Cameron Freer

Research Scientist at the @MIT @ProbCompProj studying interactions of randomness and computation

[Cameron Freer @cameronfreer]

The world is on the cusp of abundant autoformalization. This will empower anyone to verify their mathematical ideas more quickly and engage in co-discovery with computers backed by a large, growing foundation of formalized mathematics.

[Anirudh Pai @ani_pai]

Excited to share that we’re doubling down on @mathematics_inc.

AI is generating more code than ever, but correctness isn’t guaranteed and failures can be catastrophic.

Their agent Gauss delivers a missing primitive for AI: provably correct code and knowledge.

Thrilled to keep working with @jessemhan, @jdlichtman, and the @mathematics_inc team!

[Sidharth Hariharan @SidharthHariha1]

It gives the maintainers of the Sphere Packing Project great pleasure to announce a major milestone in the formalization effort: we now have a sorry-free Lean proof of the theorem that the optimal sphere packing in ℝ⁸ is the E₈ lattice packing. (1/9)

[Cameron Freer @cameronfreer]

@npparikh Others probably know more, but my sense is that some ingredients are there, but not everything -- see, e.g., https://t.co/1OjgbAkssT

[Cameron Freer @cameronfreer]

@npparikh I haven't worked with either as directly, and my sense is that it still depends a lot on the specific thing you're trying to show -- though linear algebra is probably the most mature of these in general.

[Cameron Freer @cameronfreer]

Currently coordinating with mathlib to upstream parts of the formalization.

Discussion: https://t.co/0xDbSYrHP1

(3/3)

[Cameron Freer @cameronfreer]

I'm excited to share de Finetti's theorem formalized in Lean: three different proofs that infinite exchangeable sequences are conditionally IID.

43k lines across 113 files with 600+ lemmas.

💻 code: https://t.co/thYmw3ANDj

📊 blueprint: https://t.co/GWX9vc9hCT

(1/3)

[Cameron Freer @cameronfreer]

All three proofs from Kallenberg (2005), each filling different mathlib gaps:
• Reverse martingales
• Mean ergodic theorem machinery
• Elementary L² bounds

Built over 3 months with Claude + GPT with my Skill for Lean 4: https://t.co/TMYbI7zQUV

(2/3)

[Quanta Magazine @QuantaMagazine]

Computer scientists often look to natural phenomena like the cosmic background radiation in order to obtain truly random sequences — but randomness can only take you so far. Many real-world situations depend on weighted outcomes. https://t.co/WHl0wiBLcq

[pressetext.com @pressetext]

Neuer #Algorithmus wirft effizient gezinkte Würfel. Klimaforschung, Epidemiologie, Finanzen:... https://t.co/z209OTJepX

[Cameron Freer @cameronfreer]

Our work on near-optimal exact samplers, to appear at #AISTATS2020: https://t.co/oj65vMioNA

[MIT School of Engineering @MITEngineering]

The fast and efficient generation of random numbers has long been an important challenge. For centuries, games of chance have relied on the roll of a die, the flip of a coin, or the shuffling of cards to bring some randomness into the proceedings @MITEECS https://t.co/kFenrsLBQ6

[Andy Gordon @AndrewDGordon]

One week to Oct 18 deadline for your 2-pager on Probabilistic Programming to https://t.co/STA3Mq1dlD part of #POPL2018

[WIRED @WIRED]

How life (and death) spring from disorder https://t.co/CPD6bYXoFG

[Quanta Magazine @QuantaMagazine]

Living organisms feed on "negative entropy" by capturing and storing information, Schrödinger said: https://t.co/fc3joAhjpS

[Chung-chieh Shan 單中杰 @ccshan]

#PPS2017 workshop opens with invited talk by Gordon Plotkin #POPL2017

[Diana Cai @dianarycai]

New preprint “Priors on directed exchangeable graphs" out - On what's missing in many digraph models https://t.co/cgR4Vc7Lms @cameronfreer

[Chung-chieh Shan 單中杰 @ccshan]

I don't always observe a zero-probability event, but when I do, I specify the observed variable, not just a Boolean predicate.

[Chung-chieh Shan 單中杰 @ccshan]

Workshop on probabilistic programming semantics, colocated with POPL 2016, call for extended abstracts http://t.co/WyrHsxred4

[Cameron Freer @cameronfreer]

Verifying myself: I am freer on Keybase.io. gF3yDXiH6zBiObb05Oj1grePfoFCdmtRr3_w / https://t.co/o9KXsPiQz8

[Dan Roy @roydanroy]

Recent interview with MIRI's Luke Muehlhauser on probabilistic programming and AI. http://t.co/cQMviTqzpS