Shubhendu Trivedi
@shubhendu.bsky.social
Interests on bsky: ML research, applied math, and general mathematical and engineering miscellany. Also: Uncertainty, symmetry in ML, reliable deployment; applications in LLMs, computational chemistry/physics, and healthcare.
https://shubhendu-trivedi.org
https://shubhendu-trivedi.org
Seemed to be something people were going on over on twitter. But I did not understand since I avoid scrolling entirely. For example:
November 20, 2025 at 3:31 AM
Seemed to be something people were going on over on twitter. But I did not understand since I avoid scrolling entirely. For example:
So I'll just post an older set of pictures. A small number of books here are missing. Didn't take pictures of those when I managed to get hold of them.
November 16, 2025 at 6:38 PM
So I'll just post an older set of pictures. A small number of books here are missing. Didn't take pictures of those when I managed to get hold of them.
But at least I managed to get all of them (20 or so are easy, 10-15 somewhat less so, the rest not easy). At the time it felt like a silly project, but now I am v. happy about it. I was trying to take a good picture of the lot, but not able to get them right. Softer lamps also seem to have given up.
November 16, 2025 at 6:38 PM
But at least I managed to get all of them (20 or so are easy, 10-15 somewhat less so, the rest not easy). At the time it felt like a silly project, but now I am v. happy about it. I was trying to take a good picture of the lot, but not able to get them right. Softer lamps also seem to have given up.
Programmed periodic re-reading
November 13, 2025 at 2:45 AM
Programmed periodic re-reading
The mathematical structure here is quite interesting, but probably not worth to get into for a thread. In the meantime, can just use the Conway book as a copout, which covers them in the first ~quarter or so (although it's also a bit too cute).
November 13, 2025 at 12:41 AM
The mathematical structure here is quite interesting, but probably not worth to get into for a thread. In the meantime, can just use the Conway book as a copout, which covers them in the first ~quarter or so (although it's also a bit too cute).
Another reality is people getting multiple tacky (but I like them) hardcovers, and placing them randomly next to Darwin.
November 11, 2025 at 2:42 AM
Another reality is people getting multiple tacky (but I like them) hardcovers, and placing them randomly next to Darwin.
विकलांग श्रद्धा का दौर
November 11, 2025 at 1:25 AM
विकलांग श्रद्धा का दौर
Sage era. One with mullet-like hair, one without. Same time. Back when Royal Enfields weren't too common.
November 7, 2025 at 11:50 PM
Sage era. One with mullet-like hair, one without. Same time. Back when Royal Enfields weren't too common.
Reminded me of a figure from the PhD thesis of Ferenc Huszar
November 5, 2025 at 10:21 PM
Reminded me of a figure from the PhD thesis of Ferenc Huszar
From a great book:
November 3, 2025 at 2:41 AM
From a great book:
Harvard classics also have a wall of bricks aesthetic. But they have many different versions, some of which were very well-done. But again, they were heavily marketed to be aimed at "home readers," "curious autodidacts," "precocious teenagers" and similar, so it can be forgiven.
November 2, 2025 at 3:56 PM
Harvard classics also have a wall of bricks aesthetic. But they have many different versions, some of which were very well-done. But again, they were heavily marketed to be aimed at "home readers," "curious autodidacts," "precocious teenagers" and similar, so it can be forgiven.
Talking of toothpaste-coloured bricks. The Nobel prize used to commission different "libraries" in the 70s and 80s. These aren't meant to be as expansive as Loeb. They are meant to be for everyone, have no high culture pretensions, yet can be easy on the eyes.
November 2, 2025 at 3:50 PM
Talking of toothpaste-coloured bricks. The Nobel prize used to commission different "libraries" in the 70s and 80s. These aren't meant to be as expansive as Loeb. They are meant to be for everyone, have no high culture pretensions, yet can be easy on the eyes.
Anyway, boring stuff aside, I still have my moto razr v3.
