Colibri in Astronomy, Astrophysics & Computing

• Colibri is a multidisciplinary term used in various fields such as astronomy, astrophysics, and computing to denote distinct systems and methodologies.
• In astronomy, Colibri encompasses real-time multimessenger alert platforms, fast-photometry occultation arrays, and robotic follow-up telescopes designed to capture transient phenomena rapidly.
• In computational contexts, Colibri identifies neuromorphic edge systems, manycore synchronization mechanisms, and specialized software tools that optimize data analysis and secure communications.

Colibri is a recurrent designation in contemporary research rather than a single object. In astronomy, Astro-COLIBRI is often colloquially referred to as “Colibri” in the community and functions as a real-time platform for transient and multimessenger alerts, while the same name is also used for a fast-photometry occultation array, the 1.3 m robotic COLIBRI follow-up telescope in the SVOM ground segment, and the COLIBRI code for thermally pulsing asymptotic giant branch evolution. Outside astronomy, the designation appears in neuromorphic edge systems, manycore synchronization, parton-distribution-function fitting, lightweight Internet Key Exchange, and fuzzy color representation (Reichherzer et al., 2021, Mazur et al., 2022, Marigo et al., 2013, Rutishauser et al., 2023, Costantini et al., 3 Oct 2025).

1. Nomenclature and scope

The name has distinct, domain-specific meanings. Astro-COLIBRI expands to COincidence LIBrary for Real-time Inquiry and was introduced by Reichherzer et al. and summarized further by Schüssler et al. as a modular, cloud-based alert platform for multimessenger and time-domain astrophysics (Reichherzer et al., 2021, Schüssler et al., 2021). In observational astronomy, “Colibri” also names a dedicated telescope array for Kuiper Belt and trans-Neptunian object detection through serendipitous stellar occultations (Pass et al., 2017, Mazur et al., 2022). In the SVOM mission ground segment, COLIBRI denotes a robotic 1.3 m facility at San Pedro Mártir, with DDRAGO optical channels and the CAGIRE near-infrared imager, built for rapid follow-up of gamma-ray bursts and other transients (Basa et al., 27 Apr 2026, Flèche et al., 2023).

In stellar astrophysics, COLIBRI is a fast envelope-based evolutionary code for the TP-AGB phase, tightly coupled in later work to PARSEC tracks and PARSEC-COLIBRI isochrones (Marigo et al., 2013, Marigo et al., 2017). In computing and engineering, the same designation identifies a RISC-V neuromorphic embedded platform, a UAV implementation of that platform, a distributed hardware mechanism for LRwait reservations, an open-source Python package for fast PDF fits, an implementation of the Minimal IKE protocol, and a fuzzy color model based on human categorization experiments (Rutishauser et al., 2023, Bian et al., 2023, Riedel et al., 2024, Costantini et al., 3 Oct 2025, Zuane et al., 25 Feb 2026, Shamoi et al., 15 Jul 2025).

This multiplicity is not incidental. The sources show that “Colibri” operates as a reusable project name across astronomy, astrophysical modeling, and computer systems, so the term requires immediate domain qualification in technical writing.

2. Astro-COLIBRI and the multimessenger alert layer

Astro-COLIBRI addresses what its authors describe as the “last-mile problem” in multimessenger astronomy: alert generation and machine-to-machine coordination may be automated, but human observers still need rapid, comprehensible summaries with context and observability information (Reichherzer et al., 2021). The platform evaluates alerts in real time, filters them by user-defined criteria, and places each event into its multiwavelength and multimessenger context. Supported classes include gamma-ray bursts, fast radio bursts, active galactic nucleus flares, supernovae, tidal disruption events, gravitational-wave events, and high-energy neutrino alerts from streams such as IceCube, ANTARES, and GVD (Reichherzer et al., 2021).

