HATPI: Wide-Field Time-Domain Survey Facility

• HATPI is a continuous sky survey facility that employs a mosaic of 64 camera–lens units to capture wide-field, high-cadence observations.
• It covers approximately 0.7π steradians with a spatial scale of ~20 arcsec per pixel, enabling monitoring of exoplanet transits, stellar variability, and moving objects.
• Robust data processing using techniques like EPD and TFA ensures precise photometric calibration for recovering long-period exoplanet ephemerides and comet light curves.

HATPI (Hungarian-made Automated Telescope PI-steradians) is a recently commissioned time-domain facility at Las Campanas Observatory, Chile, designed as a continuous, high-cadence, very wide-field survey of the sky visible from that site. In the current literature it is described as a wide-field photometric instrument consisting of 64 camera–lens units in a mosaic configuration, using a broad optical bandpass and a coarse spatial scale to trade angular resolution for sky coverage and cadence. Its published roles already span long-period exoplanet ephemeris recovery, stellar-variability monitoring, and moving-object time-series photometry (Hartman et al., 25 Feb 2026, Rojas et al., 13 Feb 2026).

1. Instrument concept and physical configuration

HATPI is installed at Las Campanas Observatory and is operated as a continuous survey instrument rather than a pointed follow-up telescope. One description characterizes it as “a mosaic of 64 identical camera–lens units mounted together on a single platform,” designed for “continuous imaging of the entire sky visible from LCO,” while another describes “64 fast small telescopes in a mosaic configuration to monitor the full sky visible (above the horizon by at least approximately 35 degrees) from Las Campanas Observatory in Chile” (Rojas et al., 13 Feb 2026, Hartman et al., 25 Feb 2026).

Its hardware is specified in greater detail in the moving-object study. The facility uses 64 Mitakon 154 mm f/1.6 lenses, with an effective clear diameter of 96 mm, coupled to 64 Finger Lakes Instrumentation MicroLine ML230 cameras, each carrying an E2V CCD230-42 $2048\times 2048$ back-side-illuminated CCD. The 64 lens+camera systems are termed Instrument Holder Units, or IHUs, and are attached to a single massive equatorial-drive mount (Hartman et al., 25 Feb 2026).

The facility-level geometric parameters are central to its design. The published mosaic field of view is given as 7,100 square arcdegrees in the abstract of the 3I/ATLAS paper and as 7,055 square degrees in the instrument description, corresponding to 17.1% of the full celestial sphere, or $0.7\pi$ steradians. The spatial scale is $19.7$ arcsec pixel$^{-1}$, and one exoplanet analysis rounds this to “20 arcsec pix$^{-1}$” when discussing survey characteristics (Hartman et al., 25 Feb 2026, Rojas et al., 13 Feb 2026).

Property	Published description	Source
Site	Las Campanas Observatory, Chile	(Hartman et al., 25 Feb 2026)
Optical units	64 camera–lens systems / 64 IHUs	(Rojas et al., 13 Feb 2026, Hartman et al., 25 Feb 2026)
Detector format	$2048\times 2048$ back-side-illuminated CCDs	(Hartman et al., 25 Feb 2026)
Field of view	7,100 square arcdegrees; 7,055 square degrees	(Hartman et al., 25 Feb 2026)
Spatial scale	$19.7$ arcsec pixel$^{-1}$	(Hartman et al., 25 Feb 2026)

The published descriptions therefore define HATPI as an extremely wide-field, high-cadence survey instrument whose architecture is optimized for persistent sky coverage rather than high spatial resolution.

2. Survey operation, cadence, and photometric bandpass

HATPI observes through a custom broad-band filter covering $430\,\mathrm{nm}<\lambda<890\,\mathrm{nm}$, with central wavelength $\lambda=660\,\mathrm{nm}$. The 3I/ATLAS paper notes that this bandpass is “fairly close” to the Gaia $0.7\pi$0 bandpass, though “not equivalent,” and the TIC65910228b paper lists the same bandpass in Table 1 as “Asahi Spectra 430–890 nm” for the exoplanet observations (Hartman et al., 25 Feb 2026, Rojas et al., 13 Feb 2026).