November 1, 2025 at 2:29 AM
Anyway, boring stuff aside, I still have my moto razr v3.
As an aside: I always like to say that The Waves is the almost opposite of Memoirs of Hadrian (someone tried to describe why it's the way it is very beautifully in a review screenshotted), yet they have similar effects in ways. The former has six interior voices, basically polyphonic monologues that
October 26, 2025 at 5:10 PM
As an aside: I always like to say that The Waves is the almost opposite of Memoirs of Hadrian (someone tried to describe why it's the way it is very beautifully in a review screenshotted), yet they have similar effects in ways. The former has six interior voices, basically polyphonic monologues that
Finding this specific version and cover is still a goal that I haven't given up on.
October 26, 2025 at 4:41 PM
Finding this specific version and cover is still a goal that I haven't given up on.
This book is just so beautiful
October 26, 2025 at 4:39 PM
This book is just so beautiful
Writing it seems like a bigger challenge, and needs more time, because of the cursive, even if you love cursive. So did a cop out and had read a book on its history instead.
October 24, 2025 at 10:31 PM
Writing it seems like a bigger challenge, and needs more time, because of the cursive, even if you love cursive. So did a cop out and had read a book on its history instead.
Using abigudas as a prompt for more randomness. Another reason I like these systems is because you can literally sit with the script, commit the system to memory over 10-11 sittings, and easily write whatever. e.g. here the abstract of a paper of ours in Brahmi (which I learnt during early covid).
October 24, 2025 at 10:21 PM
Using abigudas as a prompt for more randomness. Another reason I like these systems is because you can literally sit with the script, commit the system to memory over 10-11 sittings, and easily write whatever. e.g. here the abstract of a paper of ours in Brahmi (which I learnt during early covid).
A random "fun fact": English is obviously the best example of a language with a deep orthography (Guy Deutscher had a very nice coverage of the history, followable for a non-linguist), but languages written in abiguda _tend_ towards phonemic orthographies (the two things naturally [+]
October 24, 2025 at 12:12 AM
A random "fun fact": English is obviously the best example of a language with a deep orthography (Guy Deutscher had a very nice coverage of the history, followable for a non-linguist), but languages written in abiguda _tend_ towards phonemic orthographies (the two things naturally [+]
A random form momentarily took me back to when I was 17 years old and decided to decline joining the Air Force (after significant consternation, it wasn't an easy decision for me at the time. I also think my father wasn't thrilled by it, although he obviously encouraged regardless).
October 21, 2025 at 9:24 PM
A random form momentarily took me back to when I was 17 years old and decided to decline joining the Air Force (after significant consternation, it wasn't an easy decision for me at the time. I also think my father wasn't thrilled by it, although he obviously encouraged regardless).
I have a picture of Gyárfás. This photo represents 108 Erdős papers in one frame.
October 19, 2025 at 3:19 PM
I have a picture of Gyárfás. This photo represents 108 Erdős papers in one frame.
They were able to locate solutions for 10 problems. Curious which patterns it latched on to. I guess solutions to some others could be located by a model that is good at finding "analogies to analogies" in mathematical terms, but that would perhaps require significant advances.
October 18, 2025 at 5:17 AM
They were able to locate solutions for 10 problems. Curious which patterns it latched on to. I guess solutions to some others could be located by a model that is good at finding "analogies to analogies" in mathematical terms, but that would perhaps require significant advances.
Going through some Erdős problems that Mark Sellke and Mehtaab Sawhney found had already been solved using GPT5 pro for search. More precisely, these were listed as open, but solutions to them already existed in the literature, but not recognized as such. Quite remarkable.
One example: [+]
One example: [+]
October 18, 2025 at 5:17 AM
Going through some Erdős problems that Mark Sellke and Mehtaab Sawhney found had already been solved using GPT5 pro for search. More precisely, these were listed as open, but solutions to them already existed in the literature, but not recognized as such. Quite remarkable.
One example: [+]
One example: [+]