guiherme (@d4mmit) header
guiherme (@d4mmit) profile picture

tech lead, usuário de claudinho e fã de rock triste animado

Natal, Brasil
Joined November 2010
Photo by guiherme (@d4mmit)Photo by guiherme (@d4mmit)Photo by guiherme (@d4mmit)Photo by guiherme (@d4mmit)Photo by guiherme (@d4mmit)Photo by guiherme (@d4mmit)Photo by guiherme (@d4mmit)Photo by guiherme (@d4mmit)Photo by guiherme (@d4mmit)Photo by guiherme (@d4mmit)
guiherme retweeted
>ai engineer >tells ai to do engineering >it does >go home
15
565
30
10,060
104,695
- me contratan como desarrolladora - desarrollo depresión - me voy
12
609
20
6,604
42,894
ARRASCAETA > BRUNO HENRIQUE (GOAT) > SAMUEL LINO > GOL GERAÇÃO 2019 ETERNA
24
823
12
5,733
41,373
guiherme retweeted
WTF is wrong with this timeline... bruh?? 😂🔥 First we got Reverse Engineer Anything... then we got Craft Anything... then we got Rust Anything... and now we got Fast Anything. 😭 This one seems cool though... Instead of rewriting your entire app in Rust, it helps you rewrite only the parts that are actually slow. ~ Find performance bottlenecks. ~ Rewrite slow parts in Rust. ~ Verify the new code. ~ Benchmark speed improvements. ~ Keep the rest of your app untouched. Basically, don't rewrite everything. Just make the slow stuff FAST. I wonder what comes next. 💀 Follow me for more amazing AI, Coding & Web Dev insights 💎
11
51
1
681
17,239
In these Riemann-hypothesis-fueled uncertainty, the physicist in me is for the first time hopeful that, in my lifetime, I might see a reconciliation of quantum mechanics and general relativity.
63
87
16
1,350
31,427
guiherme retweeted
Independent mathematicians have now finished the full verification of the proof of the Riemann hypothesis announced yesterday by Anthropic mathematician Levent Alpöge. The results are real, they say. scientificamerican.com/artic…
63
524
99
2,449
175,768
🚨 UPDATE sur Riemann et elle est énorme Hier, Levent Alpöge, mathématicien chez Anthropic, annonçait avoir trouvé une preuve avec Claude. C'était une revendication, personne n'avait encore vérifié Aujourd'hui, la Lean FRO et la Mathlib Initiative, les deux organisations derrière Lean et sa bibliothèque de maths, publient une déclaration commune : « We hereby certify that both the statement and the proof of the Riemann Hypothesis are correct » EN 24 HEURES !! Elles écrivent avoir « full confidence » dans la preuve et félicitent Alpöge Après 167 ans cest le plus gros morceau des maths qui tombe, alors que tout le monde le voyait encore résister 30-50 ans. 1. Ce qui a changé en 24h Hier je vous disais que personne hors de son cercle n'avait confirmé que le code compile, ni que le théorème final dit exactement ce que dit l'hypothèse de Riemann Les deux points sont réglés. Des dizaines de personnes ont recompilé la bibliothèque de leur côté et obtenu le même fichier de preuve, sans aucune erreur Et l'énoncé démontré est bien celui de Mathlib, qui avait déjà été relu par des experts de Lean et de théorie des nombres 2. Comment ils ont vérifié La preuve fait 17 millions de lignes de Lean et ne dépend directement que de Mathlib La communauté a isolé l'énoncé de Riemann dans un fichier à part, qui n'importe que Mathlib, puis l'a comparé au théorème d'Alpöge avec deux outils indépendants, comparator et con-leche. Les deux confirment que c'est exactement le même énoncé La preuve a ensuite été passée dans plusieurs vérificateurs, dont le noyau officiel de Lean, con-leche et con-ron, plus d'autres écrits par des tiers dans d'autres langages Détail très fort : con-leche et con-ron ont eux-mêmes une preuve formelle qu'ils ne peuvent pas accepter une fausse preuve. Ça répond à la crainte d'une faille dans la logique de Lean que la preuve aurait pu exploiter 3. Claude des deux côtés Andrew Sutherland, du MIT, publie tous ses fichiers de vérification. Et ses contrôles ont été lancés avec Claude Code Sur sa page : les 175 187 modules compilent, la preuve n'utilise que les trois axiomes standard de Lean, et nanoda, un noyau indépendant écrit en Rust, accepte la preuve complète Donc Claude a aidé à trouver la preuve, et Claude a aidé à la vérifier. Avec des outils indépendants à chaque étape. Pour le million du Clay Institute, les règles demandent une publication dans une revue reconnue, deux ans d'attente puis une acceptation générale par la communauté. Et Anthropic n'a fait aucune communication officielle 5. Ce que ça débloque Comme je vous disait hier Des milliers de théorèmes commençaient par « si l'hypothèse de Riemann est vraie, alors… ». Ils tiennent maintenant sur une preuve vérifiée par machine Pour les gros needs, Sutherland l'a déjà montré : en combinant la preuve avec une formalisation indépendante du critère de Robin, Lean démontre l'inégalité de Robin pour tout n > 5040, sans aucune mention de la fonction zêta Le Lean blog conclut en attendant toute l'activité de la communauté pour expliquer les idées nouvelles de cette preuve. Les humains doivent maintenant comprendre ce que Claude et Alpöge ont trouvé, une première dans l’histoire L’ère de l’exponentielle vient de commencer, l’humanité entre dans un âge d’or
