j_bertolotti

Associated Professor of Physics at the University of Exeter.
Maker of typos.
Scientific visualizations (grouped under the hastag #PhysicsFactlet).
He/lui/on. All opinions are my own fault.

Public Key

npub1a4ywuu668ac6558wk53ng6nsykluuag9zpwm64nhgmjpeucc5n6skrazsc

NIP-05 Address

j_bertolotti@mathstodon-xyz.mostr.pub

Profile Code

nprofile1qqsw6j8wwddr7ud22rht2ge5dfczt07ww5z3qhda2em5dequ7vv2fagpz3mhxue69uhhyetvv9ujuerpd46hxtnfduqs6amnwvaz7tmwdaejumr0dspp4avr

Publishing to

relay.damus.io

nos.lol

atlas.nostr.land

Author Public Key

npub1a4ywuu668ac6558wk53ng6nsykluuag9zpwm64nhgmjpeucc5n6skrazsc

Show more details

Last Notes

2026-04-09 10:55:24 UTC

- reply

@nprofile…np45 Just think about all the implied knowledge in any cookbook. Recipes can be compressed to a page or two only because they assume you already know how to do a lot of the stuff they are talking about. If you have no idea how to cook, no amount of cookbooks will ever help you.
@nprofile…jnc7 @nprofile…ugpy

2026-03-03 11:56:47 UTC

I see a lot of people posting about the Greens overcoming Labour in the polls, but everything I see in those polls is Reform being the first party.
And that is f***ing scary.

2025-12-26 18:36:04 UTC

- reply

@nprofile…8eua There are infinite cool things to learn and a finite time to learn them. You have to prioritise 😉

2025-12-26 12:31:18 UTC

- reply

@nprofile…8eua I didn't mean to offend you. If I did, I apologize.
I understand this stuff on an intellectual level, but I never devoted nearly enough time to music to make this stuff less confusing at an intuitive level. There is just too much historical baggage.

2025-12-26 12:16:53 UTC

- reply

@nprofile…8eua The very fact that people decided to divide notes into "white" and "black" instead of simply say that they are ultimately the same thing is also something that will confuse me forever 🤣

2025-12-18 11:36:55 UTC

- reply

@nprofile…8eua After not touching it for >1y, I used this occasion to deactivate it (it will take 30 days before it is deleted).

2025-11-27 07:57:27 UTC

- reply

@nprofile…pklh If I split it into 3 parts, can I then recombine them to obtain 2 spheres identical to the original?

2025-11-05 11:50:11 UTC

- reply

@nprofile…tqek Danno assolutamente per scontato che sia cattolico. L'ipotesi che un bianco possa seguire una qualunque altra religione (o, orrore degli orrori, essere ateo) non li sfiora neppure
@nprofile…tzrf

2025-11-04 10:35:16 UTC

- reply

@nprofile…8eua I wonder if Meitner ever crossed paths with Noether. Both Jews women in Germany struggling to have their genius recognized by a close-minded University system. Both having to flee Germany. Roughly the same age.
They were interested in very different problems, but I can only imagine what they could have achieved if they ever worked together!

2025-10-28 14:09:37 UTC

ChatGPT costs more money and resources to run than Wikipedia. So if the answers from ChatGPT are not at least as good as the ones you can find just visiting the correct Wikipedia page, the rational move is to give all of OpenAI money to the Wikimedia foundation.

2025-10-23 19:48:06 UTC

Today I happened to have to walk for a while with a young PhD student, and while doing smalltalk, I found myself telling him about the history of home computers, the internet and videogames before he was born.
I am old 👴🏻
ps
He explicitly asked, so it doesn't count as psicological torture 🤣

2025-10-18 10:56:35 UTC

- reply

@nprofile…4stm Knitting is the kind of activity that disturbs nobody, so it would be very petty to forbid it 🙂

2025-10-18 07:40:14 UTC

There is a girl who started brining her knitting to my class, and now I wish I could spend a few lectures on knot theory and (some of) the beautiful Maths behind knitting 🧶

2025-10-15 08:45:29 UTC

- reply

@nprofile…ah9w "For instance, someone might embed such triggers in specific websites to make models unusable when they retrieve content from those sites."
I now need instructions on how to include the poisoned text in my website so it is invisible to a human reader (e.g. as tiny white text? Would that work?)

2025-10-13 10:49:52 UTC

#PhysicsFactlet
Sometimes you need to forgo the intuitive way to define stuff for the sake of actually being able to do anything useful with those definitions.
An example of this I always found funny is in knot theory, where a simple loop is considered to be a knot, while anything where the two extremes are dangling are not, including the common overhand knot.
This looks weird the first time you see it, but there is a very good reason to go with such a definition: you want to study what you can and can't do by manipulating the knot, and if you have the two extremes dangling, you can always untie any knot, making them all equivalent to a piece of string. In order to be able to say anything interesting about them you need to remove this trivial option, and thus accept the simple loop as a knot.
#knots
https://media.mathstodon.xyz/media_attachments/files/115/366/428/534/174/821/original/0c89383034a0cc2c.png

2025-10-03 11:13:44 UTC

#PhysicsFactlet
A geodesics on a surface is the shortest* curve connecting two points.
*or longest, if you want to be pedantic 😉
https://media.mathstodon.xyz/media_attachments/files/115/309/934/107/591/247/original/ed06bc39d666a81d.mp4

