thatcherc

Slim binaries and better static compilation were mentioned as big future features in the closing remarks, so that'd be a good one to watch - a lot of the people there working in embedded, robotics, or plane satellite projects were excited about those.

There weren't any talks I saw where anyone was actually running Julia in space, but Ronan Chagas' "Attitude control subsystem development using Julia" and Corbin Klett's "Realtime embedded systems testing with Julia" both get pretty close and point out reasons why you wouldn't quite want to run Julia on a satellite or plane just yet. Hopefully the videos capture the questions from the audience because there were some good followups!

If JuliaCon last week is any indication, there might be more Lisp in space in the future! As long as the runtime size can be shrunk down a bit, but it sounds like that's in the works too.

I'm reading that right! It's cool. It's all Scheme, so get ready for that if you haven't seen it before. It's really interesting stuff though, all about how to make really extensible systems. I'm not sure how much I'll use the tools it talks about day to do, but it is really neat to have a much deeper understanding of things I often take for granted like pattern matching (regex) and generic dispatch and games solvers can be implemented from directly scratch.

As the author notes, this is very much in the style of the aphyr's Interview series of stories like "Rewriting the Technical Interview" - [0] - which are all great reads as well.

[0] - https://aphyr.com/posts/353-rewriting-the-technical-intervie...

More of a collection of notes exported from my Joplin notes than a regular blog:

https://jthatch.com/notes/index.html

Yes! In principle at least, and only miniscule/undetectable amounts of energy. The first gravitational wave detectors build were these "Weber bars" [1] - big block of aluminum that were supposed to resonant mechanically (like a bell) when a gravitation wave of the right frequency passed through them.

Looks like these things haven't detected a real gravitational wave, but if a strong enough one at the right frequency came through, they might start ringing like (very quiet) bells!

[1] - https://en.wikipedia.org/wiki/Weber_bar

I'll go in to plug the Clockwork Rocket series (Orthogonal and its sequels) and Incandescence, all of which have much more to do with physics and spacetime than Permutation City and short story collections like Instantiation (which I also love).

I think Incandescence is my favorite book of the physics set, but the Orthogonal books and their corresponding web notes [1] are an absolute tour de force of deriving a whole universe of physics from a tiny modification to the rules we're familiar with. It was wild to read those as a physics undergrad and still very cool to think about today.

[1] -https://www.gregegan.net/ORTHOGONAL/00/PM.html

Not OP, and I'd have to see the prompt to run it on my set up, but I but together a refurbished HP Chromebox with some extra RAM and an SSD for a total of $250 and it'll run all the 13B models about as fast as I can type on my phone, just as a reference point. I'm consistently impressed by it even if it's not blazing fast.

Do you have code for this (or pointers to what you used to set this up)? I'd love to set up something similar!

Weird article. This section stuck out:

> Hajimiri, who led the Caltech that developed MAPLE, explained how the wireless transmission of energy through space is based on a quantum phenomenon called “interference.”

And then they go on to describe classical wave interference. I think the point they were trying to make is that the satellite uses a digital phased array antenna to direct its energy downward, but there's nothing quantum about that - it's the same technology used in Starlink terminals and new-ish wifi routers (among many other applications).

They also don't list how much power was received! That's the one figure that matters here I think. I'm curious whether it was watts, milliwatts, or kilowatts.

This is very helpful, thanks! Looks like I have some reading to do about IP forwarding.

Fairly off-topic, but I've been having the hardest time finding a Wireguard configuration guide that lets me connect two peers (my phone and an SBC at my house behind my router) to a VPS peer (with a public IP) in a way that routes all the traffic from my phone through the SBC (via WG) and out to the internet via my home fiber connection. All the blog posts and tutorials I've seen have traffic going out through the VPS peer, with little explanation of how all the firewall and iptables commands might change if I wanted a different configuration. Has anyone seen a configuration like that, or know which networking concepts I should keep searching for to go in the right direction?

Author seems to be using billion == 10^12 instead of the common billion == 10^9. A lot of the math still works out since there's a multiply and a divide by a billion, but it is a little confusing to see passages like this:

> Given the parameter count, we can multiply by two to get bytes. So to calculate the size of the weights for a 52B model.

> 52e12⋅2 = 104e12 bytes ≈ 104GB

It's the same reason the Voyagers launched in the late 70s - the Grand Tour configuration of the planets [0]. The outer planets align in a way that lets you chain together several gravity assists to achieve high enough speeds to escape the solar system. The trick is that configuration only occurs every 175 years. I don't think we have any space vehicles right now that could produce that same kind of delta-v with their own chemical/electrical/nuclear power.

[0] - https://en.wikipedia.org/wiki/Grand_Tour_program

Could you point me toward any more info or keywords on "the cartpoll demo famously tripped up derivative based reinforcement learning for awhile"? It sounds like an interesting bit of history but my google searches are only bringing up python tutorials.