Its architecture is explicitly modular. The 2021 system description centers on a Flask-based REST API, a continuously running event listener, MongoDB as the static or authoritative event store, Firebase Firestore as the real-time database, Firebase Cloud Messaging for push dissemination, and a single Flutter codebase for web, iOS, and Android clients (Reichherzer et al., 2021). The listener ingests VOEvent alerts via Comet, parses TNS notifications and GCN circulars with a custom Python email parser, and subscribes to at least three VOEvent brokers simultaneously for redundancy. The API enriches incoming notices with third-party metadata, merges updates into coherent event records, and exposes public endpoints including cone searches and visibility plots. Later descriptions add CBAT, ZTF via Fink, SIMBAD, NED, Gaia, VizieR, WISE host diagnostics, and tilepy-based tiling support for neutrino and gravitational-wave follow-up (Schüssler et al., 9 Jul 2025).

The user-facing model is built around rapid triage. Clients expose a sky map, sortable event lists, graphical event summaries, contextual links, and observability products. Cone searches use a default radius of $10^\circ$, list sources in ascending order of separation, and cross-match against a merged persistent-source catalog containing Fermi 4FGL (5788 sources), TeVCat (118 sources), and FLaapLUC targets (228 sources) (Reichherzer et al., 2021). Observability is computed for selected observatories or custom locations, with next-24-hour and monthly views, Sun and Moon altitude constraints, Moon phase, and Moon–source separation. By 2025, the platform reported visibility support for a database of over 2600 professional and amateur observatories compiled by the IAU, together with JSON visibility windows for schedulers and automatically collected photometry from ATLAS, ASAS-SN, and ZTF via Fink (Schüssler et al., 9 Jul 2025).

Measured performance in the 2021 architecture paper is oriented toward control-room and mobile use: push notification alert latency of 335 ms, cone search of a transient in 1291 ± 71 ms, manual cone search in 850 ± 292 ms, monthly visibility plots in 892 ± 59 ms, and detailed 24-hour visibility plots in 2768 ± 80 ms (Reichherzer et al., 2021). The later overview also cites practical use in the IceCube-211208A case, where immediate contextualization of the neutrino with the flaring blazar PKS 0735+178 helped trigger rapid multiwavelength follow-up (Schüssler et al., 9 Jul 2025). The platform is presented throughout as a complement to GCN, TNS, AMON, and broker systems rather than a replacement for them.

3. Observatory and instrumentation uses in astronomy

One established astronomical use of the name is the Colibri fast-photometry array for serendipitous stellar occultations by Kuiper Belt objects. The preliminary trials paper describes a dedicated array conceived to detect sub-10 km KBOs through millisecond-scale diffraction signatures, with occultation timescales set by the Fresnel scale,

$r_F = \sqrt{\frac{\lambda D}{2}},$

and a target cadence of 40 Hz to resolve $\sim 200$ ms events at optical wavelengths and $D \approx 40$ AU (Pass et al., 2017). The later technical description presents the built system at Elginfield Observatory as three 50 cm telescopes separated by 110–160 metres, streaming imagery to disk at 1.5 GB/s for next-day processing by a custom pipeline, and reaching limiting magnitudes of about 12.1 at 40 fps for temporal SNR = 5 (Mazur et al., 2022). The array uses multi-station coincidence to suppress scintillation false positives, and its detection pipeline combines wavelet filtering, diffraction-kernel matching, and strict GPS-timed coincidence.

A separate astronomical referent is the COLIBRI robotic follow-up telescope for SVOM. This facility is the French Mexican Ground Followup Telescope at the Observatorio Astronómico Nacional, Sierra de San Pedro Mártir, with a 1.3 m alt-az telescope, two simultaneous optical channels, and a near-infrared feed (Basa et al., 27 Apr 2026). Early operations reported an average end-to-end delay of about 60 s from onboard trigger to first exposure for observable SVOM alerts, follow-up efficiency of about 71% overall and about 77% excluding far-southern unobservable alerts, and detection efficiency of about 65–71% once ECLAIRs triggers are validated. The system is designed to refine $\sim 26$ arcmin space-based localizations to $<0.5$ arcsec within minutes, with real-time stacked images, calibrated photometry, and transient candidates delivered by an automated pipeline.