The Lean FRO and Mathlib Initiative have released a joint statement on the Lean verification of the Riemann Hypothesis. On November 20, Levent Alpöge released a proof of the Riemann Hypothesis as a Lean library of 17 million lines of code. Over the past day, the Lean and Mathlib communities worked together to verify it. The statement certifies the proof as correct and describes how it meets all necessary verification criteria. Congratulations to Levent on this achievement, and thank you to everyone in the community who contributed to the verification. Full statement: leanprover-community.github.…
75
234
24
1,733
148,876
guiherme retweeted
BREAKING: The Lean community declares Levent Alpöge & Claude’s proof of the Riemann Hypothesis correct, following verification of 17 million lines of mathematical code.
211
967
233
13,315
597,936
guiherme retweeted
Cómo así que AC - DC no significa: Antes de Claude - Después de Claude?
7
388
5
2,844
44,231
guiherme retweeted
terence tao just gave his take on the AI-generated Generalized Riemann Hypothesis (GRH) proof. he ranked this week's three AI-assisted math proofs by how "digested" they are >euler blow-up: a "human-readable proof", also formally checked in Lean. >navier-stokes blow-up: the strategy is "broadly clear", but it still needs "a proper, professionally written proof". >GRH: an "(enormous) Lean-only proof". mathematicians are still working through what it actually does. he says it "appears to use the full power of modern developments on the Langlands program and arithmetic geometry" in the comments he adds it's "very early days", but the AI proof seems to combine all the main approaches mathematicians thought could crack the problem. he also looked at the separate, entirely human-written proof by alessandro masullo. he says it's closer to christopher deninger's strategy than alain connes's, though neither of those approaches is close to being fully fleshed out. and his verdict: "I imagine the digestion process will take some time"
8
24
1
330
52,804
guiherme retweeted
The Lean FRO and Mathlib Initiative have released a joint statement on the Lean verification of the Riemann Hypothesis. On November 20, Levent Alpöge released a proof of the Riemann Hypothesis as a Lean library of 17 million lines of code. Over the past day, the Lean and Mathlib communities worked together to verify it. The statement certifies the proof as correct and describes how it meets all necessary verification criteria. Congratulations to Levent on this achievement, and thank you to everyone in the community who contributed to the verification. Full statement: leanprover-community.github.…
399
4,323
1,699
23,644
2,199,725
guiherme retweeted
In 1924, Indian physicist Satyendra Nath Bose sent a four-page research paper to Albert Einstein after an earlier journal submission received no response. Einstein personally translated it into German and arranged its publication. The paper introduced the counting method behind Bose–Einstein statistics. Today, an entire class of particles is named after Bose: BOSONS.
11
47
2
267
11,910
Let’s take Riemann Hypothesis off the Millennium problem list and put cancer on it and see what happens.
55
68
9
1,652
41,375
Claude explain this proof of the Riemann Hypothesis in a 5 min 3Blue1Brown-style video don't dumb it down and make no mistakes
9
34
3
851
28,062
The only proof we have for the riemann hypothesis is in lean. Look up the nationality of the guy who created lean. Nuff said
Congratulations to the Brazilian mathematician who will solve the Riemann Hypothesis in two weeks.
8
55
4
1,059
57,553
guiherme retweeted
The Collatz Conjecture will end up being harder than all Millennium Prize Problems. The next few Busy Beaver numbers include solving Collatz-like problems. Only P vs. NP might be similarly difficult, and I think it might be easier than Collatz.
