FINESSE: Resonators, Simulations & Beyond

• FINESSE is a multifaceted term describing optical finesse in resonators as well as platforms for interferometer simulations, exoplanet spectroscopy, and specialized benchmark suites.
• In optics, finesse quantifies the ratio of free spectral range to resonance linewidth, linking mirror reflectivity, photon lifetime, and intracavity field buildup.
• High finesse enables enhanced optomechanical coupling, precise laser metrology, and improved simulation of complex interferometric systems across various applications.

FINESSE appears in the cited literature in two principal senses. In optics, it denotes the finesse $F$ or $\mathcal{F}$ of a resonator: the ratio between free spectral range and resonance linewidth, and therefore a direct measure of round-trip loss, photon lifetime, spectral selectivity, and intracavity field buildup. In uppercase, FINESSE also names several research systems and workflows, including a frequency-domain interferometer simulation program, a NASA exoplanet survey concept, an FEBio shape-enforcement method for heart valves, and a hierarchical benchmark suite for financial LLMs (Rochau et al., 2021, Freise et al., 2013, Zellem et al., 2018, Laurence et al., 2024, Stanishevskii et al., 14 May 2026).

1. Optical finesse as a resonator metric

In Fabry–Pérot usage, finesse is defined as the ratio between the free spectral range and the linewidth of a cavity resonance, either in angular-frequency form,

$\mathcal{F}=\frac{\omega_{\mathrm{fsr}}}{\kappa},$

or in ordinary-frequency form,

$\mathcal{F}=\frac{f_{\mathrm{FSR}}}{\delta f}.$

The same literature also relates finesse to mirror reflectivity and total round-trip loss. For a symmetric cavity with negligible internal loss,

$\mathcal{F}=\frac{\pi\sqrt{R}}{1-R},$

while in the high-finesse limit it is commonly written as an inverse-loss relation, such as $\mathcal{F}\approx 2\pi/L$ for total round-trip power loss $L$ (Rochau et al., 2021, Hond et al., 2017, Truong et al., 2022).

These equivalent definitions encode the same physical content. A high-finesse cavity has a narrow linewidth for a given free spectral range, long photon storage time, and large intracavity intensity buildup. Several of the cited papers explicitly connect finesse to photon lifetime through ring-down time or cavity decay rate, and to the sharpness of the resonance used in spectroscopy, optomechanics, or stabilization (Jin et al., 2022, Hindle et al., 2019, Ding et al., 28 Sep 2025).

The distinction between finesse and quality factor is likewise standard. $Q$ counts optical cycles within a resonance linewidth, whereas finesse compares the linewidth to the spacing between adjacent longitudinal modes. In short cavities the two can become numerically similar, but they remain conceptually distinct because finesse is fundamentally tied to round-trip loss and free spectral range rather than to absolute optical frequency (Fait et al., 2021).

2. Reported values across cavity platforms

Reported finesse values in the cited literature span more than three orders of magnitude and cover reference cavities, fiber Fabry–Pérot resonators, open-access microcavities, photonic-crystal-slab cavities, terahertz resonators, and ultralow-loss mirror platforms.

A medium-finesse optical cavity for simultaneous stabilization of 960 nm and 780 nm Rydberg lasers was designed and characterized at $\mathcal{F}\approx 1500$ (Hond et al., 2017). At much higher values, a fiber-based Fabry–Pérot microcavity for optomechanics reached an empty-cavity finesse of $204{,}000$ and a loaded finesse of $\mathcal{F}$0, with radiation-pressure backaction demonstrated in a regime of ultrahigh finesse up to $\mathcal{F}$1 (Rochau et al., 2021). A bow-tie cavity for Rydberg atom arrays reported a finesse of $\mathcal{F}$2 at 852 nm with a waist of $\mathcal{F}$3 (Chen et al., 2022). A high-finesse free-space cavity incorporating a Si$\mathcal{F}$4N$\mathcal{F}$5 photonic-crystal slab reached $\mathcal{F}$6 at $\mathcal{F}$7 nm (Chen et al., 2016). In the terahertz domain, a 620 GHz Fabry–Pérot cavity based on an oversized corrugated waveguide and photonic mirrors obtained finesse above $\mathcal{F}$8 (Hindle et al., 2019).

