Sun Woo Kim

How does it work?

A pedantic comment: the image pictures a WarCraft III pro.

Surprisingly LLMs have only recently surpassed the usefulness threshold for me: Generating an iCal file from a pdf or a website. Few months ago they would fail miserably, but now they can do it, with some back and forth. Now I'm fully convinced that they're going to enslave us all

Is there an example where generating randomness is the bottleneck?

Wouldn't it be more like a 1d chain of springs? What makes it Ising?

Hi I'm a PhD student working on QEC and many body physics and I'd like to join! Also, I would be interested in doing an internship in the next available round.

is this vibe proving

My aching back, that's what

Okay, I should read this, but is it at all related to acceptance rate when you do MCMC? I heard that the general rule of thumb is you should be accepting ~0.2 fraction of the time, for some reason

Yes but what is the reference :( please bring me in the loop

What do the numbers mean Mason

Thanks for coming!

interesting talk about quantum hidden markov models and inference by @sunwoopkim.bsky.social

😮 i really need to start working on it 😬

Forgot to say: the quantity that is conserved is the number of greens on the even sites minus the number of blues on the odd sites. It was found using the linear equation described in the slides! (8/7)

(7/7)

(6/7)

(5/7)

(4/7)

(3/7)

(2/7)

If I gave you these evolution rules, can you tell me what will stay constant during the evolution? Very belated post on recent work on hydrodynamics of circuit systems! Work with Friedrich Hübner, Juan P. Garrahan, and Benjamin Doyon. See more here: arxiv.org/abs/2503.08788 (1/7)

Oh I did not know that the red cone has a peak at the same place where the blue does as well; I guess it means that only activating blue would correspond to a 'new' color as well..

Just for curiosity, is this comic AI generated?

Question: What was the main bottleneck for these universal wavefunctions before, and what was the main breakthrough? As an anecdote, I remember I was trying to train a single neural network for the whole parameter space of TFI model for fun, but it wasn't training very well.

Stage 1: there is Markov's inequality
Stage 2: there is Markov and Chebyshev
Stage 3: there is Markov and Chebyshev and Chernoff and Hoeffding
Stage 4: there is Markov and Chebyshev and Chernoff and Hoeffding and Bernstein and...
Stage 5: there is Markov's inequality

thanks i hate it

The point of compressed sensing is that you 'project' your data onto this basis, then set the higher order ones to zero or something like this right? I remember it was used for MRI at some point (though surpassed by DL methods now)

What's the favicon (Red/Light blue thing)? I'm guessing maybe a covariance matrix?

Le Monde reporting that a French scientist traveling to Houston to attend a conference was denied entry to US after a search of his phone & computer revealed messages critical of Trump's science cuts, "which [says CPB] conveyed hatred of Trump & could be qualified as terrorism". Computer confiscated