3 ms·
Generating conjectures on fundamental constants with the Ramanujan Machine
- dj_mc_merlin 6y agoSad to see this on the front page of HN. Their marketing campaign is working wonders. The Ramanujan Machine is fraud: https://www.galoisrepresentations.com/2019/07/17/the-ramanujan-machine-is-an-intellectual-fraud/ https://www.galoisrepresentations.com/2019/07/17/the-ramanuj...
- l-lousy 6y agoWould the new discovery for catalans number be considered a counter to: > nothing considered remotely new by an expert Or is the proof mentioned at the end of the article the proof for Catalans number (sorry, I’m a non mathematician)
- spekcular 6y agoIt seems they don't actually consult experts when saying such things. IIRC, they've made claims of novelty like this before and then been wrong because they didn't do their homework. If it truly is a new identity (and not a specialization of something already known), then that's kind of interesting. But given their track record, I find it unlikely. One much always be very careful when claiming some identity like this is new. Even excellent researchers can get burned – it has happened to Doron Zeilberger, one of the pioneers of algorithmic mathematics, at least one. (I'm looking for the details but unfortunately can't find them at the moment...) edit: The issue in the latter case may have been a very simple analytic proof existing, not novelty per se. He had posted correspondence between him and another mathematician on his website but apparently deleted it. Also, the blog post linked above has been updated: https://www.galoisrepresentations.com/2021/02/11/ramanujan-machine-redux/ https://www.galoisrepresentations.com/2021/02/11/ramanujan-m...
- Brajeshwar 6y agoI realized the source of the original paper[1] is available for a price. The dedicated website[2] has more details, including the source code[3]. 1. https://www.nature.com/articles/s41586-021-03229-4 https://www.nature.com/articles/s41586-021-03229-4 2. http://www.ramanujanmachine.com http://www.ramanujanmachine.com 3. https://github.com/ShaharGottlieb/MasseyRamanujan/ https://github.com/ShaharGottlieb/MasseyRamanujan/
- sva_ 6y agoSee also the recent discussion of another article about the paper[0]. I'm also predicting this thread will have a meta-discussion about whether mathematics is invented or discovered. [0] https://news.ycombinator.com/item?id=26044640 https://news.ycombinator.com/item?id=26044640
- jacknews 6y agohttps://www.viduba.com/video:QdVT4ZlaVdnVWplcaZkUpRWRZl3Vn1TP https://www.viduba.com/video:QdVT4ZlaVdnVWplcaZkUpRWRZl3Vn1T...
- Radim 6y agoThis whole "DISRUPTION AT 8 O'CLOCK!" economy seems out of whack. Bombastic headlines, vacuous promises… it's been the optimal strategy in business and media for a while, but now infecting the sciences apparently. I wonder how long honest players can survive in this climate. AI Winter, where are you?
- fxj 6y agoTheir paper is also free to read on arXiv: https://arxiv.org/abs/1907.00205 https://arxiv.org/abs/1907.00205 "We present two algorithms that proved useful in finding conjectures: a Meet-In-The-Middle (MITM) algorithm and a Gradient Descent (GD) tailored to the recurrent structure of continued fractions. Both algorithms are based on matching numerical values and thus they conjecture formulas without providing proofs and without requiring prior knowledge on any underlying mathematical structure. This approach is especially attractive for constants for which no mathematical structure is known, as it reverses the conventional approach of sequential logic in formal proofs. Instead, our work supports a different approach for research: algorithms utilizing numerical data to unveil mathematical structures, thus trying to play the role of intuition of great mathematicians of the past, providing leads to new mathematical research." For a live version on which you can comment and discuss with your peers and other readers, see this link on SciHive: https://www.scihive.org/paper/1907.00205 https://www.scihive.org/paper/1907.00205