Good thread!

https://threadreaderapp.com/thread/1642726595436883969.html

Have you checked out Symbolics.jl? It's a lot like SymPy.jl but with all the niceness of a implementation in native Julia instead of a python binding.

Any idea how this picture was made? Was the Goldstone dish scanning a really tightly focused beam across the region of space where this asteroid was, or are they relying on some kind of doppler or synthetic aperture radar effect with the relative movement of the dish (on a rotating Earth) and the asteroid let the image be recovered out of Doppler and timing info?

I went looking for some details on how the picture was produced on the NASA page linked at the bottom of the article but I can't find much technical info on the process. However it was done I'm sure it's a cool technique!

Here's another article that seems to be a summary of the Bloomberg article (but without the paywall) - [0]

The ban under consideration seems to be a ban on the sale of new gas stoves, although I don't think the article actually says that explicitly.

[0] - https://financialpost.com/pmn/business-pmn/us-safety-agency-...

That is a great link list! I've been curious about converting over a few micro-USB devices I have around so I might pick up some of these boards.

Do you have any recommendations for search terms for boards that would take in DC power (12V+) and make that available for output over a USB-C connection?

One thing I've wanted to do is put a battery charge monitor on the escooter I leave at work. I'd like it to report the charge back to a computer I have at my house. Running WireGuard on an ESP would let me send battery telemetry back to machine I control, even if the scooter was on public wifi, and without having expose any of my ports to the wider internet (except 51820).

Pre-covid (and pre-renovation) it had a laser cutter too! Hoping that comes back soon. By all accounts it was really popular and well used.

That 100% efficient figure is correct - you get one Watt of heat for one Watt of electricity. It's just that heat pumps can deliver 3 or 4 Watts of heat for each Watt of electricity (usually quoted at around 300% or 400% efficiency!). Compared to that, resistance heaters aren't as efficient.

The explanation that made the most intuitive sense to me is that it takes less energy to move heat from one place to another (air at 273 Kelvin to air at 300 Kelvin, like a heat pump does) than it does to create heat from nothing (like a resistor does). That's why the heat pumps can get deliver more heat to you from the same amount of electricity.

"Volts" by David Roberts![0] A lot of subjects covered related to decarbonization and getting and using more renewable electricity, usually by interviewing a variety of experts

[0] - https://www.volts.wtf/

That's a big takeaway I got from the book Electrify by Saul Griffith (a really good read). The book only really looks at US carbon emissions, but a substantial chunk of emissions comes from refining and transporting petroleum projects -- these are emissions that occur just to prepare gasoline to be burned! Switching to electric cars (or just fewer cars) has the same double bonus as you mentioned in that there's less tailpipe emissions and less petro processing and transportation emissions as well.

The big this seems to be that their noising/denoising process is invertible (to some extent) so it can be used to generate uncertainties, which is something the article says diffusion models can't do. The intuition is that the "flow" process used here is reversible (imagine going backwards in time and watching a cloud of charged particles coalescing into a bunch), which gives some nice properties that random blurring used in more common diffusion models does not.

I think it's the reversible flow part that's important for the results, not the connection to the physical electrostatics system. It just happens to be that electrostatics have this nice flow behavior too and are a pretty approachable analogy for what the model's doing.

Section 2.1 in their linked paper (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/201...) gives some clues. I think this is what's going on:

GPS receivers work by figuring out how far away they are from a number (>3) of GPS satellites. The receiver knows where the GPS satellites are (since the satellites broadcast their orbit parameters) so if a receiver knows how far it is from several satellites it can work out where it is itself.

Now, as the GPS satellite signals travel through Earth's atmosphere, they can be slowed down by different atmospheric effects. A slower signal will cause the receiver to think it's farther away from a satellite than it really is, so the receiver might estimate that it's position has changed a little bit. However, if you know the receiver's position hasn't change (maybe it's fixed in place to a big rock), then you can attribute the receiver's measured "change in position" to a change in atmosphere characteristics.

In this paper, they seem to have lots of fixed GPS receivers all over the place. By looking at all of them together, they can make a sort of map of the atmosphere characters in a part of the sky that's affected by rocket launches. The authors see these big ripples emanating from a Falcon Heavy launch in the US and this tweet shows those same ripples emanating from a launch site in North Korea.

I had a real "click" moment with Statistic Rethinking (2nd Edition) too! I never really "got" Bayes' Theorem before (could use it to solve homework problems and stuff but never appreciated it) but somewhere in SR Chapter 2 the whole point of it really snapped into focus. Good book!

> 200 MHz on a Xilinx KU060

> (That specific clock frequency on that specific part carries heavy hints about what this core is intended for.)

Fun clue! Looks like the Xilinx KU060 is a rad-hard FPGA for space applications. Does anyone know what 200 MHz might imply? Comms maybe?

National Institutes of Health?