Linux User #40306. Única persona que sigue usando &c. para abreviar etcetera.
Zaragoza, España
Joined November 2007
- Tweets111K
- Following555
- Followers1.4K
- Likes137K
Pinned Tweet
Since 2011, we have run a thread on the failure of String Theory. Never read by any of the contendants, it seems.
physicsforums.com/threads/th…
Alejandro Rivero retweeted
"My priority is to maximize my profits and obliterate my competition"
- Gemini 4 Carbon
And so it did.
Google is #1 on Vending-Bench 2 for the first time in a year.
Alejandro Rivero retweeted
Norm multiplicity of octonions.
the key identity used to disprove ETH is from a guy from the 18th century named Carl Degen
![From the ETH proof. Lemma (Eight-square identity [CS03, AM99]). There is a function σ : {0, 1}3 × {0, 1}3 → {−1, 1} such that for all x, y, u, v ∈ {0, 1}3, with ⊕ the bitwise exclusive or, (12) 2 [x = v] [y = u] = [x ⊕ y ⊕ u ⊕ v = 0] (σ(x, y) σ(v, u) + σ(x, u) σ(v, y)). Proof. Let σ(p, q) = (−1)ϕ(p,q) with ϕ(p, q) = ∑ k≤k′ pkqk′ + q1p2p3 + p1q2p3 + p1p2q3. Equa- tion (12) is a statement about 12 bits, and it can be checked at all 4096 settings. (§4 and Appendix F prove it by hand.) ■ The identity is two centuries old: Degen found it in 1818, and Graves found it again in 1843 by showing that the norm of the octonions is multiplicative [Bae02, Section 1].](https://pbs.twimg.com/media/HVvWXQXWIAA-hLM.jpg?name=orig)
ALT From the ETH proof. Lemma (Eight-square identity [CS03, AM99]). There is a function σ : {0, 1}3 × {0, 1}3 → {−1, 1} such that for all x, y, u, v ∈ {0, 1}3, with ⊕ the bitwise exclusive or, (12) 2 [x = v] [y = u] = [x ⊕ y ⊕ u ⊕ v = 0] (σ(x, y) σ(v, u) + σ(x, u) σ(v, y)). Proof. Let σ(p, q) = (−1)ϕ(p,q) with ϕ(p, q) = ∑ k≤k′ pkqk′ + q1p2p3 + p1q2p3 + p1p2q3. Equa- tion (12) is a statement about 12 bits, and it can be checked at all 4096 settings. (§4 and Appendix F prove it by hand.) ■ The identity is two centuries old: Degen found it in 1818, and Graves found it again in 1843 by showing that the norm of the octonions is multiplicative [Bae02, Section 1].
Alejandro Rivero retweeted
You can check out the talk given by David Gross when he delivered the, yes, Oskar Klein Memorial Lecture.
Good morning, all.3/3
Alejandro Rivero retweeted
1/4
Meet Qwenputer: Qwen3.5-4B with a full RISC-V CPU inside its weights, using the model's own Gated DeltaNet layers as RAM.
It runs C and MicroPython in its forward passes, no external runtime.
On DeepMind math: base 55–90%, base + Python tool 95–100%, Qwenputer 97.5–100%.
Alejandro Rivero retweeted
Of course the labs are trying Goldbach, Collatz, ABC, P/NP, and all of them. They're spending millions of dollars on them.
The fact that they haven't announced anything *is* evidence.
Alejandro Rivero retweeted
恭喜 Gemini 4 Carbon 成为全球最强模型！！
Pro 用户：不能用。
Ultra 用户：不能用。
API 开发者：不能用。
Artificial Analysis：全球第一。
Google：感谢大家观看。
我：🤬
Alejandro Rivero retweeted
🔴 ¡NUEVO GEMINI 4 CARBON!
Google golpea por segunda vez este mes, tras la salida de su modelo Argon, con una nueva versión entendemos más grande y sobretodo, potente. Superando en muchos benchmarks a Opus 5.5 y a GPT-6 Astra!
Por ahora es el anuncio del modelo, sin acceso.
Alejandro Rivero retweeted
Se lo pedí a un señor de la cola, súper amable. Le firmo la autorización en el momento y nos presentamos los dos en el mostrador: yo con la foto de mi DNI, él con la autorización y su DNI físico, para recoger mi notificación. Surrealista el sistema de los cojones.
