Labo333

Thanks a lot :)

No but you can install any library from github with

pip install --upgrade git+[URL]

I don't think Sage has the Jacobi symbol (just the Legendre), but Jacobi and Solovay-Strassen are less than 20 lines.

Anyway, the goal was to provide very simple and readable implementations. Sage is far more powerful.

Also, I think Sage uses square free factorisation algorithms that run much faster than Berlekamp.

Anyway, we had much fun coding that and I really didn't expect to receive so many uovotes!

I just let this here, I feel it was missing as another failed attempt to prove P = NP.

I don't think he is a jerk, but must admit I was surprised by his reaction.

I wouldn't entrust him with the business part is he was too soft either, so I guess it is ok.

So, I investigated the split between Jobs and Woz, and it seems that it was 65.4/34.6 at the date of the IPO (p. 22 of https://wiki.epfl.ch/a3-angels/documents/reunions/a3angels07...).

But when the company started, the split was 45/45 (10 for Wayne) (http://www.foundersatwork.com/steve-wozniak.html).

Steve Wozniak was not interested in money, that is how I explain the difference. Still, even at the time of the IPO, the split was slightly more than 33/66.

In fact, it is possible for us to find money in a lot of other places, and I think his work is at least as important as mine. It would be easy for us to get 500k from external investors.

Plus, I'm not in need for money, so I don't care about a rate. For the moment he is CEO and I'm Chief Scientist.

That is why I don't think the money is the big deal here, and even he didn't talk much about it because he knows that.

It is a software related company, but much more a scientific company (about cryptography and machine learning).

Yeah the split between Jobs and Woz was one of his arguments, but do you think it was fair? Would you do the same as Woz?

Besides, I am still studying (while he is "unemployed") and want to study next year (September 2018) in another country (1 hour by plane). He was worried a lot about the fact I wouldn't be able to work on the idea, but I promised him I would be as much available as I can, and that I just needed to go to the university for the exams (which is true) and could handle my thesis by emails. So I'll be a but sacrificing my degree for our project.

Now, we are discussing about the shares, more precisely what our relative repartition should be after some rounds of founding.

I said 40/60 while he said 33/66. I feel bad about the fact he would have twice as much as me.

He said he was the one with the ideas, and that although I always had all the answers to his questions, the questions are more valuable. I don't think it is true, IMO I asked him the business related questions while he asked me the technical ones. It is just that he had a lot more questions because… well… business people don't know anything, while business is easily understandable and I have a broad experience that allows me to understand it. So in a sense his argument is just that BECAUSE I have more knowledge I contributed less.

He said he only came to see me as a technical consultant and that 33% was already a lot.

He also gave me a lot of examples of companies where the technical guys don't get much shares (Amazon, Facebook, etc…), and I must say I begin to have doubts.

What do you think of:

- the right repartition of equity

- how I should react

Thanks :)

Cool page! It would be really great to have the code shipped in a single file so that people can use it!

I love Jupyter notebooks because 1) they are beautiful 2) if you download them, you can play :)

Doug's implementation seems indeed really fast.

Hi, sorry, I don't have anything in particular. However, it seems that http://snnynhr.github.io/ParallelSuffixArrays/ contains interesting algorithms and links.

I'm not sure about the difference, but yes :) I wanted an implementation that could fit on my screen and still be efficient. Of course Python is slow and even my algorithm is not the best (I explained why I chose it). But I believe this is one of the best implementation you can make of prefix doubling, especially against random data and I have never seen my tricks anywhere.

OP and author here :)

Python is not the fastest programming language in the world either. As I said, I wanted something simple (less than 20 lines and quite readable) that performs well against random data. It is NOT intended to be used in production. You would have to add some checks for border cases and things like that. Plus it uses a lot of memory (although linear) because of the set and the dict, which is a drawback that the original implementation (suffix_matrix_stanford) does not have.

I clearly took simplicity over efficiency, and the advantage of this particular algorithm is that you compute the whole suffix matrix that you can reuse for the LCP.

Thank you for divsufsort, it seems to be really fast! During my search, I found this algorithm that does really well against the others, including divufsort: https://arxiv.org/abs/1307.1417

Without thinking too much, I think it is quite easy to make the prefix doubling algorithm parallel if you have a good parallel sorting algorithm, but it would of course not be the best solution…

I think the best possible laptop for development that is not too expensive is simply a macbook pro unibody 2012, with a SSD and 16GB RAM. I still costs about 1000$ and you have to buy a SSD and RAM though they are reusable.

Connectivity : It's the last macbook pro to have both a ethernet port and a DVD slot. And you have one of the new thunderbolt ports and a firewire port.

The i5 processor is still good. The only drawback is the screen (and the GPU). But for development it doesn't matter, and for my part, I watch movies on it without any problem. With SSD and RAM you have a computer that is as fast as the latest macbook pro.

It is solid and quite beautiful like all Apple portable computers. The battery is not very powerful (about 5 hours web browsing with chrome), but you can easily get external batteries (like the ThinkPad).

It is easy to install linux on it with the new boot system, though I like macOS because it is still Unix and has a lot of convenient apps and more softwares (and I can still install most of linux apps with macports or compile them myself).