mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Miscellaneous Math (https://www.mersenneforum.org/forumdisplay.php?f=56)
-   -   Checking of Collatz problem / conjecture (https://www.mersenneforum.org/showthread.php?t=24661)

 Dylan14 2019-09-08 13:17

1 Attachment(s)
From the attached image, the client was able to connect successfully to the server, download some workunits and start working on them. It hasn't finished them yet, so I don't know about the uploading of the results.

 dabler 2019-09-09 06:14

As I look into the server log, I see that these assignments were successfully returned in approximately 2 hours and 12 minutes. That is perfect, thank you!

 R. Gerbicz 2019-09-13 09:44

Almost solved(!), Terence Tao has made a big progress on Collatz conjecture: [url]https://arxiv.org/pdf/1909.03562.pdf[/url]

 retina 2019-09-13 10:05

[QUOTE=R. Gerbicz;525767]Almost solved(!), Terence Tao has made a big progress on Collatz conjecture: [url]https://arxiv.org/pdf/1909.03562.pdf[/url][/QUOTE]Tao's page:
[url]https://terrytao.wordpress.com/2019/09/10/almost-all-collatz-orbits-attain-almost-bounded-values/[/url]

 Dr Sardonicus 2019-09-13 12:12

[QUOTE=retina;525768]
[QUOTE=R. Gerbicz;525767]Almost solved(!), Terence Tao has made a big progress on Collatz conjecture: [url]https://arxiv.org/pdf/1909.03562.pdf[/url][/QUOTE]
Tao's page:
[url]https://terrytao.wordpress.com/2019/09/10/almost-all-collatz-orbits-attain-almost-bounded-values/[/url][/QUOTE]In the arxiv preprint, Definition 1.2 has

$$\text{for all }A\;\subset\;S$$

which threw me for a loop, since "S" was not defined.

It was a typo. Tao's page (thanks, [b]retina[/b]!) says it should be

$$\text{for all }A\;\subset\;\mathbb{N}\;+\;1$$

 dabler 2019-09-16 15:52

Any help still welcome

Any help with this distributed computation is still welcome. All you need to do is to compile the client from [URL="https://github.com/xbarin02/collatz"]this GitHub repository[/URL]. The source codes require a GNU toolchain. They can also be compiled on MS Windows (see above discussion).

 R. Gerbicz 2019-09-16 16:51

[QUOTE=dabler;525936]Any help with this distributed computation is still welcome.[/QUOTE]

I could be boring, but this type of task is ideal for the much faster gpu.

 dabler 2019-09-17 11:28

I'm sure you're right about this. However, now I don't known any person able to implement it on the GPU right now.

PS: Linux binaries are also available on the project page.

 dabler 2019-10-14 06:28

My project has evolved a lot in the last month. Now I use modern (massively parallel) GPUs to verify the problem convergence. This brings an acceleration of more than two orders of magnitude. At this moment, I would welcome any help regarding the GPU implementation. Could anyone [URL="https://codereview.stackexchange.com/questions/230606/computational-verification-of-collatz-conjecture-using-opencl"]review my code in OpenCL on StackExchange[/URL]?

 dabler 2020-07-02 07:51

After less than a year, I managed to verify all numbers below 2[SUP]68[/SUP] in this way. For those who are interested, the results are presented in [URL="https://rdcu.be/b5nn1"]this article[/URL].

 moebius 2020-07-02 09:29

Poll:
Which of the previous avatars looks most like a mugshot from sheriff's department?

All times are UTC. The time now is 12:29.