gugagore

I am familiar with the situation you describe, and it's a good point.

However, it does suggest that there is an opportunity for factoring "Dog" out in the code, at least by name spacing (e.g. Dog.Model).

By the way, I love your blog post https://alexkritchevsky.com/2024/02/28/geometric-algebra.htm...

I didn't understand this part:

> I strongly believe that if GA would make this distinction they would lose a lot fewer people. It is a completely interesting and useful thing to talk about “a representation of a particular class of operations that makes composition and inversion easy”, and completely offputting when you blur the distinction between operators and geometric objects themselves, and write every operation in terms of the geometric product when only a few of them are really compositions of operators.

I can't tell what "a few of them" refers to. What is this potential distinction between operators and geometric objects? Sounds like the the distinction between a group action and a group object?

I am willing to believe that GA is an unnecessary renaming of other simpler things, and also that it has these kind of culty vibes, but I'm focusing on the claim that (I understood as) "unifying the operators and geometric objects" is a bad feature rather than a good feature.

Totally agree! And this is why I sympathize with GA fanatics. Although it might not really make sense to do the rebranding or be maximalist about the geometric product, I do think there is a shared goal of making these things easier to understand.

Happy to have agreement that this goes beyond "traditional linear algebra".

But, pedagogical treatment is a separate question from what is linear algebra.

Should these all be the same wikipedia page?

- https://en.wikipedia.org/wiki/Exterior_algebra

- https://en.wikipedia.org/wiki/Multilinear_algebra

- https://www.georgehart.com/research/multanal.html (okay, not mainstream enough to have a relevant page, but it is extremely relevant in any engineering practice of linear algebra)

That's fine. That's still a step beyond just "linear algebra".

What other way is there around avoiding polar vs axial vectors looking the same but behaving differently?

Or normal vs tangent vectors transforming differently?

This is reminiscent of MATLAB where there is no distinction between a scalar and a 1x1 matrix.

They are likely talking about "cued speech".

So when people call the (smaller) PC serial port connector DB9, they're technically wrong? TIL

I'm confused that you're bringing in case sensitivity into the discussion. For simplicity, early systems could not be case sensitive because they didn't have a large enough character set.

The comment about crashing due to spelling error really should be addressed by some form of static analysis, not by making name lookups more relaxed, in my opinion.

Technically, image resizing does in general care about the color channel ordering, because color spaces are in general not linear. https://www.alanzucconi.com/2016/01/06/colour-interpolation/

Maybe someone can explain why the modulation scheme matters, instead of the frequency range. AM less sensitive to phase noise? Or is it that AM radio typically means some frequency range?

The integral term handles an unknown (or unmodeled) additive error. So in that sense, you don't have the differential equation and you still apply PID.

Floating point addition, for example, is not associative, so the order of taking a sum affects the result. If the summation were sequential and single threaded, it would be deterministic. But it happens in parallel, so timing variations affect the result.

But there is probabilistic sampling that happens (see "temperature").

I can understand in principle but not actually. Converting between snapshot and patch representations is not an exponential effort operation. You could convert a git history to pijul (yes, expensive, but not exponential), do the conflict resolution in pijul land, and then handwaved convert back to snapshots.

Is sapling compatible? It gets rid of your .git directory.

This doesn't seem to be true. I can use git rebase to change the order of commits, and I don't get conflicts if the corresponding patches are independent.

If you can losslessly convert between the two models, then it seems like you could make git's conflict resolution smarter without changing the underlying data model.

Couldn't git just recreate patches from snapshot? I mean, this is what happens when you show a commit or a diff.

This is probably what you mean by how it's invisible to the user. When does it actually make a difference?

> when more reasonable modern default choices for all purposes outside of high performance applications would either be a rational composed of bigints, or a bigdecimal, or maybe a generating function for the coefficients of a Taylor series expansion or something like that if you really want to get nuts.

Rationals of bigints doesn't work in general. First of all, you still have to do approximations if you take e.g. a square root or a sine. And even if you have an algorithm that is closed under rational numbers (like the Simplex method for solving linear programs), the number of digits you need in your integers blows up unless you have a tiny problem.

My rough sketch is that inv(A) involves det(A) in the denominator, and the number of digits in det(A) is exponential in nature (n! if A is n-by-n).

The first thing the page shows is generating configurations in existing formats like YAML. For most of my use cases, there is tooling that does some updates of the configuration (e.g. bumping dependencies), so I'd really need bidirectional support to convert the updates back into Pkl.

I had a Internet connectivity problem on a JetBlue flight recently that I couldn't figure out. I could see the captive portal, with a URL like "planSelectionPage". I would check the agreement and hit "Let's go", and some JavaScript would trigger and look like a new page was loading, but would just stay on the same page. It happened on both Firefox and Chrome. My phone connected without a problem.

I wasn't sure how to debug it.

You mean it's conjugated like the third person singular, i.e. like você or ele or "a coisa" (the thing).

I would like to understand why JavaScript semantics have allowed for a lot of performance, but Python semantics hasn't.

OO is a lot of things to a lot of people, but it's not "structured control flow".

https://polyscope.run/data_adaptors/ has an interesting approach to handling custom C++ types. I wonder if that could carry over to this project.

The main thing I found to be an improvement is that Graphite maintains all the stacking metadata in git refs. With the other tools, I did not know how to develop a stack across multiple development machines.

I would be happy for someone to prove me wrong and show how, e.g. `git branchless` lets you work on a stack from multiple machines.

many problems seem to have the property that it's easier to verify a solution than to come up with one. If someone provides a filled-out sudoku puzzle, it's relatively straightforward to check if they've followed the rules and completed it correctly. However, actually solving the puzzle from scratch requires a different kind of thinking and might take more time.

Totally. Thank you for expanding on "typically".

If I can expand on your "kind of", it would be that because of the kernel trick, it actually does matter that the data itself can determine the "linear" (in an infinite dimensional space, that would require infinitely many parameters under the primal formulation) model.

SVMs typically have weights per data point. I.e. nonparametric/hyper parametric. Modern machine learning doesn't really work like that anymore, right?

One thing that I have seen in the Julia ecosystem, where multiple dispatch is pervasive, is that function names can get away with only being verbs. Without dispatch, you often see an approximation by baking in (Hungarian notation) the name of a type into the function.

This is most clearly seen with conversation functions between types defined in different libraries. A collection class might have a method like "from_Array".