The cadence is described in two closely related ways. The TIC65910228b paper states that HATPI enables continuous imaging with cadences of 30 and 45 seconds, while the 3I/ATLAS paper describes regular synchronized operation at a cadence of 45 s, with all 64 cameras observing simultaneously (Rojas et al., 13 Feb 2026, Hartman et al., 25 Feb 2026). This suggests that 45 s is the published operational cadence for the recent survey-mode moving-object demonstration, while the exoplanet paper preserves the broader description of the facility as supporting 30 and 45 s cadences.

The mount tracks the sky at the sidereal rate for one hour, then slews back to its initial hour angle position and tracks again. Each IHU includes motors for micro-tracking and focus control, so that pointing is stable at the sub-pixel level at the centers of the individual fields and to the pixel level across the mosaic over each one-hour tracking period (Hartman et al., 25 Feb 2026).

The single-exposure photometric reach is also documented. In uncrowded regions at new moon, HATPI reaches a $0.7\pi$1 detection threshold of $0.7\pi$2 mag in a single exposure, while sources with $0.7\pi$3 mag are saturated. This limiting magnitude is directly relevant to the 3I/ATLAS results, where the transition from nightly-stack detection to single-exposure detection occurs at $0.7\pi$4 mag $0.7\pi$5 mag systematic uncertainty (Hartman et al., 25 Feb 2026).

3. Data products, reduction, and calibration

The exoplanet and moving-object papers together give a relatively complete picture of the HATPI processing chain. For stellar light curves, “high-precision light curves were generated using aperture and image subtraction photometry,” and are made available in three forms: raw photometry, photometry detrended via External Parameter Decorrelation (EPD), and photometry with additional systematic correction using the Trend Filtering Algorithm (TFA) in the EPD light curve. The EPD step removes variability correlated with source position, horizontal angle, airmass, and PSF shape; TFA removes common-mode systematics using many stars’ light curves. In the TIC65910228b analysis, the adopted product was explicitly “the subtracted photometry TFA light curve” (Rojas et al., 13 Feb 2026).

The moving-object pipeline is described in more detail. Calibrated frames are produced with standard CCD steps—overscan correction, trimming, bias subtraction, dark subtraction, bad-pixel masking, and flat-fielding—implemented with FITSH. Astrometric solutions are derived against Gaia DR2, and for the 3I/ATLAS observations the median astrometric error is reported as $0.7\pi$6. Optimal image subtraction is then carried out against an empirical reference image built from $0.7\pi$7 observations of the same field, obtained near new moon under clear conditions. For moving objects, photometry is measured as forced aperture photometry at the ephemeris-predicted position on the subtracted images (Hartman et al., 25 Feb 2026).

Three fixed circular apertures are used for the comet analysis, with radii 1.45, 1.95, and 2.35 pixels, corresponding to 28.6, 38.4, and 46.3 arcsec. The relative photometry is tied to the Gaia $0.7\pi$8 system by fitting a polynomial transformation from instrumental reference magnitudes to Gaia $0.7\pi$9, with dependence on source $19.7$0 and $19.7$1 position and on $19.7$2 color. The average uncertainty on this transformation is $19.7$3 mag. The same calibration is written as a flux scaling relation,

$19.7$4

where $19.7$5 is the measured flux on the reference system and $19.7$6 is the magnitude on the Gaia $19.7$7-band system (Hartman et al., 25 Feb 2026).

The papers also show that HATPI reduction is strongly instrument-aware. In the exoplanet joint fit, the large pixel scale is handled with an explicit dilution prior,

$19.7$8

motivated by the expectation that dilution should not exceed 10% and conservatively allowing for modest contamination. A photometric offset and a jitter term are also fitted for HATPI,

$19.7$9

together with quadratic limb-darkening parameters in the Kipping reparametrization (Rojas et al., 13 Feb 2026).