The CAGIRE imager is the infrared component of this telescope. Before installation, it was specified as a wide-field J/H camera with a 21.7 arcmin square field, 0.65 arcsec per pixel sampling, a 2048 × 2048 ALFA HgCdTe detector, and 1.33 s non-destructive full-frame reads (Flèche et al., 2023). An end-to-end detectability study later derived $ZP_J = 20.90 \pm 0.03$ mag and $ZP_H = 21.26 \pm 0.04$ mag under representative conditions, and found that, among nine $z \ge 6$ GRB afterglows, four would be recovered in an early-catch scenario, remaining detectable up to $z = 9.6$ in J and $r_F = \sqrt{\frac{\lambda D}{2}},$0 in H for the brightest cases (Fortin et al., 2024). The same 1.3 m COLIBRI telescope has also been used beyond GRB work: a 121-day $r_F = \sqrt{\frac{\lambda D}{2}},$1 campaign on SN 2025bvm measured $r_F = \sqrt{\frac{\lambda D}{2}},$2 mag and $r_F = \sqrt{\frac{\lambda D}{2}},$3 mag, consistent with a normal, slowly declining Type Ia supernova (Gonzalez-Buitrago et al., 12 Jan 2026).

4. COLIBRI in stellar evolution and stellar populations

In stellar astrophysics, COLIBRI is a TP-AGB evolutionary code introduced as a fast, envelope-based alternative to full 1D calculations while retaining much of their physics (Marigo et al., 2013). Its core strategy is to integrate the stellar structure equations from the atmosphere down to the bottom of the hydrogen-burning shell, with on-the-fly equation-of-state and Rosseland-opacity calculations synchronized to the evolving surface composition. The code computes effective temperature, atmosphere and convective-envelope structure, hot-bottom burning energetics and nucleosynthesis, the pulse-driven convective zone composition, and the onset and quenching of third dredge-up. It also solves a 25-isotope nuclear network and was described as the first evolutionary code to use accurate on-the-fly computation of the equation of state for roughly 800 atoms, ions, and molecules together with on-the-fly Rosseland mean opacities (Marigo et al., 2013).

A closely related paper coupled COLIBRI to stationary wind and dust-growth calculations for TP-AGB outflows (Nanni et al., 2013). There the major physical distinction was between a low-condensation-temperature silicate prescription that includes chemisputtering and an alternative high-condensation-temperature case without chemisputtering. The latter yields silicate condensation temperatures up to $r_F = \sqrt{\frac{\lambda D}{2}},$4 K rather than $r_F = \sqrt{\frac{\lambda D}{2}},$5 K, shifts the condensation radius from $r_F = \sqrt{\frac{\lambda D}{2}},$6 to $r_F = \sqrt{\frac{\lambda D}{2}},$7, and allows the models to reproduce the observed trend between terminal velocities and mass-loss rates of Galactic M-giants. For carbon stars, the same framework emphasizes the carbon excess $r_F = \sqrt{\frac{\lambda D}{2}},$8 as the governing quantity for homogeneous amorphous-carbon growth.

The code subsequently became a standard TP-AGB backend for population synthesis. The PARSEC-COLIBRI isochrones extend over 15 initial metallicities, $r_F = \sqrt{\frac{\lambda D}{2}},$9, and include complete thermal pulse cycles, surface H+He+CNO abundances, long-period variability models, and dust-growth plus radiative-transfer calculations (Marigo et al., 2017). In white-dwarf initial–final mass relation work, PARSEC evolves stars up to late TP-AGB conditions and COLIBRI completes the remaining pulses, mass loss, envelope removal, and core growth. That study uses the dredge-up efficiency

$\sim 200$0

and argues that a small but non-zero $\sim 200$1 together with a mass-dependent $\sim 200$2 is required to reproduce the observed IFMR kink at $\sim 200$3–$\sim 200$4 (Addari et al., 2024). COLIBRI’s role there is not long extrapolation but physically consistent completion: typical $\sim 200$5 is about 1–2 pulses, though a few cases reach about 8–10.

