thatcherc

This looks fantastic! The table and formula extraction features are especially interesting. My immediate question is: can this be integrated into Zotero? Most of the PDFs I read are research papers and extracting tables and formulas directly from my zotero collection would be super handy.

Can you explain that one (the logarithm trick)? I have no idea what it's trying do to. Is it saying that it's convenient or useful to know the order of magnitude of a number rather than its value?

It's weird to see a new vehicle announced like this that's not an EV. I wonder what it's like launching a gas truck when new battery-powered trucks are looming in the distance (or already here).

Link to the paper: https://www.nature.com/articles/s41377-025-02119-y

From the abstract:

> Here, we demonstrate a covert communications method in which photon emission is rapidly electrically modulated both above and below the level of a passive blackbody at the emitter temperature. The time-averaged emission can be designed to be identical to the thermal background, realizing communications with zero optical signature for detectors with bandwidth lower than the modulation frequency

It sounds like maybe they're modulating the emissivity of a diode up and down so that over time, its IR spectrum looks like black body radiation. Only someone looking at the intensity of the thermal radiation coming from the diode at really fast timescales (kilohertz or megahertz) would notice that there was a signal being transmitted.

This looks great! I've been using ThumbsUp[1] for a similar purpose (creating a gallery of photos I can push S3), but adding album and photo captions required some un-ergonomical tricks. I'll try this out!

[1] - https://github.com/thumbsup/thumbsup

https://jthatch.com/ !

You might have been thinking of https://earth.nullschool.net/

I appreciated the Pluto.jl mention! Going from Pluto notebooks that understand data flow to Jupyter notebooks where you have to tell the computer which order to run the cells in is always baffling to me. Why doesn't Jupyter know the run order and dependencies already? The way Pluto handles dependencies between code cells is really just so nice.

Kinda weird - Coolify doesn't come up except in the first and last paragraphs. Seems like the page is incomplete or just mis-titled.

Yup, slight mix-up. Gravity waves are waves in the ocean and atmosphere (or other fluid bodies) where Earth's gravity is the restoring forces that causes wave propagation. Gravitational waves are the waves in spacetime caused by powerful astronomical events like black hole mergers.

Cool! It looks like the Wansview Q5 has a similar SoC/camera/wireless setup as the Wansview W7, which as an installer guide on the Thingino wiki [1]. I wonder if that same installation process (but with the q5 firmware) would work. For $16 I'm inclined to try it out.

[1] - https://github.com/wltechblog/thingino-installers/tree/main/...

Thanks! The author's blog is a great find as well - https://www.dm5tt.de/

> The analog ones are easy to play with. You just need a DAC to drive their VCO and then can sample the I/Q pins with an ADC

Do you have any reference or notes on how to access the IQ pins on one of these devices (ideally one of the FMCW ones)? I've been wanting to play around with one of these 24 GHz or 60 GHz units for coherent radar but it seems like most of the boards only report on distances over serial links. If there's an easy way to tap into the analog IF signal after down conversion I'd love to see how to do that!

Cool! Do you like that approach? I've thought about setting up that exact thing but I wasn't sure how well it would work in practice. Are there any pitfalls you ran into early on? I might give it a shot after your "very easy to set up and operate" review!

I would recommend just trying it out! (as long as you have the disk space for a few models). llama.cpp[0] is pretty easy to download and build and has good support for M-series Macbook Airs. I usually just use LMStudio[1] though - it's got a nice and easy-to-use interface that looks like the ChatGPT or Claude webpage, and you can search for and download models from within the program. LMStudio would be the easiest way to get started and probably all you need. I use it a lot on my M2 Macbook Air and it's really handy.

[0] - https://github.com/ggml-org/llama.cpp

[1] - https://lmstudio.ai/

> you can even chroot into foreign CPU architectures, which is a handy thing to have when you mount your phone's eMMC

This sounds very interesting! What's the scenario where you'd do this? Would you be, for example, emulating an ARM processor with qemu on an x86 computer and chrooting into Android on an eMMC?

