass-like behavior can also be engineered in purely electromagnetic degrees of freedom. In a superconducting two-mode circuit, a high-frequency Kerr cavity with undriven resonance $\mathcal G<0$2, linewidth $\mathcal G<0$3, and Kerr constant $\mathcal G<0$4 is coupled by photon pressure to a low-frequency mode with $\mathcal G<0$5 and $\mathcal G<0$6. Under strong driving, the high-frequency mode develops signal and idler quasimodes; the idler is described by a generalized susceptibility

$\mathcal G<0$7

with $\mathcal G<0$8 for the effective negative-mass-like branch (Rodrigues et al., 2022). In the supplemental formulation, the corresponding oscillator Hamiltonian is $\mathcal G<0$9, and blue-sideband pumping produces damping rather than antidamping because

$E_n=-(n+\tfrac12)$0

Experimentally, this gave blue-sideband normal-mode splitting and sideband cooling of the low-frequency mode from $E_n=-(n+\tfrac12)$1 to $E_n=-(n+\tfrac12)$2 quanta (Rodrigues et al., 2022). The conceptual importance is that the “mass” sign can be implemented entirely as a drive-induced inversion of susceptibility.

An all-optical version of the same idea replaces the ancilla oscillator by a detuned optical cavity mode. In the all-optical ENMO scheme, the ancilla mode $E_n=-(n+\tfrac12)$3 and meter mode $E_n=-(n+\tfrac12)$4 are coupled through a down-conversion interaction and a beam-splitting interaction, with matching conditions

$E_n=-(n+\tfrac12)$5

and, for coherent quantum noise cancellation,

$E_n=-(n+\tfrac12)$6

The reported in-situ parameter extraction gave $E_n=-(n+\tfrac12)$7, $E_n=-(n+\tfrac12)$8, $E_n=-(n+\tfrac12)$9, $\mathcal H=-\hbar \omega \hat b^\dagger \hat b$0, and thus $\mathcal H=-\hbar \omega \hat b^\dagger \hat b$1 (Johny et al., 11 Nov 2025). The ideal CQNC phase-quadrature spectrum contains a residual back-action term proportional to $\mathcal H=-\hbar \omega \hat b^\dagger \hat b$2, so perfect matching removes it identically (Johny et al., 11 Nov 2025). With the measured ENMO parameters cascaded with a simulated matched optomechanical sensor, the projected performance was a broadband quantum noise reduction of $\mathcal H=-\hbar \omega \hat b^\dagger \hat b$3 dB, corresponding to a $\mathcal H=-\hbar \omega \hat b^\dagger \hat b$4 reduction in quantum back-action noise at the optimal frequency of maximum reduction (Johny et al., 11 Nov 2025). This extends the negative-mass-like oscillator concept from matter or spin systems to susceptibility-engineered optical networks.

4. Band-engineered matter waves and metamaterial oscillators

In matter-wave systems, negative-mass-like oscillators arise from band curvature rather than from Hamiltonian sign reversal in a fixed bare mass. For a one-dimensional collisionless Bose–Einstein condensate in an optical lattice,

$\mathcal H=-\hbar \omega \hat b^\dagger \hat b$5

the envelope dynamics are governed by the effective-mass Hamiltonian

$\mathcal H=-\hbar \omega \hat b^\dagger \hat b$6

with $\mathcal H=-\hbar \omega \hat b^\dagger \hat b$7 (Zhang et al., 2013). In the first band near the zone edge $\mathcal H=-\hbar \omega \hat b^\dagger \hat b$8, the condensate can be prepared with $\mathcal H=-\hbar \omega \hat b^\dagger \hat b$9. Stable trapping then occurs near a maximum of $m^*=\hbar^2/\mathcal E''(q_0)<0$0, and the quantized envelope modes have

$m^*=\hbar^2/\mathcal E''(q_0)<0$1

The cavity-coupled collective excitation is therefore a negative-frequency optomechanical oscillator (Zhang et al., 2013). In the Tsang–Caves construction used in the paper, two oscillators of equal $m^*=\hbar^2/\mathcal E''(q_0)<0$2 and opposite mass signs admit collective variables $m^*=\hbar^2/\mathcal E''(q_0)<0$3 and $m^*=\hbar^2/\mathcal E''(q_0)<0$4 with $m^*=\hbar^2/\mathcal E''(q_0)<0$5, enabling a quantum-mechanics-free subsystem (Zhang et al., 2013). Here the negative mass is an emergent band property of the condensate wave packet.

A mechanically explicit reduced equation appears in a one-dimensional phononic metamaterial built from mass-in-mass resonators and chiral couplings. After eliminating the internal translational and rotational coordinates, the coarse-grained displacement $m^*=\hbar^2/\mathcal E''(q_0)<0$6 obeys

$m^*=\hbar^2/\mathcal E''(q_0)<0$7

with

$m^*=\hbar^2/\mathcal E''(q_0)<0$8

Both $m^*=\hbar^2/\mathcal E''(q_0)<0$9 and $\chi_-=-\chi_+$0 are frequency-dependent and can become negative, zero, or divergent (Zhang et al., 2019). The dispersion relation

$\chi_-=-\chi_+$1

then supports single-negative gaps, double-negative pass bands, a flat band when $\chi_-=-\chi_+$2 and $\chi_-=-\chi_+$3 diverge simultaneously, and a Dirac-like zero-index point when $\chi_-=-\chi_+$4 and $\chi_-=-\chi_+$5 coincide (Zhang et al., 2019). In this setting, “negative mass” means dynamic effective inertia of a reduced lattice degree of freedom, not a sign-flipped bare mass of any component.