5. Computing, control, and information-science uses

In edge AI and neuromorphic engineering, Colibri denotes systems built around the Kraken low-power RISC-V SoC. ColibriES integrates a DVS interface, an eight-core PULP cluster, the SNE spiking accelerator, and the CUTIE ternary-CNN accelerator, and was evaluated on DVS Gesture with 164.5 ms end-to-end latency, 7.7 mJ per inference, and 83% accuracy (Rutishauser et al., 2023). ColibriUAV extends the same design to a drone platform with event-based and frame-based cameras; its SAER event path reaches 7200 event-frames per second at 10.656 mW, with end-to-end sensing and processing below 50 mW and a 163 ms closed-loop latency in the demonstrated configuration (Bian et al., 2023).

In computer architecture, Colibri is the scalable implementation of LRwait reservations within the LRSCwait proposal for manycore systems (Riedel et al., 2024). Its distributed linked-list reservation queue replaces retry-heavy LR/SC polling with sleeping waiters and FIFO service. On a 256-core open-source RISC-V platform, the reported area overhead is about 6%, while throughput improves by a factor of 6.5 over LR/SC-based implementations and energy efficiency by a factor of 7.1. The same name also appears in network security: Colibri is an open-source C implementation of Minimal IKE, including a post-quantum variant that replaces X25519 with ML-KEM-512 while preserving the lightweight INIT/AUTH structure (Zuane et al., 25 Feb 2026). In the post-quantum configuration, the reported average memory usage is $\sim 200$6 MB for Colibri versus $\sim 200$7 MB for StrongSwan.

Other uses are methodological rather than systems-oriented. In high-energy phenomenology, Colibri is an open-source Python tool for parton distribution function fits, built on JAX, supporting Hessian, Monte Carlo replica, and Bayesian nested-sampling methodologies, and writing output PDF sets in LHAPDF format (Costantini et al., 3 Oct 2025). In perception-oriented modeling, COLIBRI stands for “Color Linguistic-Based Representation and Interpretation,” a fuzzy color model based on HSI, human categorization experiments, and partitions over nine hue, four saturation, and five intensity categories, with a total reported sample of $\sim 200$8 (Shamoi et al., 15 Jul 2025). The reuse of the name in these cases is literal rather than conceptual: the projects are technically unrelated.

6. Disambiguation in scholarly usage

The sources make clear that “Colibri” is not a stable identifier across the literature. Even within astronomy, it can denote Astro-COLIBRI, the Colibri occultation array, the SVOM/FM-GFT COLIBRI telescope, or the COLIBRI TP-AGB code (Reichherzer et al., 2021, Mazur et al., 2022, Basa et al., 27 Apr 2026, Marigo et al., 2013). Outside astronomy, it additionally denotes hardware platforms, algorithms, software packages, and fuzzy knowledge representations. Capitalization only partly helps: both “Colibri” and “COLIBRI” occur across several unrelated projects.

A common misconception is therefore to treat Colibri as a single telescope or a single software framework. The record is more heterogeneous. In some cases the name is an acronym—COincidence LIBrary for Real-time Inquiry in multimessenger astronomy, and Catching Optical Light and Infrared BRIGHT transients for the SVOM ground telescope—whereas in other cases it functions simply as a project name (Schüssler et al., 2021, Flèche et al., 2023). This suggests that precise bibliographic qualification is indispensable: the unqualified term carries little technical meaning without its domain, paper, or collaboration context.