A challenge there is that there are very few stable lunar orbits! High orbits are perturbed by Earth's gravity (3-body problem) and low lunar orbits are perturbed by the lumpy distribution of mass in the Moon's interior [0]. Lunar GNSS satellites with a little bit of onboard propulsion could probably correct for some of these perturbations but once they ran out of fuel they would have a limited orbital lifetime.

[0] - https://en.wikipedia.org/wiki/Lunar_orbit#Perturbation_effec...

This sounds like something I'd want to do! Is the idea that you'd have a public domain name like "internal.thatcherc.com" resolve to an internal IP address like 10.0.10.5? I've wondered about setting this up for some local services I have but I wasn't sure if it was a commonly-done thing.

Yup, my data is nicely gridded so I can use the convolution approach pretty easily. Agreed though - missing data at the edges or in the interior is annoying. For a while I was thinking I should recompute the SG coefficients every time I hit a missing data point so that they just "jump over" the missing values, giving me a derivative at the missing point based on the values that come before and after it, but for now I'm just throwing away any convolutions that hit a missing value.

I actually have one for this! Last week I had something really specific - a GeoTIFF image where each pixel represents the speed in "x" direction of the ice sheet surface in Antarctica and I wanted to get the derivative of that velocity field so I could look at the strain rate of the ice.

A common way to do that is to use a Savitzky-Golay filter [0], which does a similar thing - it can smooth out data and also provide smooth derivatives of the input data. It looks like this post's technique can also do that, so maybe it'd be useful for my ice strain-rate field project.

[0] - https://en.wikipedia.org/wiki/Savitzky%E2%80%93Golay_filter

This is my favorite write up of someone truly building a GPS receiver from scratch: http://www.aholme.co.uk/GPS/Main.htm

A real tour de force of DSP - really cool stuff, and well-written

Sounds like a cool model. I'd love to try it but it seems like they're not releasing it (yet?). I've really gotten spoiled with the recent language models where I can download and run any new model or fine-tune I hear about. It's gotten to the point where I don't feel a model is very relevant unless I can run it locally. Hopefully a local version of this becomes available because I have plenty of time series data I'd like to run through it!

This list is weirdly not that helpful. `planor`, the planetary orbital calculator, sounds cool, but it's actually "A TUI client for cloud services for aws, vultr, heroku, etc." No orbits in sight! Now I'm not sure which of the descriptions are accurate.

> System76 claims 14 hours, and I managed 11 hours in our battery drain test (looping a 1080p video). In real-world use, I frequently eked out over 13 hours. That’s off the charts better than any other Linux laptop I’ve tested recently.

This is the most intriguing part to me. I've been mulling a Framework for a while but what's held be back so far is that the battery life is 9-10 hours after tuning, and I'd love it to be longer. 11 hours of continuous video on this laptop is pretty impressive for a Linux laptop I think. My Dell XPS 13 feels like it needs to go onto the charger every 4 hours so this would be a huge improvement.

Also curious. I don't see anything like that listed on the Ballotpedia page [0].

[0] - https://ballotpedia.org/Maine_2023_ballot_measures

They work phenomenally well on induction stoves! I use a single-burner plug-in induction unit in my small apartment kitchen and it works so well cast iron pans and an iron dutch oven.

Really cool that the image patches are converted to tokens with just a linear projection instead of a big embedding model! I wonder if that trick will prove viable for other multimodel media like audio.

Are the pictures in that second link from a diffusion model? They look flashy but they don't look like any spacecraft I recognize (and are weirdly asymmetric) and especially don't look like the Starlink satellites SpaceX has launched so far, which look more like [1].

[1] - https://spacenews.com/wp-content/uploads/2019/05/Starlink-Sc...

I don't really know how to read this page, but it looks like the time of closest approach has already passed -

> Closest approach time of impactors: t_min = 2023/08/14 04:48 TDB and t_max = 2023/08/15 12:22 TDB

If so then we're all good. Can the poster or anyone else chime in with background on this?

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!