wthomp

This is maybe not the place, but we did some apples to apples comparisons between PyMC, Dynesty, and the Julia Turing.jl package.

A little to my surprise, despite being a Julia fan, Turing really outperformed both the Python solutions. I think JAX should be competitive in raw speed, so it might come down to the maturity of the samplers we used.

I visited the Large binocular telescope just a month or two ago. A very impressive facility, and one can only imagine the image quality if they were captured using both mirrors coherently.

That doesn’t seem to match what I understood from the article. At one point they say explicitly that it will enable brighter displays.

My understanding is that the AO system for GMT is going to pose quite a challenge. A big topic of ongoing research is dealing with “petaling” where the separate primary mirrors in phase due to the atmosphere.

For combining the projects, it does seem like that may be the only option funding wise, but it’s hard to imagine what the resulting observatory would look like. Maybe it would have to be a completely new design?

There’s a bit more to it than that. GMT is further along but its design has many compromises versus TMT. Also, a second ELT in the southern hemisphere is less useful than one in both Hemispheres. The situation for TMT on Maunakea is definitely tricky, but it’s also a better site than either of GMT’s or EELTs.

Worth mentioning that essentially all professional astronomers / astrophysicists consider this person a grifter. It’s amazing that he keeps getting so much press though, would love to learn his secret.

Yes exactly. Which means it's extremely, extremely low.

Even though the probably of a meteorite hitting somewhere is high, the probability of hitting somewhere in particular is tiny.

I’m using this from Julia and both the user and developer experience is great. It’s much more limited than publication style plotting libraries but the instant 60fps reactivity is amazing.

#3 spot at atmospheric pressure

Coming from matplotlib, I found Makie such a breath of fresh air. The API is just as (if not more) flexible but way more predictable. Their layout system in particular is amazing. I think it bundles it's own constraint engine? Congrats on the new website!

PS. Thanks to the Makie team for the shoutout to my corner plot package in the ecosystem section!

As has been pointed out elsewhere, this is the first image of our galaxy in something other than light (radio, infrared, x rays, gamma rays are all photons).

Would you mind explaining this further? Since sound is literally pressure waves, I don't understand how reducing sound can still leave "sound pressure".

This is not a “nature publication” in the common sense. It’s in their commentary section.

This is true but the poster above doesn’t mean Monte Carlo integration. Rather, testing those methods on simulations from generative models.

Small correction, no one will be able to pay for time on JWST. But if you put in a proposal for time and it's accepted, they will pay you. That's to make sure there is sufficient funds available to properly make use of the data you proposed for.

If you were to fly into these nebula in some kind of spaceship they wouldn't be any brighter than they appear in the night sky from Earth. They would just look way way bigger. The frustrating thing is that our eyes start to respond differently to colours when the light is really really faint. So we would probably perceive them as a grayish green haze. If the image was brightened artificiallythen we would see it as mostly red, with some browns and blues.

This is really impressive, and a great write up. I’ve been following the work on static compiling Julia to x86 libraries from the GPUCompiler.jl folks but I didn’t expect to see Julia working on an Arduino any time soon. With some kind of basic GC support (even if just using a bump allocator) it seems like a good fraction of the language could actually be available. Most tight loops hopefully don’t allocate so it would mostly just be necessary for creating initial arrays and mutable structs.

As others have mentioned, this is similar to how the event horizon telescope works today!

However, there’s no free lunch. By using arrays of telescopes instead of a single filled dish/mirror, they are missing a lot of information. Imagine a telescope the size of the earth, but you only use light from a few dozen spots on the surface and let the rest fall through. This is why they had to do all that complicated image reconstruction processing to create the image shown in the papers.

No, unfortunately those measurements have to be taken at the same time. That said, as the Earth rotates the distance between any two pairs of antennas changes which can be used to add additional information to the images (those new pairs of measurements again have to be taken simultaneously). From what I understand, this is less useful for looking at Sgr A* since the scene isn’t static and changes on a roughly 10 minute timescale.

For those wondering if we could get sharper images with JWST, here’s the previously imaged black hole (same angular size as our own) compared to a single pixel from Hubble’s wide field camera 3:

https://twitter.com/alex_parker/status/1116070667068170240?s...

JWST will have smaller “pixels” but is in the same ballpark.

Unfortunately JWST won’t come close to being able to resolve Sgr A*. They measured its diameter in micro-arcseconds, whereas JWST’s limiting resolution will be measured in 10s-100s of milli-arcseconds.

For mercury, the L1 and L2 points would be closer to the planet.

Many circumstellar disks have been detected, but this is the first unambiguous detection of a circumplanetary disk. This is believed to be material swirling around a protoplanet, accreting on to it, and perhaps, forming moons. The actual processes of how planets form out of protoplanetary disks are still not well understood, so observations like this will help narrow things down. Quite impressive that they were able to resolve it with ALMA.

The diffraction limit is the fundamental resolution limit of a telescope. This is the size of "spot" that will be created on the camera sensor for a single point of light like a star [1].

Its easy to calculate, just take the wavelength of the of light you want to observe and divide it by the diameter of the primary mirror (and multiply by ~1.2).

For example, for JWST observing in the mid-infrared, say 4micron, with a 6.5 meter diameter mirror, has a resolution limit of: 4e-6 / 6.5 = 6.15e-7 Or about 0.6 micro-radians (astronomers would normally use arcseconds but leaving in radians for clarity).

This is just the theoretical limit though, it's reduced by any imperfections in the optics, and for telescopes on the ground, it's limited by the blurring of the Earth's atmosphere to about 4 micro-radians.

For narrow fields of view, however, ground-based telescopes can use adaptive optics to compensate for this shimmering/blurring in real time and reach close to their theoretical diffraction limit. Plus, they can be much bigger since we don't have to launch them into space. I'm not familiar with the MGT but this might be how it will surpass JWST in terms of resolution (which again also depends on the wavelength).

For infrared observations though, a huge effect that can't be compensated for is sensitivity. At mid-infrared wavelengths, the Earth's atmosphere actually glows and makes it much harder to see faint sources. This is one of the ways JWST will really shine.

[1] Note however that you can still do things like measure the position of an object to less than the diffraction limit using e.g. centroiding. But you can't tell if there are two objects or one below this limit.

This related blog post also talks about the history of the symbol and programming:

https://cormullion.github.io/pages/2020-10-09-asterisk/