2025-08-21 08:33:59 UTC

- reply

@nprofile…rwjw Abbiamo (più o meno) la stessa età e leggere il suo CV non aiuta la mia autostima 🤣

2025-08-08 13:38:20 UTC

Something I often wonder but I have no way to easily check: how do #LLM fare with languages that have a smaller number of speakers (i.e. for which the training dataset is much smaller)?
E.g. if you ask ChatGPT a question in Basque, how well does it work?
(I only know that, to this day, Google Translate is hilariously bad at translating from English to Italian)

2025-07-14 09:15:33 UTC

- reply

@nprofile…uxf3 Point 2 is VERY important, because it is both a major component of why 25 degrees are lethal instead of a just a minor annoyance, and it is something that can't be changed on a reasonable timescale, even if there was a political will (and there isn't one).

2025-06-06 13:39:30 UTC

- reply

Look mom, I got a semi-viral post on Mastodon!
I don't have a soundcloud (or whatever people who regularly go viral tend to advertise), so just be nice to each other!

2025-06-06 08:50:06 UTC

"AI companies claim their tools couldn't exist without training on copyrighted material. It turns out, they could — it's just really hard. To prove it, AI researchers trained a new model that's less powerful but much more ethical. That's because the LLM's dataset uses only public domain and openly licensed material."
tl;dr: If you use public domain data (i.e. you don't steal from authors and creators) you can train a LLM just as good as what was cutting edge a couple of years ago. What makes it difficult is curating the data, but once the data has been curated once, in principle everyone can use it without having to go through the painful part.
So the whole "we have to violate copyright and steal intellectual property" is (as everybody already knew) total BS.
https://www.engadget.com/ai/it-turns-out-you-can-train-ai-models-without-copyrighted-material-174016619.html?src=rss

2025-05-30 23:50:09 UTC

- reply

@nprofile…m2ru @nprofile…f7v6
It depends a lot on exactly what you want to do, but to create a simple focus like in this visualization you can get surprisingly good result with very poor precision, as long as you are stable and you can control a large number of degrees of freedom.

2025-05-30 10:37:43 UTC

#PhysicsFactlet
Scattering scrambles coherent light into a speckle pattern, where the field at each point can be seen as the superposition of a large number of random phasors. At some point the result is brighter, and at some points the result is dimmer, creating the "speckly" pattern.
By changing the phase of the incident light one can change the phase of the phasors making up the resulting field, and since elastic scattering is linear, changing the phase of different input modes is going to rotate different phasors without cross-talk.
As a result it is possible to find an incident wavefront such that all the phasors making up the field at one point are in a straight line (constructive interference), resulting in a single bright dot (a focus) through a completely scattering material.
#Optics #Physics
https://media.mathstodon.xyz/media_attachments/files/114/596/310/441/654/962/original/0241c23b0d5bf003.mp4

2025-05-06 13:35:42 UTC

- reply

@nprofile…fa4k "but whenever I try to explain it online, a bunch of men get offended" which is not independent from the reason why it is predominantly men 😉
@nprofile…s3kl

2025-03-12 07:39:43 UTC

- reply

@nprofile…fa4k @nprofile…m7f4 Anything bigger than a standard Russel's teapot is a moon 🙃

2025-03-02 17:14:12 UTC

- reply

@nprofile…e5g0 The sad part is that she used to be very reasonable (although I never appreciated her use of "Physics" to mean "the tiny sub-sub-fieldnI used to work in")

2025-02-28 15:45:36 UTC

- reply

@nprofile…e5g0
Blocked and reported.
@nprofile…59k6

2025-02-27 16:01:20 UTC

- reply

@nprofile…3ug8 Immagino che tu conosca già il "password game", ma giusto nel caso te lo metto qui che magari ti serve per fare allenamento 😉
https://neal.fun/password-game/

2025-02-27 09:36:31 UTC

I would expect all my fellow scientists to be able to spot a dishonest chart from 10 miles away, but it is still good to get a quick refresher from time to time.
Also a good resource to give to students, who are often very naive on these things.
https://flowingdata.com/projects/dishonest-charts/

2025-02-12 09:30:17 UTC

- reply

@nprofile…9374
London -> Paris -> Milan -> Venice requires 3 changes, but each individual leg of the journey works quite well.
@nprofile…c5n8

2025-02-05 09:51:38 UTC

#PhysicsJournalClub
"The Tumultuous Birth of Quantum Mechanics"
by Philip Ball
https://physics.aps.org/articles/v18/24
"quantum mechanics wasn’t created all at once. It took several decades and was a messy, confused process, during most of which the true nature of this revolution was obscure. In some ways it still is."
#QuantumMechanics #HistoryOfScience #Physics

2025-01-24 11:10:38 UTC

- reply

@nprofile…yu7x
https://media.mathstodon.xyz/media_attachments/files/113/883/018/208/883/008/original/80b43cc30b979be6.png

2023-08-03 17:50:28 UTC

- reply

@npub1znz…8l64 Super-naïve question: does wishlisting on Steam helps the developers? (I technically have an account on Steam, but I have never used it for anything)

2023-07-20 12:41:49 UTC

- reply

@npub1knz…c73p Well, not that we have one unit for the x-axis and one (different) unit for the y-axis, I guess what we really need is a unit for the z-axis.