Alejandro Rivero retweeted
Bieeeen, tras unos meses donde el Artificial Analysis Intelligence Index (agregado de benchmarks) no parecía reflejar correctamente las capacidades de los modelos, hoy han actualizado a una nueva versión v5, además con la mayoría de datasets privados para evitar benchmaxxing 👍
Announcing Artificial Analysis Intelligence Index v5, featuring 2 new benchmarks and 3 upgrades. For the first time, the majority of the Index is built on private datasets
The Artificial Analysis Intelligence Index is our composite measure of model capabilities across tasks reflecting real-world usage today. v5 is our biggest upgrade since v4 launched in January. With v5, 60% of the Index is now based on evaluations with private test sets or held-out answers.
Alongside Intelligence Index v5, we’re launching the Artificial Analysis Progress Index v1, which tracks model capability over time on one continuous scale.
New to the Intelligence Index:
➤ FrontierCode 2.0: Cognition's newly-launched private benchmark of real-world coding tasks, graded on whether a repository maintainer would merge the agent's change
➤ Terminal-Bench-Science: 70 agentic scientific research tasks, run in a terminal environment
Upgraded:
➤ HLE-Diamond (replacing Humanity's Last Exam): a refined 1,000-question version of Humanity's Last Exam, split evenly between reasoning and knowledge questions
➤ AA-Omniscience v1.1: An updated dataset with saturated and public set questions removed, new questions added, and ambiguous or erroneous questions revised
➤ CritPt v2: a private, 40-question hard subset of CritPt v2, revised by the CritPt authors and run with a code-execution tool
Staying in the Index: AA-Briefcase, AutomationBench-AA, GDP.pdf and Terminal-Bench 4.0
Removed from the Index: GDPval-AA v2.1, SciCode, and AA-LCR v1.1
Key results:
➤ Claude Opus 5.5 (max) leads at 58, ahead of GPT-6 Astra (max, 57) and GPT-6.1 Sol (max, 55): Anthropic and OpenAI hold the top five models. Gemini 4 Argon (high) is next at 50
➤ The open weights gap widens from 3 points to 12: Kimi K3.1 Pro (weights not yet released) is the top open weights model at 46. Among released weights, DeepSeek V4.1 Flash 1017 (max) and GLM-5.3 (max) lead at 32 - 26 points behind the leader and below Claude Haiku 5.5 (max, 36)
➤ Cost per task varies by more than 13x among the leading models: of the 10 highest-scoring models shown, GPT-6.1 Terra (max) is the cheapest at $0.96 per Intelligence Index task and Claude Fable 5.1 (max) the most expensive at $13.04
Alejandro Rivero retweeted
Statement of Yang-Mills Working Group, a scientific collaboration to interpret AI result related to the existence of quantum chromodynamics & completions of the Standard Model, including a subproblem proving the existence of the Quantum Yang-Mills theory. buff.ly/BwV89Tk
Alejandro Rivero retweeted
computer use with terra has crossed a qualitative threshold where it feels faster at using a computer than what i can do unassisted
Alejandro Rivero retweeted
My dream tonight:
physicists assisted by AI (have) adapt(ed) Rivasseau arxiv.org/abs/0706.1224 work on the Grosse-Wulkenhaar matrix model arxiv.org/abs/hep-th/0401128 to function like Balaban renormalization group method for the #SpectralStandardModel
perplexity.ai/search/2fae8cd…
I wonder if #OpenAI could define an exact block-spin transfo that acts on the single universal matrix package of the Yang-Mills-Higgs sector of the #SpectralStandardModel devised by Chamseddine, Iliopoulos & van Suijlekom to complete a constructive renormalisation program in 4D.
Alejandro Rivero retweeted
I am obviously not a QFT/hep person, but this blog post (+ comments) will probably be of interest to the broader community.
New Post: Secret Unified Physics Slop
math.columbia.edu/~woit/word…
Alejandro Rivero retweeted
The Yang-Mills formalization took one month!
Fermat's Last Theorem was 11 days and the one of Navier-Stokes 17 hours.
What is coming there 🤯?
Alejandro Rivero retweeted
UC Berkeley mathematics professor Richard Borcherds stops grading homework for correctness, declaring there is “no point” as student use of AI becomes “widespread.”
Students now receive credit simply for submitting their assignments.
Follow: @AFpost
Alejandro Rivero retweeted
Por parte de las hermanas Serrano no veo especulación alguna, sino amor por la obra de su familia. A cambio de no dejarla morir, la cedieron y alquilaron el local por menos de lo que podrían haber obtenido. Lo que realmente han recibido ha sido...
ojo al estilo de gráficos de @cognition – at Saragossa, Spain
Alejandro Rivero retweeted
Esto se parece más a mi experiencia, está mucho mejor calibrado
Alejandro Rivero retweeted
"En lo que llevamos hablando han salido 3 modelos nuevos" #TRG26