-
Notifications
You must be signed in to change notification settings - Fork 2
Development Plan
This document tracks the major development challenges foreseen for astrohack.
| Milestone | Challenge | Status | Collaborators |
|---|---|---|---|
| VLA support | Holography | Final stages of Validation | PB |
| Antenna position corrections | waiting validation | GM, PB | |
| Pointing corrections | Not started | ||
| ALMA Support | Near Field Holography | Started | JM, JAGV |
| Far Field Holography | Started | JM, PJ | |
| ngVLA Support | Optical Design | Completed Summer 2025 | ngVLA team |
| (Ant. Proto. Mar 2026) | Panel layout | Completed Summer 2025 | JM |
| Aperture Models | Completed Summer 2025 | SkS | |
| Other Telescopes | Different Antenna designs | Not Started | |
| Single Dish | Not Started | ||
| Demo Release | Initial code sprint | Completed February 2023 | URV |
| API Development | API Design | Completed April 2023 | URV |
| CI and Tests | Setting up test pipeline | Completed October 2023 | SC, NS, AW |
| Benchmarking | Uncertain | AM |
Holographies with the VLA are well supported and in the final stages of validation. Tests against AIPS have resulted in corrections being very similar, with 95% agreement in screw corrections between the two softwares. Another validation test that has been carried out is comparing antenna surfaces between On-the-fly and step holographies. This AIPS independent test shows a recoverability of screw adjustments that is > 95%.
The antenna position correction determination piece of astrohack, locit, has been tested against AIPS attaining very similar results and its validation is slated to start in the last quarter of 2024.
The need for the inclusion of antenna pointing corrections in astrohack has been identified in the first quarter of FY2025. Currently there is no implementation of this in astrohack. Effort on this front is slated to begin in 2026, Probably around the time of the initial round of tests with the ngVLA antenna prototype.
Due to the high frequency of ALMA receivers (> 70 GHz) it is hard to find strong enough sources in the sky for high precision astronomical holography. Hence panel adjustments are limited to using Near field holography using a transmitter on site. Astronomical holography is still performed at ALMA but its usage is limited to finding dish deformations with elevation and similar lower resolution measurements.
There are two main challenges foreseen in tackling ALMA near Field holography, taking into account the near field effects caused by the proximity of the transmitter and the fact that the asdm file produced by ALMA near field holography is non standard. ALMA near field holography is really important as it will enable us to validate panel adjustments with 5 points.
The Near Field ASDM does not contain a main table making a simple call to importasdm in CASA impossible. Work on how to read this non-standard ASDM is ongoing, We have contacted ALMA operations and they have also displayed interest in being able to read this non-standard ASDMs.
The proximity of the transmitter implies in non-plane parallel wavefronts at the antennas which modifies the antenna beam pattern. These effects are well understood and it is possible to correct for them. The modeling and correction of these effects will be done in astrohack.holog.
This point might actually not be relevant as the degrees of freedom of the panel fitting algorithms are actually independent of the number of adjusters on a panel.
ALMA far field holography data contains only 3 antennas which breaks the regular calibration schemes. Efforts on how to properly calibrate this kind of data is ongoing.
ngVLA Support has been added to astrohack in the summer of 2025. When holographic tests start several new challenges that can’t be completely foreseen might appear, but the original chalenges due to the new optical design and the panel layout have been solved.
ngVLA holography tests will probably start in spring 2026 for band X and in fall 2026 for the band Q.
The new optical design introduces several new challenges, the most impactful for astrohack is the fact that the new design uses shaped optics. The usage of shaped optics requires that we compute light paths numerically, or do a generic aproach by using Zernike Polynomials.
The path chosen by astrohack was to use Zernike Polynomials as the fitting results for the aperture can be used to create a phase model that can then be subtracted from the aperture phase to remove larger scale phase fluctuations, making it suitable for panel adjustments.
Contrary to the VLA, the ngVLA design includes a segmented secondary mirror with adjustable panels. It still remains unclear if we will be able to distinguish the surface deviations from the secondary from the those of the primary mirror as they will probably appear superimposed over each other on the aperture image.
All the currently supported telescopes have panels that are arranged in rings around the center of the primary mirror. This symmetry has allowed us to implement optimizations on the assignment of pixels to panels that will no longer be valid for the NGVLA. The NGVLA panels will be approximately rectangular and arranged on a grid, requiring a new strategy on how to assign pixels to panels.
This new strategy has been implemented with Shapely using polygons to represent the shape of the panels. Assign a pixel to a panel is now done by testing if the pixel is inside the panel polygons.
The creation of aperture models for the telescopes in intended to help in advanced imaging techniques. For this purpouse and also for a new way to correcting aperture phases, astrohack has implemented Zernike Polynomial fitting of apertures. Shortly after the creation of the apertures Zernike Polynomial models are created for each of the available correlations.
Antennas with complicated optical designs and panel arrangements might difficult to support. However a few, such as Meerkat, and GBT have optical designs somewhat similar to the ngVLA and this might simplify the process of incorporating these observatories.
We would like to be able to support single dish antennas to be able to perform holographies of the far flung ngVLA antennas, but also other observatories, such as the VLBA in the shorter term.
Astrohack’s first demo release was published on April 2023 and presented at DSOC.
The initial code sprint consisted in a concentrated effort by JH, JWS and VSM where they sat down in CV for two weeks in February 2023 to get a deliverable mostly ready for the demo release in April.
The design of the API was mostly based on the AIPS implementation, in which the holography is divided in 3 tasks: UVHOL, HOLOG and PANEL.
- UVHOL: This task extracts the holography information from the raw
visibilities and stores it for later use, in ASCII file, in
astrohack this is accomplished by 2 functions:
- extract_pointing: Extract the pointing information from the MS
- extract_holog: Extract the visibilities from the MS and then match them to the pointing information.
- HOLOG: This task takes the ASCII file with the holography information grids the visibilities and then Fourier transforms them to get the apertures, In astrohack this is accomplished by 1 function: holog.
- PANEL: This task splits the aperture phase into the panels that compose the antenna surface and then fits each panel individually to suggest adjustments to the screws in each panel to improve the antenna surface, in astrohack this is accomplished by the pane function.
Each of the astrohack functions returns an object that is a mirror of the data on disk. These objects can then be called to produced desired derived products such as plots and screw adjustments.
The test pipeline was put in place in the second half of 2023 and is based on the github Continuous integration API. The tests are run on Linux and Mac with 4 python versions: 3.8, 3.9, 3.10 and 3.11. Tests are automatically triggered at the creation of a pull request or a push to the main branch.
AM created a basic framework of benchmarking tests, however these have not (as of yet) been fully implemented.