Open-access and microfabricated platforms push farther. An O-band open microcavity platform reported finesse approaching $\mathcal{F}$9 together with mode volumes $\mathcal{F}=\frac{\omega_{\mathrm{fsr}}}{\kappa},$0 (Fait et al., 2021). Mid-infrared supermirrors near $\mathcal{F}=\frac{\omega_{\mathrm{fsr}}}{\kappa},$1 yielded $\mathcal{F}=\frac{\omega_{\mathrm{fsr}}}{\kappa},$2 for all-crystalline mirrors and $\mathcal{F}=\frac{\omega_{\mathrm{fsr}}}{\kappa},$3 for hybrid crystalline–amorphous mirrors (Truong et al., 2022). Micro-fabricated mirrors produced compact Fabry–Pérot resonators with finesse values as high as $\mathcal{F}=\frac{\omega_{\mathrm{fsr}}}{\kappa},$4 million and measured excess loss $\mathcal{F}=\frac{\omega_{\mathrm{fsr}}}{\kappa},$5 ppm, with an average coating-limited finesse of $\mathcal{F}=\frac{\omega_{\mathrm{fsr}}}{\kappa},$6 million across 43 devices (Jin et al., 2022). Buckled dielectric membrane mirrors extended this regime to visible/near-infrared microcavities, achieving a record finesse of $\mathcal{F}=\frac{\omega_{\mathrm{fsr}}}{\kappa},$7 at 780 nm and packaged devices with linewidths of $\mathcal{F}=\frac{\omega_{\mathrm{fsr}}}{\kappa},$8 MHz and $\mathcal{F}=\frac{\omega_{\mathrm{fsr}}}{\kappa},$9 kHz (Ding et al., 28 Sep 2025).

This numerical spread reflects differing design targets rather than a single monotonic performance hierarchy. Medium finesse can be optimal where robustness, broad lock range, or affordability dominate; ultrahigh finesse becomes decisive when weak light–matter interaction, path-length enhancement, or extreme spectral selectivity is the limiting constraint (Hond et al., 2017, Truong et al., 2022).

3. What high finesse enables in experiment

In cavity optomechanics, high finesse directly boosts circulating photon number and narrows the optical linewidth. In an ultrahigh-finesse fiber microcavity coupled to a silicon nitride membrane stripe, this enabled static optomechanical coupling as large as $\mathcal{F}=\frac{f_{\mathrm{FSR}}}{\delta f}.$0, observable radiation-pressure backaction, self-oscillation on the blue side, and cooling of a mechanical mode from $\mathcal{F}=\frac{f_{\mathrm{FSR}}}{\delta f}.$1 K to an effective $\mathcal{F}=\frac{f_{\mathrm{FSR}}}{\delta f}.$2 K on the red side (Rochau et al., 2021). The same work treats single-photon cooperativity as a key parameter and emphasizes that high finesse improves sensitivity by increasing phase fluctuations in the output field.

