JBorrow

You could ship that known environment as an optional download.

This is great! The ending phase really gives you the same feeling of “screw everything else, I need to finish this paper” rushing to get the final stuff out.

That is incorrect for scientific data. The limitations are:

a) Massive data volumes (~100 Gb - 1 Pb/project)

  ai) This means that data is typically stored on limited access machines like HPC clusters

  bi) This also means that shipping this data around is financially expensive, and cannot be supported purely by small client machines

b) A low number of seeders; scientific data is not exactly popular, and there may be network restrictions on uploads through the typically used networks;

c) The requirement for a data legacy; torrents are fantastic for ephemeral data (e.g. operating system builds), but are terrible for data that must be archived and kept for potentially decades to centuries.

What are you talking about? This is an ongoing global public health crisis: https://www.who.int/news/item/23-10-2022-almost-1-million-pe...

It says that in TFA :)

No it’s not, this even goes to things like reference letters. US letters for the same candidate are always much stronger and use very different language.

Cubehelix is not a matplotlib colour map, though it is available in their defaults :)

https://www.mrao.cam.ac.uk/~dag/CUBEHELIX/

Because MAME is effectively an emulator for arcade games and dolphin is a state-of-the-art console emulator? It is generally not be possible to get the level of fidelity and performance of a custom emulator out of a general-purpose framework like MAME.

"Cleanliness and modern plumbing" include massive water treatment facilities, digging up every street, laying billions of miles of pipe.

For part a) that is only true for a small selection of observations. A huge proportion of astronomical science comes out of surveys that typically use large fields of view. Additionally, observation times can be very long, so even if you are looking at a small bit of sky you can still end up having things ruined by a satellite.

For part b) you are completely incorrect in your assumptions here. The majority of the cost and effort of telescopes is really in the instrumentation (the 'cameras' that are strapped to the back of the telescope). To get the most out of an observing facility, we need access to these to change them, fix them, and improve them. This is not possible with a space-based mission.

What you want is a Withings ScanWatch

That’s just not true, most of Eastern Europe sees extreme temperature variation similar to New England.

Implementing O(N ln N) or O(N) gravity on GPUs (which I guess is what is underpinning this) is both much more difficult and it's unclear whether it would actually run faster for any reasonable number of particles. Don't forget about that constant :)

I worked a while ago on something similar, using a state-of-the-art universe simulation model (including hydrodynamics too). The simulations don't run in real time, but people may be interested as we provide a bunch of information as you go along:

http://www.galaxymakers.org/galform.php

In TFA they state that the weight loss was an incidental part of the experiment and in fact they thought they would gain weight.

$n00k of compute time is nothing, sorry. This is the kind of thing that academic institutions can give out for free…

Ah, we use the ML methods to understand what happens to a large collection of galaxies (maybe 100k - 1 million), all co-evolving together. Trying to use these methods v.s. direct simulation on individual galaxies is a no-go, too many nonlinearities exist.

Your comment about a numerical gravitational simulation being 'wrong' because of chaos may be true on an individual resolution element level, yes, but our simulations are made up of many billions of individual particles. Like traditional CFD simulations we can also demonstrate some level of convergence.

Similar things are happening in galaxy formation too; numerical models here cost tens of thousands - tens of millions of cpu hours to run, and have maybe 8-10 free parameters. We use emulators based upon a training set of maybe 100-200 real simulations to calibrate out those free parameters to scaling relations (e.g. sizes of galaxies v.s. their mass).

Odd that they did not mention these when discussing sub-grid modelling, as afaik these Astronomy simulations (along with planetary impacts) are probably some of the first to have used them…

depends, if you are bisecting the commit messages are very helpful

This depends on the region/university/timescale/etc.

Consider this: student at institution A (UK) writes code, postdoc at institution B (USA) modifies it, shared with researcher at institution C (China) who runs it for a paper first-authored by student at institution D (Germany).

Who has the rights to share the code and data? Which jurisdiction would this fall under?

From my perspective (postdoc in astrophysics) things are quite complicated.

I agree generally with data and code sharing policies (all of my recent papers include zenodo deposits of data, and I work on a large open source simulation code). However, there are some serious issues here:

1. Code licensing is not always clear. Sometimes you are working with a code that some PhD student wrote, who left the field, whose advisor shared it with you. Can you share this code? Not necessarily... but this problem will go away with time.

2. Some codes include proprietary components, and chopping them out can make the code inoperable (or useless). One notable instance of this is a very famous hydro code that was used to produce predictions for nukes, that was then repurposed for running simulations of planetary collisions.

3. Sharing data is prohibitive, and if you don't have the data having the code might not really help. Our projects can produce Pbs of data, costing 10s-100s Ms of CPU hours, and as such making the data open source can be really tricky. This then means you can only share a selection of runs. On the other hand, even if you had the code, you can't reproduce a lot of the results, because of the cost of running the simulations... This actually will get _worse_ over time as we run bigger and bigger simulations.

Yes, basically a bunch of huge telescopes across the planet have their images combined to give an effective mirror size of the entire earth, so one satellite is not going to cut it (even including the fact that IR has a shorter wavelength).

Can anyone say strawman?

Having labs is not equal to having bioweapons

The next tweet responds to your point 3.

https://twitter.com/nick_beauchamp/status/147355576211076710...

For your point 1, this is a smoothed average. I suppose they should have used wider bins, but "failing the eyeball test" doesn't say much. The point isn't to fit the data, but to collate it so that you can clearly see the overall pattern.

There's the additional prior here that we know there were significant increases in COVID infection at the same time, which has a well-studied symptom of loss of taste and smell, that nicely correlates with these.

Sorry I'm a bit late to this, but those simulations take 10s - 100s of millions of Cpu hours (i.e. costs of millions - 10s of millions of dollars), so that's not practical.

It doesn’t really work like that. For instance, imagine you have a simulation with billions of particles in it. To construct reduced data you may need to use many fields (position, temperature, composition) of all particles over many outputs (usually at different times).

I have a PhD in astronomy :)

SpaceX is an environmental disaster that I guarantee we will look back on in 50 years as the company that kick-started kessler syndrome.

They have already, objectively, caused huge environmental light-pollution based damage.

If you want a python unit system that is backed by sympy you can use `unyt`: https://unyt.readthedocs.io/en/stable/

Well these are scientific articles and as such must be preserved, the cost of which must be paid for sooner rather than later. So it's not just the 'wire cost' of sending PDFs around, but the archival work too that needs to be paid for.

I also think that journals are too expensive, but that's why I only publish in society journals.

Is it just me or is this article entirely unreadable due to the in-page scrollable HP ads in the background? The page breaks aren’t calculated correctly so I can’t see half the text. Sigh.