For moving objects, the error model is similarly conservative. Formal aperture-photometry errors are supplemented by a floor based on robust scatter estimates, and large-scale quality flags reject frames with poor subtraction statistics, satellite-trail contamination, or strong blending with bright or variable neighbors. This treatment is necessary because the coarse pixel scale makes crowding and background variability important observational systematics (Hartman et al., 25 Feb 2026).

4. Role in long-period exoplanet discovery and ephemeris recovery

HATPI’s most clearly documented exoplanet role is in the discovery and characterization of TIC65910228b, a transiting warm Jupiter found initially as a single TESS transit. In that study, the central problem was that TESS provided only one event, so the orbital period was not directly known. The published observing strategy was sequential: radial velocities from FEROS and PLATOSpec were used to identify a $^{-1}$0-day signal; HATPI then supplied wide-field continuous photometry over multiple seasons; once HATPI caught a second transit egress, the refined ephemeris was used to schedule a higher-precision egress with Observatoire Moana; and a joint fit of TESS, HATPI, OM-ES1, FEROS, and PLATOSpec data was then performed with juliet (Rojas et al., 13 Feb 2026).

The HATPI light curves for TIC65910228 span three observing seasons:

• 2022/08/17–2023/06/02
• 2023/08/17–2024/06/02
• 2024/08/17–2025/06/02

Within that baseline, HATPI observed an egress of TIC65910228b on 2024/12/09. The paper states that this egress “is consistent with the period found in Section 3.2 (GLS on RVs).” In the joint narrative of the paper, this is the first ground-based confirmation of a second transit and the key event that converted a single-transit TESS detection plus an RV periodogram peak into a phase-coherent ephemeris (Rojas et al., 13 Feb 2026).

HATPI also contributed to activity vetting. The paper presents generalized Lomb–Scargle periodograms of the PLATOSpec radial velocities, bisector span, CCF FWHM, and the HATPI light curve, with the explicit statement that the HATPI GLS can uncover variability linked to stellar rotation. The result was that the HATPI photometry did not show strong periodicities at the planet’s orbital period, and this, together with the activity diagnostics, supported the planetary interpretation of the RV signal (Rojas et al., 13 Feb 2026).

In the final transit-plus-RV model, HATPI entered as an independent photometric data set alongside TESS and OM-ES1, using batman within juliet. The paper does not provide a side-by-side “with HATPI” versus “without HATPI” comparison, but it does state that the 2025 OM-ES1 egress was predicted using previous analysis with HATPI data and radial velocities, and that this combination constrained the period and refined the ephemeris sufficiently to schedule the observation. A plausible implication is that HATPI’s main exoplanet contribution is not merely photometric confirmation, but ephemeris recovery over multi-year baselines for sparse single-transit detections (Rojas et al., 13 Feb 2026).

5. Moving-object time-series photometry and the 3I/ATLAS demonstration

The first published moving-object case study is “HATPI Pre-Perihelion Time-series Photometry of the Interstellar Comet 3I/ATLAS” (Hartman et al., 25 Feb 2026). That paper explicitly presents the 3I/ATLAS analysis as “the first report of moving object time-series photometry from the HATPI facility,” and it uses the comet as an early demonstration of the survey’s capabilities for Solar System work.

The data volume is large. Using JPL Horizons ephemerides, the authors identified 15,317 HATPI images containing 3I/ATLAS between 2025 May 1 and 2025 Sep 13. After applying quality cuts for problematic images, satellite contamination, and blending with bright or variable stars, 7,294 clean observations remained. The comet moves very little within a single 45 s exposure: the maximum motion over the observing window is $^{-1}$1, less than $^{-1}$2 pixels, so no elongation correction was required (Hartman et al., 25 Feb 2026).

