Astraeus: Radio Interferometry Simulation

Radio interferometry arrays such as the Very Large Array (VLA), Atacama Large Millimeter/submillimeter Array (ALMA), and Event Horizon Telescope (EHT) are made up of a number of antennas that are distributed over large distances, but used collectively as a single, synthetic aperture. The arrays use principles in interferometry to achieve angular resolution much higher than each individual antenna would be able to achieve separately. By correlating signals from every pair of antennas, they construct high-resolution images of the radio sky, which reveal structure on scales from galactic jets to black hole shadows. This project provides an interactive visualization tool aimed at helping both scientists and educators explore how different array configurations affect imaging quality.

Visualization in radio astronomy is not just crucial to interpreting outcomes, but also central to decision-making for array planning and instruction—linking complicated Fourier-based theory of imaging and spatial intuition. This software employs exploration and interactivity to enhance comprehension of interferometry's ability and limitation, notably in imaging from partial data with algorithms such as CLEAN.

In 2019, the Event Horizon Telescope collaboration released the very first image of a black hole. This image achieved a resolution of approximately 25 microarcseconds, enabling the ability to see the event horizon shadow of a blackhole for the very first time.

Several projects aim to improve upon the resolution of the EHT. The ngEHT is an upgrade from the EHT that will include more dishes and larger baselines, ultimately allowing for higher resolution. Another example is BHEX, a space-based interferometer that aims to reveal the photon ring, which is the ring of light produced by light orbiting the blackhole before escaping into space.

This raises the question: how can scientists and decision makers make informated, cost-effective decisions about where to place new telescopes to get the best science out of them? Should we place them in orbit or add 3 more telescopes on Earth? Which scientific questions would each help answer? This is where this project comes in. The motivation of this project is to simulate how reconstructed images would look like for a given configuration of telescopes, which could be placed in real time to see their impact on image resolution.

Interferometry superimposes multiple incoming wave signals to recover spatial information about a distant source. Let us consider a well known experiment as an analogy. In the double slit experiment, light through two slits produces an interference pattern due to constructive and destructive interference between waves. Fringe spacing depends on slit separation and wavelength.


For radio interferometry, the slits are replaced with radio antennas. Two antennas a distance apart (a baseline) record the incoming wave at nearly the same time, but not exactly. The slight delay is interpreted as a phase difference or a slanted incoming wave. This time delay, τ (tau), is used to calculate the source’s position on the sky. One of the main differences from the double-slit experiment is that radio interferometry does the opposite: instead of deconstructing patterns to infer structure, we observe several interference patterns (visibilities) and use those to construct an image of the sky.


Notable interferometer arrays include the Very Large Array (VLA) in New Mexico with 27 antennas, the Atacama Large Millimeter/submillimeter Array (ALMA) in Chile with 66 antennas, and the Event Horizon Telescope (EHT)—a global network that captured the first image of a black hole. Interestingly, ALMA itself is an interferometer that also serves as a single element within the larger EHT array, demonstrating the hierarchical nature of modern radio astronomy.

A radio interferometric array is a network of antennas arranged over some geometry. Different arrangements have different capabilities. Famous examples include:

• VLA – Y-shaped array in New Mexico
• ALMA – High-altitude array in Chile
• EHT – Global array spanning continents

Some instruments, such as ALMA, are interferometers on their own, but also function as sub-arrays within larger systems like the EHT.

Converting interferometer data into a recognizable image is a big challenge. When antennas sample the visibility function (which is the Fourier transform of the sky's brightness distribution), they do so incompletely—some spatial frequencies are not sampled due to the finite number of antennas and their physical arrangement. This makes this an ill-posed inverse problem. This incomplete sampling results in artifacts in the reconstructed image, often called the "dirty image."

Mathematically speaking, the dirty image is the true sky image convolved with a dirty beam. This dirty beam represents the artifacts and sidelobes due to the unsampled frequencies.

Astronomers use sophisticated reconstruction algorithms to address this problem, with CLEAN being one of the most popular. Developed by Jan Högbom in 1974, CLEAN works on the principle that most astronomical images can be represented as a collection of point sources.

The algorithm iterates:

1. Find the brightest point in the dirty image
2. Subtract a fraction of that brightness (controlled by the "loop gain" parameter)
3. Record the position and brightness of that point
4. Repeat until a specified threshold is reached

The final "CLEAN image" is constructed by convolving the recorded points with an idealized beam pattern and adding back residual noise. Our tool allows you to adjust CLEAN parameters and see the effect immediately.

Beyond CLEAN, there are other algorithms as well, such as regularized maximum likelihood estimation and Bayesian inference.

The next technological evolution in radio interferometry is to take arrays out into space. Concepts like the projected Black Hole Explorer would locate radio telescopes in low Earth orbit, augmented by ground stations to provide unprecedented baselines and resolution.

Space-based interferometry possesses certain major advantages, including baselines that are larger than Earth's diameter and observation of frequency bands masked by Earth's atmosphere. Our program specifically includes the capability of simulating satellite-based telescopes such that one can analyze the influence of orbital parameters on image quality.

Ultimately, as arrays grow larger, interferometry will allow for more finer details to be unlocked.

Our tool allows users to simulate the image produced by an interferometry array. They can simulate how a model image will look given the placement of telescopes.

There are numerous inputs to the tool. The model image can be changed by uploading any image (such as a blackhole simulation image or a collection of various point sources). The image can be in the form of a FITS file or a png file. Array parameters (such as wavelength and duration) can be changed. Telescope locations can be moved, deleted or added (including the addition of satellites). Satellite parameters can be changed (such as eccentricity and arc perigee). Source direction can be changed as well. Additionally, as the image will be put through the CLEAN algorithm, the parameters of the CLEAN algorithm can be updated as well (such as the loop gain or the threshold).

In terms of outputs, the beam pattern, the fourier transform of the model image, the UV coverage and the dirty image generated by the pattern can be viewed. The dirty image will be put through the CLEAN algorithm and the ehtim algorithm. The outputs can be viewed as well.