Hacker News new | ask | show | jobs
by astro123 2323 days ago
Yeah, I'll second what others have said that designing (good) telescopes is not a quick/easy thing. Also, any telescope that is easy to slap together is probably so many orders of magnitudes worse that existing ones that you don't get any usable data from it.

I'll throw a couple examples out for you:

DESI[1] has 5000 individually movable fiber optic cables. It's also sitting on a 4m mirror. It is a massive engineering project, but also would make anything with similar goals that a handful of people could hack together for a few million dollars obsolete. It can get 5000 spectra every 15 mins and a small, slit based telescope can maybe do a handful per night. Large scale projects like "mapping the night sky" are going to be dominated by massive projects. See also GAIA [6]

Space based makes things way harder. Let's take JWST[2] as an example. Someone else has mentioned the tolerances on the mirror. You also need to keep everything cold (7 kelvin!). You need to be able to control this, keep it pointed in the right direction with stunning accuracy, etc, etc. And all this needs to work in space, after being shaken around through a rocket launch. You also need a really compelling reason to go to space for a telescope. Those reasons include, observing things that you can't see from the ground (X-rays for example). You need really good seeing (no atmosphere). You need really low noise observations. I'd be surprised if those were what a small operation needed. Especially when it makes all the other things (control, servicing, etc) so much harder.

In fact, there are gaps where pretty simple ground based hardware can do good scientific work. Though, it is usually for pretty specific goals. [3] is a bunch of DSLRs that is one of the best instruments for finding really diffuse galaxies which are really interesting systems at the moment.

On the ML techniques, those are definitely being used in astronomy. One recent example [4] but go to ADS and look for things with ML in the title and you'll get a lot.

[1] https://www.youtube.com/watch?v=g1LVMox0KNc [2] https://jwst.nasa.gov/content/observatory/ote/mirrors/index.... [3] https://www.dragonflytelescope.org/ [4] https://www.kaggle.com/c/PLAsTiCC-2018 [5] https://ui.adsabs.harvard.edu/search/filter_database_fq_data... [6] https://en.wikipedia.org/wiki/Gaia_(spacecraft)