The first robust recovery was obtained on the night of 2025 Jul 2, one night after discovery, at a Gaia $^{-1}$3-band magnitude of

$^{-1}$4

with an additional $^{-1}$5 mag systematic uncertainty. The comet subsequently brightened to

$^{-1}$6

$^{-1}$7 mag by 2025 Sep 13, after which it became unobservable by HATPI as it approached perihelion. Before the comet reached

$^{-1}$8

$^{-1}$9 mag on 2025 Aug 6, it could be detected only in nightly stacks; after that date it was bright enough to be detected in individual 45 s exposures (Hartman et al., 25 Feb 2026).

The time-series behavior after Aug 6 is also informative. The paper states that it does not detect evidence for significant short-time-scale variations in the brightness of 3I/ATLAS after that date, and the plotted 45 s light curves show no clear intra-night variability at the $^{-1}$0 mag level. The interpretation given in the paper is that by Aug 6 the coma dominated the light from the system, suppressing any rotational signal from the nucleus (Hartman et al., 25 Feb 2026).

For the photometric evolution, the paper adopts the standard comet magnitude law

$^{-1}$1

and for small phase angles assumes

$^{-1}$2

Using a fixed linear aperture of $^{-1}$3 km and setting $^{-1}$4, the combined fit to HATPI and literature photometry yields a heliocentric index

$^{-1}$5

and a phase coefficient

$^{-1}$6

By contrast, the HATPI photometry alone is fit by a steeper heliocentric dependence,

$^{-1}$7

over $^{-1}$8 (Hartman et al., 25 Feb 2026).

The paper discusses several possible reasons for this steeper HATPI-only rise, including the larger linear aperture required by HATPI’s lower spatial resolution, the different bandpass relative to other surveys, and a possible transition toward stronger $^{-1}$9-driven activity. Because these explanations are framed as interpretation rather than direct measurement, the secure instrumental result is that HATPI can deliver calibrated, multi-month moving-object light curves dense enough to constrain heliocentric brightening laws and phase coefficients (Hartman et al., 25 Feb 2026).

6. Scientific niche, demonstrated strengths, and observational limitations

The published HATPI papers define a specific observational niche. HATPI is an all-sky, high-cadence, low-spatial-resolution survey system whose primary time-domain goals include long-period transiting giant planets, bright fast transients, and small near-Earth asteroids. In practice, the current literature shows two especially strong use cases: recovery of poorly predicted long-period exoplanet transits, and archival plus real-time light curves for moving objects (Hartman et al., 25 Feb 2026, Rojas et al., 13 Feb 2026).

Its main demonstrated strength is persistence of coverage. In the TIC65910228b study, HATPI functioned as a continuously observing “safety net” that captured an otherwise difficult second transit egress for a $2048\times 2048$0-day warm Jupiter. In the 3I/ATLAS study, the same wide coverage enabled rapid recovery one night after discovery and retrospective examination of earlier survey images. This suggests that the facility is particularly valuable where cadence and baseline matter more than angular resolution.

The main limitation is equally explicit in the published analyses: the $2048\times 2048$1 arcsec pixel$2048\times 2048$2 scale makes crowding, dilution, and blending unavoidable parts of the measurement problem. In the exoplanet fit this appears as an explicit dilution prior. In the comet analysis it required aggressive cleaning: 8,023 of 15,317 measurements were flagged as affected by problematic images, satellite trails, bright neighbors, or variable stars, leaving 7,294 clean observations. This is not an incidental detail but a defining property of HATPI data products (Rojas et al., 13 Feb 2026, Hartman et al., 25 Feb 2026).

The observational trade-off is therefore clear. HATPI is not a narrow-field precision imager; it is a very wide-field survey facility whose science return depends on continuous monitoring, robust detrending, and careful treatment of contamination. Within that regime, the published record already shows that HATPI can supply the second transit needed to phase-connect a single-transit TESS warm Jupiter, can participate directly in joint transit-plus-RV inference, can provide multi-year stellar-variability baselines, and can produce Gaia-calibrated moving-object time series at 45 s cadence. That combination of survey breadth, cadence, and archival depth defines its present astronomical significance (Rojas et al., 13 Feb 2026, Hartman et al., 25 Feb 2026).