5. Formal oscillator models with negative spectra or ghost sectors

Several mathematically controlled models realize negative-mass-like oscillator structure without direct experimental embodiment. A one-dimensional harmonic oscillator subjected to simultaneous non-Hermitian transformations of coordinate and momentum,

$\chi_-=-\chi_+$6

preserves $\chi_-=-\chi_+$7 and produces the transformed Hamiltonian

$\chi_-=-\chi_+$8

Choosing the representation frequency so that either $\chi_-=-\chi_+$9 or $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$00 removes one off-diagonal ladder term and yields the exact spectrum

$m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$01

with real-axis eigenfunctions proportional to $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$02 for $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$03 (Rath, 2015). The central feature is the coexistence of a negative discrete ladder and Gaussian decay on the real line. The central caveat is equally explicit: the spectrum is unbounded below, and the paper does not construct a full metric-operator or CPT inner product (Rath, 2015).

In higher-derivative theories, the negative-mass-like aspect appears as an alternating-sign decomposition into oscillator sectors. The $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$04 supersymmetric Pais–Uhlenbeck oscillator with distinct frequencies reduces to

$m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$05

so some sectors contribute with the opposite sign to the Hamiltonian (Masterov, 2015). Upon quantization, the fermionic anticommutators also alternate in sign, $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$06, and the Fock space contains negative-norm states (Masterov, 2015). The broader review literature stresses the same structural point for the bosonic PU oscillator: in second-order form it behaves as one positive-energy and one negative-energy oscillator, so the free theory can be consistent with positive norms and indefinite energies, but interactions that couple the sectors are unstable unless the interaction potential is bounded from below and above (Pavšič, 2016).

Relativistic oscillator analogies sharpen the limits of the concept. In the one-dimensional Dirac oscillator,

$m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$07

the nonrelativistic limit gives

$m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$08

which resembles a pair of oscillator sectors with opposite mass sign (Zhang et al., 2018). However, the would-be QMFS variables obey $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$09, not zero, and Zitterbewegung or virtual pair creation feeds measurement back-action back into the mean signal beyond the strict nonrelativistic limit (Zhang et al., 2018). This identifies a recurrent theme: many formal negative-mass-like oscillators are exact as algebraic constructions but nontrivial as physical subsystems.

6. Conceptual boundaries, stability, and common misconceptions

A central misconception is to identify every negative-mass-like oscillator with literal negative inertial mass. The surveyed literature repeatedly rejects that identification. The spin and cavity realizations define negative mass through excitation energy and susceptibility sign, not through $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$10 in a Newtonian kinetic term (Kohler et al., 2017). The BEC realization defines it through Bloch-band curvature and negative-frequency envelope quantization, again not through negative bare atomic mass (Zhang et al., 2013). The circuit and all-optical implementations are explicitly response-engineered analogs in which the operational content is $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$11 or $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$12 (Rodrigues et al., 2022). Even the non-Hermitian harmonic-oscillator construction is presented as a PT-symmetric/non-Hermitian analog, “not a literal negative-mass oscillator” (Rath, 2015).

A second misconception is to conflate negative-mass-like behavior with any instability or runaway. Some models are indeed unstable or unbounded below: the non-Hermitian oscillator has $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$13, the supersymmetric Pais–Uhlenbeck oscillator has an energy spectrum unbounded from below together with negative-norm states, and interacting positive/negative-energy PU sectors are unstable unless the potential is bounded from below and above (Rath, 2015). Other models, however, are stable precisely because the sign reversal is effective and restricted: the spin oscillator near the highest-energy Zeeman state is stabilized by preparation near a finite-spin extremum, and the BEC negative-effective-mass mode is stabilized by trapping near a maximum of $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$14 (Kohler et al., 2017). In a dissipative stochastic environment, the environment-induced mass correction

$m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$15

is negative under broad conditions, but the paper emphasizes that this is a negative correction to the inertial term, not a truly negative total mass, and within the controlled adiabatic regime $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$16 (D'Alessio et al., 2014).

A third boundary concerns sign mimicry. One conceptual paper argues that in a charged spin-$m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$17 system, under the condition $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$18, the replacement $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$19 can reproduce the same reduced dynamics as $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$20, and for a null dielectric function $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$21 a plasma of negatively charged particles with positive mass can behave like a positively charged plasma with negative mass (Campos, 2024). The same paper shows that for de Broglie matter waves, taking $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$22 can produce a negative dispersion relation and negative-index-like behavior without assuming $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$23 (Campos, 2024). These are effective-sign analogies, not universal mechanical equivalences. By contrast, a paper on literal negative mass adopts the rule $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$24 for negative mass. A plausible implication is that attaching such a particle to an ordinary spring with $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$25 gives an anti-restoring equation $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$26, so stable oscillation would require a corresponding sign reversal of the restoring structure (Hammond, 2013). That inference helps explain why most practical realizations do not implement literal negative mass at all.

Finally, some superficially similar pathologies are structurally different. The massless harmonic oscillator in the real-time path integral has no negative kinetic term and no inverted potential; its divergent transition element $m^*_{n,q_0}=\hbar^2/\mathcal E_n''(q_0)$27 is attributed instead to contributions from large constant-action manifolds, producing a “quantum runaway” distinct from either a literal negative-mass oscillator or a standard inverted oscillator (Modanese, 2014). This suggests that the encyclopedia entry for negative-mass-like oscillators must be organized not by phenomenological resemblance alone, but by the mechanism of the sign reversal: energetic ordering, susceptibility inversion, band curvature, higher-derivative ghost structure, or non-Hermitian canonical transformation.

In that technical sense, the term denotes a family of sign-reversed oscillator constructions rather than a single model. What unifies them is that the oscillator’s effective response is opposite to the conventional one; what differentiates them is whether the sign reversal lives in the Hamiltonian, the susceptibility, the band curvature, the metric of state space, or the reduced effective medium.