tangentstorm (@tangentstorm) 's Twitter Profile
tangentstorm

@tangentstorm

I like to learn cool things and explain them to people. I also write software and sometimes I make video games.

ID: 358055613

linkhttp://tangentstorm.com/ calendar_today19-08-2011 09:14:30

1,1K Tweet

380 Followers

930 Following

tangentstorm (@tangentstorm) 's Twitter Profile Photo

Not 100% $GME related, but it has to do with another ticker that has an unusually high level of DRS ownership: So there's a company called Fundrise that basically has been running a real estate investment trust for years. A while back, they started a new fund that invests in

tangentstorm (@tangentstorm) 's Twitter Profile Photo

Using Claude to decipher and retcon the ideas behind an inscrutable (to me) paper from 1867... Often Claude's descriptions are also inscrutable, but I just keep asking questions, and it just keeps getting clearer. There's a really neat story to tell here, and I think it's going

Using Claude to decipher and retcon the ideas behind an inscrutable (to me) paper from 1867... Often Claude's descriptions are also inscrutable, but I just keep asking questions, and it just keeps getting clearer.

There's a really neat story to tell here, and I think it's going
Tansu Yegen (@tansuyegen) 's Twitter Profile Photo

A new system prints full size boat hulls in one piece using a fully automated 3D process, creating strong vessels up to 12 meters long with no cutting or assembly, mainly for defense and work boats 🚤

tangentstorm (@tangentstorm) 's Twitter Profile Photo

This is a fun #jlang technique: replace + and * with verbs P and M that "add" and "multiply" strings. Polynomials are just vectors of coefficients in J, so: 'abc' represents the polynomial a + bx + cx² , and: 'abc' PPR 'def' ... "multiplies" that polynomial by d + ex + fx².

This is a fun #jlang technique: replace + and * with verbs P and M that "add" and "multiply" strings.

Polynomials are just vectors of coefficients in J, so:

'abc' represents the polynomial a + bx + cx² , and:

'abc' PPR 'def'

... "multiplies" that polynomial by d + ex + fx².
tangentstorm (@tangentstorm) 's Twitter Profile Photo

The math problem that I got nerd-sniped on has lead me to vibe code a custom SAT solver in rust. For this particular problem - or at least the smallest versions of the problem that I can get to run, it's actually faster than CaDiCaL, a state-of-the-art solver written by an

tangentstorm (@tangentstorm) 's Twitter Profile Photo

TT(26) now down to just under 19s. TT(28) takes about 40 (a solution just happens to be early in the search space). TT(30) has been running for over an hour. Many of the techniques I'm using break down for TT(56) simply because the search space is gigantic. I'm not really

TT(26) now down to just under 19s. TT(28) takes about 40 (a solution just happens to be early in the search space). TT(30) has been running for over an hour.

Many of the techniques I'm using break down for TT(56) simply because the search space is gigantic. I'm not really
tangentstorm (@tangentstorm) 's Twitter Profile Photo

I found this to be quite an inspiring story, especially as I've been reading Cedric Villani's "Birth of a Theorem" and feeling like I'm not even in the same intellectual (or at least educational) ballpark as even an "entry-level" mathematician. I don't really even know that much

tangentstorm (@tangentstorm) 's Twitter Profile Photo

I am having such a fun time with this math/SAT problem. This is the workflow I always dreamed I'd have with bex, where I have a single process I'm trying to optimize, and I just make it run faster and faster, by continuously looking for the constraint, and breaking it. I *have*

tangentstorm (@tangentstorm) 's Twitter Profile Photo

What astounds me is that even though this dang robot seems to misunderstand everything I say and only implement the dumbest possible version, it's still getting way, way faster at each step.