If you are comfortable with Python, you can write code that massively parallelises with numba, numpy, or cuda. The vast majority of ML, BLAS and similar libraries in Python are simple wrappers written in underlying C/C++ or CUDA. IMHO the simplicity and easy of use, as well as documentation of them, far outweighs trying to squeeze a tiny bit of performance out of them by moving entirely to C++. Having said that, if the code

is not part of Monero, then why bother about C++?

Rucknium Sorry I haven't had much time to read the papers, been trying to find a job. You've given me enough material to read through for a while, but if there's anything I should be working on ATM, let me know

tonz0fphun: Alternative plan: Stop your job search and work for the Monero Project. Half-serious. Especially if you can code C++.

tonz0fphun: As you read through papers could you jot down a few notes on the accompanying moneroresearch.info page. Any thoughts, questions, or summary of what the paper was about would help a lot even if it’s just a sentence or two. We are trying to make it a source of knowledge surrounding all Monero research for the community. Rucknium can get you set up with an account if you don’t have one already.

Finally, the write-up of my mining pool transaction confirmation delay research: rucknium.me/posts/monero-pool-transaction-delay

Reddit post (the blog post ^ has more detail): reddit.com/r/Monero/comments/10gapp…ed_mining_pools_are_delaying_monero

nice thanks

> <@tonz0fphun:monero.social> If you are comfortable with Python, you can write code that massively parallelises with numba, numpy, or cuda. The vast majority of ML, BLAS and similar libraries in Python are simple wrappers written in underlying C/C++ or CUDA. IMHO the simplicity and easy of use, as well... (full message at <libera.ems.host/_matrix/media/v3/do…0e5d7b586f1f81d5be87efa3ecb878e5042>)

<xmrack[m]> "tonz0fphun: As you read through..." <- Moneroresearch.info is a gold mine, but I've been discouraged by not finding papers I'd like to read (with limited time). Any recommendations?

ghostway: How could we make it more useful? Definitely open to ideas. It runs on WIKINDX. It has a lot of features that we could try

have stuff that orders stuff from the big picture to the details in every subject for example (could be provided, but really hard to do with the existing things). maybe also tags like keywords for other papers... but I don't really know

current search gives irrelevant results

I agree with the search results. It is too tokenized at the moment and appears to use a single ngram which is very verbose.

Imo that's the biggest pain point for me. I don't know if it is for everyone

without the imo*

to put things in perspective, it took a little under 2 weeks to get the titles bold

oh, well

Are there better options for us than WIKINDX?

no idea if you ask me. anyway, is there anyone that needs 10 helping fingers? currently just working on a weekend project, so not that busy

'monero focused' 'not monero focused' at least lets me see 'everything' moneroresearch.info/index.php?action=browse_BROWSECATEGORY_CORE

IMHO, WIKINDX is very good. Just need to learn to harness its power

<xmrack[m]> "I agree with the search results...." <- I have no idea what this means. Is it significant?

If you need more ways to classify papers, WIKINDX has them.

Rucknium: let me see if I can find a good example. I noticed it while importing the other day

Chain analysis by itself gives nothing

I don't think it searches the text of the PDFs. Just abstracts, titles, and authors. I may be able to turn on PDF searching

Wouldn't indexing that take a lot of space you could've used for other papers? Is space redundant?

We have enough space except when our log files blow up for some reason

Ok then heh

Rucknium... (full message at <libera.ems.host/_matrix/media/v3/do…9a4b3a5181a14fcdc018f6fe11f6bd0f1b5>)

There are some config options here we could set: wikindx.sourceforge.io/web/trunk/help-topics/search

In summary, the search is bad. plowsof and I can work on that

Thank you!

I am happy that there are enough papers to get lost in search. A year ago, the website did not exist. Thank you everyone who has contributed so far.