Happy New Year from Reykjavík, Iceland!

Hmm, good idea. There are names that have the exact same pronunciation yet have different patterns of declension, for example:

- Ástvaldur -> ur,,i,ar - Baldur -> ur,ur,ri,urs

The "aldur" ending is pronounced in the exact same manner, but applying the declension pattern of "Ástvaldur" to "Baldur" would yield:

- Baldur - Bald - Baldi - Baldar

The three last forms feel very wrong (I asked my partner to verify and she cringed).

Spoken Icelandic is surprisingly close to its written form. I wouldn't expect very different results for the trie if a "phonetic" version of names and their endings were used instead of their written forms

It would be good to cover more names for sure -- that's an ongoing process at DIM. Names are frequently added to the approved list of Icelandic names, so there's always going to be some lag.

I would not be confident enough myself to add the data myself since I'd probably be wrong a lot of the time. When reviewing the results for the top 100 unknown names I frequently got results that I thought _might_ be wrong, but I wasn't sure. For those, I looked up similar names in DIM to verify, and often thought "huh, I would not have declined those names like this". For that reason, I rely on the DIM data as the source of truth since it's maintained by experts on the language.

As a native Icelandic speaker, I have an intuition for how to decline names -- I don't really think about it consciously. I'd assume that for most people it's just pattern matching.

Native speakers very frequently decline names in ways that are not technically perfect but sound correct enough. For example, my name (Alex) should not be declined, but people frequently use the declension pattern (Alex, Alex, Alexi, Alexar).

There's some parallel to be drawn with how the compressed trie applies patterns that it's learned to names. That's at least how I thought about it when designing the library.

Hmm, good point. I’m abroad this week and don’t have a laptop, but I’ll look at adding an MIT licence to the shader itself when I’m home. (I don’t want the post contents to be MIT licensed, but the website code and examples should be)

Thanks for raising this

Thanks a lot! Yeah tons of work went into this post, I’ve been working on it since November. The shader itself was about 2-3 weeks of evenings to get the effect 90% of the way

Thanks! I've been trying to limit excessive detail and tangents in my writing (my posts tend to be on the longer side). Great to hear I managed to strike a good balance here.

This was easily the greatest article ever written on the web clipboard interfaces.

I didn't even know about isTrusted until a couple of weeks ago when I was writing some widgets to speed up some tasks by automating a bunch of web controls. I couldn't get my auto-paste to work, discovered that flag and went down a rabbit hole of trying to over-ride a safety mechanism on my own local browser (not easy!).

They're created with ThreeJS and react-three-fiber. My website is open source, you can find all of the 3D scenes here: https://github.com/alexharri/website/tree/master/src/threejs...

Nice work, one thing (which you'll see everywhere once you've fixed it yourself) though: in TeX you want to escape functions like cos, i.e. use "\cos" not "cos", which will get rendered as a product of 3 variables c, o and s.

Just pushed a fix changing "cos" to "\cos", looks much better. Thank you!

I haven't heard of this before, I'd be interested in learning more about how this is useful. Do you know of a good resource to read more about this?

I honestly don't know, I encountered the concept during my study, but it took lots of practice to get comfortable with it. Most articles I encounter are too practical to really foster understanding, or too technical to work as an introduction.

You can start with a description of how projective matrices work (and how translation and rotation are related to it). After that, best tips I can give are start with 2D until you can't bear to see another cross product. Then get familiar with Cramer's rule and higher dimensions. You'll need sone fluency in linear algebra.

My first practical use of the concept was to rectify photographs where e.g. a building was not quite upright. That might be a good starting point.

Thanks a lot, will use this as a starting point

This post was intended as a source of information for developers who might want to implement vector networks in their own projects.

Back in 2018 I wanted to support vector networks in my animation editor project, but had a really hard time finding information. I didn't have a math or CS background, so just knowing what concepts to Google was somewhat of a brick wall. Googling "vector networks" didn't yield any useful results, because there weren't any resources discussing them (aside from marketing content from Figma). So in writing this post, I wanted to create the resource I needed back then.

This is a nice writeup, but the definition and usage of determinant makes my nails curl.

A determinant is defined for matrices, not for vectors and from the purpse of your article I think you skipped over that part for simplicty/brevity. But really, introducing it once properly, then going to the more pragmatic definition of yours aknowledging that it is skipping some parts would be so much better for someone not knowing this field in depth. Or just not calling it determinant. And mentioning that determinants show a sort of scale factor of transformation, also for 3D, would be nice to connect to 3D rendering and such.

Since it is meant as intro/tutorial on implementation I would really point into all the right math-relates directions as a reference. Whoever wants, can read or skip those parts.