HN user

hardmath123

1,692 karma
Posts140
Comments110
View on HN
twitter.com 2y ago

A Transformer Quine that generates a lossy copy of its own weights

hardmath123
1pts0
arxiv.org 3y ago

Gradient Descent: The Ultimate Optimizer

hardmath123
6pts1
arxiv.org 4y ago

Designing Perceptual Puzzles by Differentiating Probabilistic Programs

hardmath123
1pts0
kach.github.io 4y ago

Show HN: Shabdle Is Wordle in Hindi

hardmath123
1pts1
web.archive.org 4y ago

Searching for Playtronics

hardmath123
1pts0
www.cityu.edu.hk 4y ago

Finding a Horseshoe on the Beaches of Rio [pdf]

hardmath123
1pts0
hardmath123.github.io 4y ago

PeLU: Porcelain-Emulated Linear Unit

hardmath123
3pts1
www.jamestanton.com 4y ago

Teacup Twists and impossible braids [pdf]

hardmath123
1pts0
scribomechanica.weebly.com 4y ago

Scribomechanica

hardmath123
1pts0
www.pencilpages.com 4y ago

Seeking Pencils

hardmath123
3pts1
kach.github.io 5y ago

Generative Adversarial Web Development

hardmath123
10pts3
hardmath123.github.io 5y ago

Learning to Play the Chaos Game

hardmath123
149pts9
hardmath123.github.io 5y ago

Systems for Coordinating Coordinate Systems

hardmath123
2pts0
hardmath123.github.io 5y ago

Clogs and Leaks: Why My Tensors Won't Flow

hardmath123
1pts0
hardmath123.github.io 5y ago

Public Time and the Privacy of Time

hardmath123
1pts0
www.cryptologie.net 5y ago

Cute Cryptography Stories

hardmath123
1pts0
hardmath123.github.io 5y ago

Sol LeWitt and the Soapy Pit

hardmath123
2pts0
hardmath123.github.io 5y ago

Hybrid Hyperoptimization: The most ambitious crossover event in autodiff history

hardmath123
6pts0
github.com 5y ago

Show HN: Generate dependency diagrams like from today's xkcd

hardmath123
2pts0
hardmath123.github.io 5y ago

GPT-3 and Claude Shannon Visit the Tartu-Moscow School

hardmath123
1pts0
hardmath123.github.io 6y ago

A Differentiable Anti-Anti-Aliasing Adventure

hardmath123
2pts0
hardmath123.github.io 6y ago

Conway's Gradient: Reversing Life with differentiable programming

hardmath123
28pts2
github.com 6y ago

Hoover Tower Hyperlapse

hardmath123
1pts0
hardmath123.github.io 6y ago

Spaghetti's Shadows are Spirali-Shaped (reasoning about PCA on trajectories)

hardmath123
4pts0
hardmath123.github.io 6y ago

Words that do Handstands

hardmath123
324pts51
robertheaton.com 7y ago

Third-Party Dream Cookies

hardmath123
3pts0
chromium.googlesource.com 7y ago

(Chromium's) Guidelines for URL Display

hardmath123
1pts0
www.cryptologie.net 7y ago

Proof of Elgamal's semantic security using a reduction to DDH

hardmath123
1pts0
hardmath123.github.io 7y ago

Tuning a Typewriter

hardmath123
59pts7
www.lockhaven.edu 7y ago

Physics puzzles

hardmath123
140pts7

A lot of replies here are missing the key idea behind Penrose: that the "substance" and "style" of any mathematical diagram are separable. For example, the "substance" of a diagram might be a collection of sets with known subset relationships:

Set A, B, C

IsSubset(B, A)

IsSubset(C, A)

That same substance can then be rendered in many different styles, e.g. as a Venn diagram or as a tree diagram (substance vs. style is a lot like HTML vs. CSS). Importantly, Penrose's vision is that experts will author libraries of domains and styles, and end-users need only express the substance of their diagrams (i.e. the three lines of code above).

The second beautiful idea in Penrose is that diagram generation is expressed as a constrained optimization problem. This lets you easily experiment with layouts by writing constraints and sampling a variety of potential diagrams via stochastic search.

These two ideas set Penrose apart from most other diagram software out there. I really hope it gains wider adoption. Give it a try!

How cool! I was on the fence about whether or not to put up the source code— I'm _glad_ I did!

Do you have a recommendation for a good reference that teaches about the various metrics for point-cloud distance? (I only used Chamfer distance because I hazily recalled it from some undergrad class taken a while ago...)

Actually, he’d be very unlikely, even with the imagined high tech that I gave him, to be able to control whether the coin came down heads or tails—especially if you provide the coin.

It's funny, Dennett talks about a "coin tossing machine" as a virtual impossibility, but Persi Diaconis actually had one built! It is described in this paper: https://statweb.stanford.edu/~susan/papers/headswithJ.pdf

Physicists have calculated that if you really want to predict coin flips, you have to know the location of every electron in the visible universe and its gravitational effect on the coin.

This is not really true either. Coin tosses leak a lot of information into the physical world. For example, here is a short paper on trying to predict the outcome of a coin toss by analyzing the sound of the "ping!" it makes: https://cs.stanford.edu/~kach/can-one-hear-the-fate-of-a-coi...

As children, we do not understand that the things we love do not belong only to us. Then, I was not prepared to relinquish music to anyone else’s ownership. Now, I have finally loosened my grip.

Re: Earley being forgotten, I've been working on the "nearley" Earley parsing library for JS for many years now and it now has a very solid user-base (200+ dependents on npm). It's probably not the fastest a JS-based parsing library could be, but it's certainly no longer forgotten! Here's a small sample of some of the amazingly varied projects using Earley parsing (via nearley) today… https://nearley.js.org/#projects-using-nearley