Review plonk_tutorial

[ethorhil]

I have I have started a new notebook with a working python version of Plonk in the plonk_tutorial folder. Would be great to get some feedback about the general direction. The code correctly verifies the example problem and I have finished about half the explanatory text so far.

I did try to continue the existing notebook plonk.ipynb but quickly realized that it would become too long for a tutorial at that level of detail. I think it could be great as a separate series of deepdive into FFTs etc.

What do you guys think? @barryWhiteHat @ChihChengLiang

[barryWhiteHat]

+1 on breaking the tutorial up into multiple sections.

[barryWhiteHat]

1. Its better to build with pull requests rather than pushing directly to master. It makes it easier to review and commet on changes.
2. I'd rather keep the style of the origianl tutorial where we ask the participant to build the code in order to continue the tutorial. Originally the tutorial has most of the code already but am planning to remove that.
3. The orginal tutorial was aimed at people who did not really understand zkps deeply. I am concerned this new version of it may be confusing to them. We use teh build to continue pattern to force them to figure things out / show us where people are having trouble.
4. I do like breaking the tutorial into multiple secions. WOuld it be better to take the original tutorial and split it?
5. What you are saying here is technically correct but I am worried it will be most accessible to people who already understand how this owrks. Here we want to target people who are starting out and try to educate them.

[ChihChengLiang]

Thanks for working on this issue! Here are some feedbacks.

• +1 on doing a pull request.
• We also have continuous integration set up for this repository, it automatically runs the tests to make sure they all pass before the pull request is merged.
• It's better to install py_ecc than include all its source code to the codebase.

Feel free to ping me for the engineering side of questions or even some questions you might feel a little bit trivial.

[ethorhil]

Ok, removed the files from master and did a PR. What we have so far could be split up into something like this:

Basics:

• Polynomials + FFT + Roots of unity

Plonk Specific Basics:

• Plonk Arithmetization
• Kate Commitments

Plonk: Putting it all together
edit: deleted the PR because there is a conflict with the py_ecc version

[barryWhiteHat]

Cool. Its better to build on the workbook we already have because there are tests and we worked a bunch on the flow.

Could also break out copy_constraints which is the most complicated peice IMO. So users would work thought polynomial + fft + ROU and then build towards getting copy constraints done before putting it al together.