Horizon-Scale Interferometric Imaging Data

Two exact Event Horizon Telescope products support this guide: the calibrated M87 visibility deposit 2019-D01-01 and the imaging-pipeline source release 2019-D01-02 at one pinned Git commit. A visibility file is not an image, an imaging pipeline is not a frozen executable environment, and a reconstructed emission ring is not by itself a measurement of a critical curve, shadow, or event horizon. Product identities and official records were checked on 2026-08-25.

Helpful background. Probability and Inference for Gravity introduces priors, likelihoods, posterior checks, and uncertainty propagation. Those ideas become essential after calibration and reconstruction choices enter the analysis.

Calibrated M87 visibilities

The accepted measurement product is First M87 EHT Results: Calibrated Data, product 2019-D01-01, DOI 10.25739/g85n-f134, released through CyVerse Data Commons on 2019-04-10. It is a total-intensity visibility product derived from the April 2017 campaign, not raw telescope voltage or unaveraged correlator output.

Field	Exact public-product value
Source and wavelength	M87 at 1.3 mm
Observing dates	2017 April 5, 6, 10, and 11; day-of-year 095, 096, 100, and 101
Role of M87	Target on April 5, 6, and 11; seven calibration-source scans on April 10
Frequency bands	Low band centered at 227.1 GHz; high band centered at 229.1 GHz
Released stations	AA ALMA; AP APEX; AZ Submillimeter Telescope; JC James Clerk Maxwell Telescope; LM Large Millimeter Telescope; PV IRAM 30 m; SM Submillimeter Array
Calibration state	EHT-HOPS fringe fitted, a-priori calibrated, independently flux calibrated, and network calibrated
Averaging	10-second time averaging; all 32 IFs averaged across each 2 GHz band
Polarization product	Derived Stokes I; cross-hands RL and LR and their errors are zero; RR and LL both contain estimated Stokes I, with errors sqrt(2) times its statistical error; JCMT single-polarization baselines retain their parallel-hand treatment
Formats	UVFITS measurement files; CSV and text convenience exports

EHT Paper III, §§2–8 documents the observation path, correlation, three independent reduction pipelines, fringe fitting, flux and network calibration, averaging, and cross-pipeline validation. The public deposit contains the network-calibrated EHT-HOPS product rather than all three independently reduced trees. Paper III’s reported 2% amplitude and 1 degree phase baseline systematic-error scales belong to its stated validation scope; they are not a universal error model for every visibility or later image.

Finite inventory and retrieval fingerprints

The official INVENTORY.txt lists 34 files: eight day/band UVFITS files and a conversion script; eight text exports and a dump script; eight CSV exports and a dump script; four root files; and three convenience archives. The CyVerse record shows that some convenience material changed after 2019, including a text archive and README in 2025. The DOI identifies the deposit but does not, by itself, freeze every retrieved byte.

Official convenience archive	Bytes	Retrieval-local SHA-256 on 2026-08-25
EHTC_FirstM87Results_Apr2019_uvfits.tgz	1,606,171	f36f77fe4c0cf772625dcf424ed34ba3d1b5766475c64ed90319b0000f1ed463
EHTC_FirstM87Results_Apr2019_txt.tgz	1,524,282	2e3edeb06826bfe3686930d6c677d68985ed2f655bd8d303cb9ae4638ed5c0ab
EHTC_FirstM87Results_Apr2019_csv.tgz	1,453,358	8fa8964ad5b1c80dd78bc3e023f849bb33312b2d8af95d6718439f529f361911

These are locally recomputed retrieval hashes, not publisher-signed checksums. Use the individual UVFITS files as the visibility authority, preserve their names, and treat the CSV and text branches as derived views.

Data rights and citation

The CyVerse LICENSE.txt assigns the data files to the Open Data Commons Public Domain Dedication and License 1.0 (ODC-PDDL). It assigns GPLv3 to the Python and shell scripts inside that deposit. Cite the Event Horizon Telescope Collaboration (2019), First M87 EHT Results: Calibrated Data, CyVerse Data Commons, DOI 10.25739/g85n-f134, and cite the analysis paper used for any scientific conclusion.

Pinned imaging-pipeline revision

The accepted source product is First M87 EHT Results: Imaging Pipelines, 2019-D01-02, at commit 80d230e76c08edd31548e7cfe61dc7cf94300780. The commit has tree e1837e2d75932a5efbb4888702bf82fdc1e8c04b, date 2020-02-20 19:21:30 UTC, and message “Improvement: use ‘Agg’ backend so the script works without X11.” It was the official repository’s master head when checked on 2026-08-25, but the commit is unsigned; the full SHA, not the branch name, is the durable revision reference.

The tree contains 14 tracked blobs under the DIFMAP, eht-imaging, and SMILI branches plus root launch material. The repository does not contain the visibility data. Its pinned source archive was 17,202 bytes with retrieval-local SHA-256 00ed9b9272c0973c684ab94d70006215cbb2ec14684adb8c0e145127a3539f32. The commit and tree identify the source; the archive hash only fingerprints that retrieval.

