This is amazing and super awesome, and I will be using it in all my talks
from now on. :) It makes me more excited for getting that site rewrite
project moving forward...

On Tue, Jun 4, 2024 at 12:27 AM Matthias Vogt <[email protected]>
wrote:

> Hi everyone,
> I wrote a cool visualizer for metamath proofs :3
>
> [image: Screenshot from 2024-06-03 20-54-43.png]
> It can...
>
>    - Show the proof steps as an interactive graph
>    - Show the proof as interactive nested proof blocks, from conclusion
>    to premises
>    - Summarize proofs and explain proof steps (I only hand-wrote
>    explanation-functions for a few theorems, e.g. "by transitivity..." below)
>    - Hide proof steps that you consider trivial, i.e. have a theorem
>    number less than x
>
> You can view it (please use desktop) at https://vo.gt/mm/mpeuni/2p2e4 or
> add it as a tampermonkey extension <https://vo.gt/mm/extension.js>. You
> can replace "2p2e4" in the URL with any theorem you like.
>
> This is at the moment a super slow and hacky proof of concept and there
> are likely many bugs. The site above mirrors metamath.tirix.org (if it is
> not OK that I am mirroring math.tirix.org, please tell me and I will of
> course take down this demo site. It is not indexed by search engines or
> linked anywhere except here.). It also makes lots of requests to
> metamath.org to get the theorem numbers. Moreover, it runs the
> expln.github.io <https://expln.github.io/metamath/asrt/2p2e4.html> react
> code to get the substitutions that show us how theorems are used in proof
> steps. The code is basically an unmaintainable fever dream :p If there is
> interest, I might rewrite it in react at some point in the future. But I
> hope you like it anyway :)
>
> The basic idea for summarizing proofs is that smaller theorem numbers
> correspond to logic glue, so we can just not mention these theorems
> explicitly in the proof. Instead, when we have a non-trivial step, we just
> silently add the application of the trivial theorem to the resulting
> expression. The threshold for what is trivial customizable with the slider.
> Swallowing logic glue helps for readability, but still, in reality, I think
> what part of a proof is verbose is more of a stylistic decision by the
> author.
>
> What would make the visualizer even better is if there was a way to get
> LaTeX or Unicode output for the terms in the substitutions. But I don't
> know how to do this. If expln.github.io
> <https://expln.github.io/metamath/asrt/2p2e4.html> had LaTeX/Unicode
> support, that could be used for the visualizer for example.
>
> Best
> Matthias
>
> --
> You received this message because you are subscribed to the Google Groups
> "Metamath" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/metamath/6fd3b538-ac91-431f-a476-9d79e4754b14n%40googlegroups.com
> <https://groups.google.com/d/msgid/metamath/6fd3b538-ac91-431f-a476-9d79e4754b14n%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>

-- 
You received this message because you are subscribed to the Google Groups 
"Metamath" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/metamath/CAFXXJStDO%2BphLRPGV63VGHzrc18vjej-72y2sfME8_QEMerCUw%40mail.gmail.com.

Reply via email to