27
31
1
635
20,914
guiherme retweeted
SITUATION UPDATED: The Riemann Hypothesis proof has passed the independent kernel check. MIT's Andrew Sutherland exported Alpöge and Claude's full proof and ran it through separately written kernels. nanoda (Rust) accepted it in about 28 minutes. con-leche accepted all 1.47 million declarations and confirmed the five theorems, including RH itself, match Mathlib-only statements. lean4lean passed all 175,187 modules, and Lean's own kernel replay passed with zero rejections. A Lean bug is now essentially ruled out, since every checker would have to share it. Two slower checks are still running, and nothing has failed. What's left is the human part: no readable proof exists yet, and the Clay prize needs publication plus two years of review. The Riemann Hypothesis is proven, but not yet understood.
SITUATION DETECTED: The Riemann Hypothesis proof was verified overnight. MIT's Andrew Sutherland and others independently compiled Levent Alpöge and Claude's Lean code, about 17 million lines across 175,000 files, and the Lean kernel accepted it using only the three standard axioms. I went through the code myself: the final theorem uses Mathlib's own definition of RiemannHypothesis, nothing in the project redefines it, there's no sorry and no custom axioms. What's left is an independent kernel check to rule out a bug in Lean itself and a human-readable version of the proof.
142
742
190
9,550
485,258
JUST IN🚨: Scientists found a way to help the brain wash out waste while we're awake, which it mostly does while we sleep: a short-acting blood pressure drug plus whole-body vibration. In 29 healthy adults, Alzheimer's-linked proteins like beta-amyloid rose in the blood, a sign they were cleared from the brain.
2
6
1
50
3,669
In 1687, Sir Isaac Newton published his masterpiece, 𝘗𝘩𝘪𝘭𝘰𝘴𝘰𝘱𝘩𝘪æ 𝘕𝘢𝘵𝘶𝘳𝘢𝘭𝘪𝘴 𝘗𝘳𝘪𝘯𝘤𝘪𝘱𝘪𝘢 𝘔𝘢𝘵𝘩𝘦𝘮𝘢𝘵𝘪𝘤𝘢 ("𝘔𝘢𝘵𝘩𝘦𝘮𝘢𝘵𝘪𝘤𝘢𝘭 𝘗𝘳𝘪𝘯𝘤𝘪𝘱𝘭𝘦𝘴 𝘰𝘧 𝘕𝘢𝘵𝘶𝘳𝘢𝘭 𝘗𝘩𝘪𝘭𝘰𝘴𝘰𝘱𝘩𝘺").
6
28
1
97
11,009
guiherme retweeted
If the Generalized Riemann Hypothesis (the stronger version of the Riemann Hypothesis) is proven, these become proven facts: Algorithms that no longer need randomness - Primality testing: Miller's fast test becomes proven correct with no randomness - Square roots mod a prime: becomes solvable fast and with no coin flips (today that's an open problem) - Cube roots and other roots mod a prime: same. - Factoring polynomials over finite fields: almost fast (quasi-polynomial) with no randomness, instead of needing randomness - Building finite fields: you can construct any finite field directly, with no random search - Factoring vs. φ(n): factoring a number becomes provably as easy as computing φ(n), even without randomness - Primes with a given remainder: a prime with any allowed remainder mod q exists that's at most about q², so searching for one is fast Complexity theory - Polynomial equations: "does this system of equations have a solution?" drops from PSPACE to a much lower class (AM) - Algebraic P vs NP: Valiant's algebraic version of P vs NP gets tied to the normal P vs NP over the complex numbers - Turing machines: a known 744-state Turing machine is proven to run forever Number theory software - Math libraries: PARI/GP and Magma results computed "assuming GRH" become fully proven - Class groups: algorithms that compute these number-field structures get guaranteed running times - Elliptic curve generation: the algorithms used to create curves for crypto get guaranteed running times Cryptography (no new attacks, only proofs) - Elliptic curves: proven that if discrete log is hard on one curve, it's hard on almost all similar curves - Lattice crypto: known quantum attacks on some lattice schemes become proven to work - Isogeny crypto: known quantum attacks on some isogeny schemes get proven speeds - Code-based crypto: schemes like BIKE get the guarantee that the special primes they need exist Prime gaps - Next prime: there's always one within about √p·log p of p, so "find the next prime" has a guaranteed worst case What it does NOT do - It doesn't solve P vs NP - It doesn't make factoring easy, so RSA and elliptic-curve crypto stay safe - It doesn't make anything faster in practice, because randomized versions already work today - Ramanujan graphs, expanders and AG codes are unaffected: they use a finite-field version of RH that was proven decades ago
75
571
55
4,506
439,194