Pipeline branch	Declared dependency boundary
DIFMAP	Caltech DIFMAP 2.5b or newer
eht-imaging	eht-imaging 1.1.0; no complete environment lock
SMILI	SMILI 0.0.0; eht-imaging 1.1.0 for pre-imaging calibration; Python 2.7; tested on Ubuntu 18.04 and macOS 10.13/10.14

This is parameter-survey source, not a container, complete dependency lock, image archive, posterior, or end-to-end metric-inference workflow.

Rights boundary — checked 2026-08-25. The official 2019-D01-02 tree contains no LICENSE file. GitHub’s repository-license endpoint returned 404. No repository-level reuse license is asserted. GPLv3 from the separate 2019-D01-01 CyVerse deposit applies only to scripts inside that deposit and must not be transferred to this repository. Public readability permits citation; it does not by itself grant redistribution or modification rights.

From visibility to physical interpretation

Each layer adds assumptions. Keeping the layers separate prevents an image feature from silently becoming a statement about spacetime geometry.

Layer	What is fixed here	What still has to be supplied
Calibrated visibilities	Exact 2019-D01-01 UVFITS identity, array, dates, bands, averaging, and Stokes-I reduction	File selection, uncertainty treatment, flagging, and any additional calibration
Imaging pipeline	Exact 2019-D01-02 source commit and declared parameter survey	Executable environment, dependency resolution, command, seed where relevant, and reproduction checks
Reconstructed image	One regularized or model-based solution compatible with declared data and settings	Robustness across algorithms, hyperparameters, synthetic data, and held-out visibility-domain checks
Emission-model inference	A map from brightness structure to plasma and radiative-transfer parameters	Model family, priors, variability, scattering, calibration nuisance terms, and model comparison
Gravitational angular scale	A relation between image-domain scale and a mass-to-distance angular scale	Calibrated mass and distance information plus an emission and spacetime model
Critical curve or shadow	A spacetime-geometric prediction linked to photon propagation and source illumination	Metric, observer, plasma opacity, source distribution, and a validated mapping from observed emission to the geometric feature
Horizon or Kerr interpretation	A causal or metric-level claim	Competing spacetimes, plasma systematics, resolution criteria, and a dated inference analysis

An emission ring can be robust across a declared set of reconstructions while its relationship to a critical curve remains model dependent. A shadow is an observer- and illumination-dependent dark region, whereas an event horizon is a global causal boundary. Neither is literally a bright ring in a reconstructed image.

Reproduction limits

The public visibility deposit contains derived Stokes I, while EHT Paper IV used full SR1 parallel-hand RR and LL inputs. The deposit README reports no net change to the DIFMAP and eht-imaging conclusions and a small SMILI self-calibration change, but that collaboration assessment does not make the inputs byte-equivalent. JCMT handling also differs between the public Stokes-I product and the Paper-IV input.

Single-precision UVFITS conversion and differences in NumPy, Astropy, SciPy, eht-imaging, SMILI, DIFMAP, and operating systems can change a reconstructed image. Record the environment and compare outputs rather than assuming a modern install reproduces 2019 behavior.

EHT Paper IV, §§4–9 and Appendix C describes independent imaging methods, synthetic-data validation, and feature-stability tests. Its parameter survey is explicitly coarse, not a posterior distribution, and not exhaustive over imaging choices. Reproducing a fiducial image therefore does not prove a unique morphology, emission mechanism, gravitational scale, critical curve, shadow boundary, event horizon, or Kerr metric.

From visibility data to physical conclusions

Use Very-Long-Baseline Interferometry for visibility formation, closure observables, calibration, and array limitations. Use Image Reconstruction and Model Fitting for regularization, self-calibration, synthetic-data tests, and visibility-domain forward modeling. Consult the dated Horizon-Scale Imaging Tests of Kerr for the physical interpretation of those observations and the limits of the conclusions at its stated cutoff date.

References

• Event Horizon Telescope Collaboration. Official Data Products Registry. Event Horizon Telescope public data registry (2019), registry, checked 2026-08-25.
• Event Horizon Telescope Collaboration. First M87 EHT Results: Calibrated Data. CyVerse Data Commons (2019), DOI, including the official README, inventory, and license.
• Event Horizon Telescope Collaboration. First M87 EHT Results: Imaging Pipelines. Event Horizon Telescope software archive (2019), product 2019-D01-02, official repository at commit 80d230e76c08edd31548e7cfe61dc7cf94300780.
• Event Horizon Telescope Collaboration. “First M87 Event Horizon Telescope Results. III. Data Processing and Calibration,” The Astrophysical Journal Letters 875 (2019) L3, DOI.
• Event Horizon Telescope Collaboration. “First M87 Event Horizon Telescope Results. IV. Imaging the Central Supermassive Black Hole,” The Astrophysical Journal Letters 875 (2019) L4, DOI.