In cavity QED and atom–photon interfaces, finesse enters through the cavity decay rate $\mathcal{F}=\frac{f_{\mathrm{FSR}}}{\delta f}.$3 and therefore through cooperativity. The bow-tie cavity for Rydberg arrays was designed for strong single-atom–single-photon coupling and a cooperativity per traveling mode of $\mathcal{F}=\frac{f_{\mathrm{FSR}}}{\delta f}.$4 at the cavity waist (Chen et al., 2022). In laser metrology, a much lower finesse can still be sufficient when matched to the servo architecture: the Rydberg-laser reference cavity with $\mathcal{F}=\frac{f_{\mathrm{FSR}}}{\delta f}.$5 produced a cavity linewidth of about $\mathcal{F}=\frac{f_{\mathrm{FSR}}}{\delta f}.$6 MHz and reduced the linewidths of the 960 nm and 780 nm lasers to about $\mathcal{F}=\frac{f_{\mathrm{FSR}}}{\delta f}.$7–$\mathcal{F}=\frac{f_{\mathrm{FSR}}}{\delta f}.$8 kHz and $\mathcal{F}=\frac{f_{\mathrm{FSR}}}{\delta f}.$9 kHz, respectively, with residual drift limited to about $\mathcal{F}=\frac{\pi\sqrt{R}}{1-R},$0 MHz/day (Hond et al., 2017).

In spectroscopy, finesse controls both effective path length and detection floor. The mid-infrared mirror work translated low ppm losses into an effective path length of order hundreds of kilometers in a $\mathcal{F}=\frac{\pi\sqrt{R}}{1-R},$1 m cavity and a minimum detectable absorption of $\mathcal{F}=\frac{\pi\sqrt{R}}{1-R},$2 at 8 s averaging (Truong et al., 2022). The terahertz Fabry–Pérot resonator provided an equivalent interaction length of about one kilometer, line-intensity access down to $\mathcal{F}=\frac{\pi\sqrt{R}}{1-R},$3 at signal-to-noise ratio 3, and Lamb-dip spectroscopy with absolute frequency accuracy better than $\mathcal{F}=\frac{\pi\sqrt{R}}{1-R},$4 kHz (Hindle et al., 2019).

In molecular strong coupling, the cited literature adds a further refinement: strong coupling in reflectance does not automatically imply strong coupling in emission. A multimode analysis of cavity photoluminescence showed that lower finesse reduces the extent of light–matter mixing in polariton states and proposed an “effective strong coupling” condition requiring both

$\mathcal{F}=\frac{\pi\sqrt{R}}{1-R},$5

and

$\mathcal{F}=\frac{\pi\sqrt{R}}{1-R},$6

Under that criterion, sufficient finesse is necessary for a clearly dispersive lower polariton to appear in emission (Menghrajani et al., 2022).

4. FINESSE as interferometer simulation software

FINESSE also denotes “Frequency domain INterferomEter Simulation SoftwarE,” a fast interferometer simulation program for steady-state optical systems (Freise et al., 2013). For a given optical setup it computes light-field amplitudes at every point in the interferometer by translating the optical description into a set of linear equations and solving them numerically. The program can work in the plane-wave approximation or with Hermite–Gauss modes, allowing explicit treatment of mode matching, mirror angular positions, telescopes, and higher-order spatial modes.

The software automates several standard analyses: modulation–demodulation error signals, transfer functions, shot-noise-limited sensitivities, and beam shapes (Freise et al., 2013). Its component-based description uses spaces, mirrors, beam splitters, lasers, modulators, lenses, gratings, and detectors connected by nodes, while Gaussian-beam propagation is handled through the usual $\mathcal{F}=\frac{\pi\sqrt{R}}{1-R},$7-parameter and ABCD-matrix formalism. In this usage, FINESSE is not a cavity metric but a numerical environment for interferometric modeling.

A separate validation study compared FINESSE simulations to analytical solutions for gravitational-wave responses in a single space, mirror-reflected arm, Fabry–Pérot cavity, Michelson interferometer, and Sagnac interferometer, with and without arm cavities (Bond et al., 2013). In that work gravitational waves are represented as phase modulations of “space” elements, and the simulated amplitude and phase of the induced sidebands agree with the analytic response functions. The software therefore occupies a distinctive role in precision interferometry: it is a modeling layer that connects cavity physics, RF sensing, and full interferometer topology.

5. FINESSE as a space mission concept for exoplanets

In exoplanet science, FINESSE denotes the “Fast Infrared Exoplanet Spectroscopy Survey Explorer,” a proposed NASA Medium-Explorer mission that had advanced to Step 2 of the Explorer program at the time of the cited white paper (Zellem et al., 2018). Its central objective is a large, uniform spectroscopic survey of exoplanet atmospheres that is sufficiently homogeneous to support comparative planetology rather than one-off case studies.

The mission concept specifies a $\mathcal{F}=\frac{\pi\sqrt{R}}{1-R},$8 m space telescope feeding a single $\mathcal{F}=\frac{\pi\sqrt{R}}{1-R},$9–$\mathcal{F}\approx 2\pi/L$0 spectrograph in an L2 halo orbit (Zellem et al., 2018). The survey plan calls for transmission spectroscopy of at least $\mathcal{F}\approx 2\pi/L$1 planets and phase-resolved emission spectroscopy for about $\mathcal{F}\approx 2\pi/L$2 planets. The wavelength coverage is designed to include major bands of H$\mathcal{F}\approx 2\pi/L$3O, CH$\mathcal{F}\approx 2\pi/L$4, CO$\mathcal{F}\approx 2\pi/L$5, CO, NH$\mathcal{F}\approx 2\pi/L$6, Na, K, TiO, VO, and several disequilibrium species, enabling simultaneous constraints on metallicity, C/O ratio, aerosols, albedo, and heat redistribution.

The mission paper frames FINESSE as a way to turn exoplanet atmosphere studies into a statistical science (Zellem et al., 2018). Its explicit scientific targets include mass–metallicity relations, C/O-driven formation inferences, climate and energy-balance diagnostics, disequilibrium chemistry, and population-level trends across hot Jupiters, warm Neptunes, sub-Neptunes, and super-Earths. In this sense, FINESSE is a survey architecture rather than a data-analysis code or a cavity parameter.

6. Other acronymic usages

The term is also used for a biomechanical pipeline and for an LLM benchmark suite.

Usage	Expansion or title	Scope
FEBio FINESSE	Finite Element Simulations with Shape Enforcement	Heart valve strain estimation from 3D echocardiography
FINESSE-Bench	Hierarchical benchmark suite for finance	Financial domain knowledge and technical analysis in LLMs

FEBio FINESSE is an open-source finite-element workflow built on FEBio for estimating in vivo heart-valve strains from 3D echocardiography (Laurence et al., 2024). It runs a forward valve-closure simulation from an open reference configuration and then enforces agreement with the image-derived closed geometry by a sliding-elastic surface-to-surface penalty contact. Synthetic test cases suggest that the method can estimate first principal leaflet strains within $\mathcal{F}\approx 2\pi/L$7 strain, and pediatric patient studies reported target-surface matching with median errors similar to or less than the smallest voxel dimension (Laurence et al., 2024). The method was presented specifically as a way to proceed in the absence of patient-specific material properties, leaflet thickness, and chordae tendineae structures.

FINESSE-Bench is a benchmark suite of eight specialized datasets comprising 3,993 questions for hierarchical evaluation of financial competencies in LLMs (Stanishevskii et al., 14 May 2026). It combines CFA-like Levels 1–3, a CMT-like Level 2 dataset, a CFTe-like Level 1 dataset, applied trading collections, and a Russian-language olympiad benchmark. The evaluation protocol covers multiple-choice questions, numerical answers, and short open-ended responses, with automated free-form scoring based on the LLM-as-judge paradigm (Stanishevskii et al., 14 May 2026). In this usage, FINESSE names neither an optical quantity nor a physical instrument, but a structured evaluation framework for professional finance competence.

Across these usages, the shared label does not imply a shared technical domain. In the optical literature, finesse is a measurable resonator property tied to linewidth and loss; in the software, mission, biomedical, and benchmark literatures, FINESSE is an acronymic name applied to specific platforms, workflows, or surveys.