Zero K proof that a chess position contains a checkmate

AI Thread Summary
The discussion revolves around building a zero-knowledge proof (ZKP) to demonstrate that a specific chess position contains at least one checkmate without revealing the checkmate itself. The RISC Zero zkVM is highlighted as a tool that can create such proofs, with an example provided in a GitHub repository that utilizes the shakmaty crate to validate checkmate conditions. The author seeks a detailed understanding of the process, particularly how to convert a chess position represented in FEN notation into a graph suitable for a three-coloring algorithm. This transformation is crucial for proving the existence of a checkmate. The author notes that while using shakmaty can assert the presence of a checkmate, the RISC0 framework offers a more comprehensive zero-knowledge setup, which is currently beyond their understanding.
fluidistic
Gold Member
Messages
3,928
Reaction score
272
Hi people,
It's been years I wanted to post this question here. I would like to build a zero knowledge proof that a given chess position contains at least one checkmate. I know that anything provable admits a zero k proof. I know about https://crypto.stackexchange.com/questions/110939/zero-knowledge-proof-applied-to-a-chess-position and .
I know it's been done already (see https://github.com/risc0/risc0/tree/main/examples/chess):
This code demonstrates a minimal example of how to use the RISC Zero zkVM to make ZK proofs about chess.


The demo uses the shakmaty crate to prove that a chess position has a checkmate without revealing what that checkmate is.
But I would like to understand exactly how to do so, every single step. There are other ways to accomplish it. I think I should be able to transform the problem into a graph/map with 3 colors scheme. I.e. if I can convince the Verifier that I can color the map with 3 colors such that no 2 colors are adjacent, then the proof would be complete. The hard part I don't know how to do is to apply an algorithm to transform a given FEN (or chess position) into such a graph. For example, this position
Code:
6Q1/8/8/8/8/8/5K2/7k w - - 0 1
contains 4 ways to checkmate in 1.
 
Computer science news on Phys.org
You need to feed the FEN into a chess engine, and get it to tell you what all of the legal moves are and what the resulting positions are.
 
pasmith said:
You need to feed the FEN into a chess engine, and get it to tell you what all of the legal moves are and what the resulting positions are.
Not exactly. I could use shakmaty (not even a,chess engine) like risc0 does, to assert whether the FEN contains a checkmate. You don't even need to invoke the command to check all legal moves. However risc0 does more than this, for it is designed to be a real 0k setup. It's over my,head for now.
 
I came across a video regarding the use of AI/ML to work through complex datasets to determine complicated protein structures. It is a promising and beneficial use of AI/ML. AlphaFold - The Most Useful Thing AI Has Ever Done https://www.ebi.ac.uk/training/online/courses/alphafold/an-introductory-guide-to-its-strengths-and-limitations/what-is-alphafold/ https://en.wikipedia.org/wiki/AlphaFold https://deepmind.google/about/ Edit/update: The AlphaFold article in Nature John Jumper...
Interesting article about an AI writing scandal at Sports Illustrated: https://www.cnn.com/2023/11/29/opinions/sports-illustrated-ai-controversy-leitch/index.html I hadn't heard about it in real-time, which is probably indicative about how far SI has fallen*. In short, the article discusses how SI was caught using AI and worse fake reporter photos/profiles to write game summaries. Game summaries are the short articles that summarize last night's Phillies game. They are so formulaic that...

Similar threads

Replies
25
Views
4K
Replies
77
Views
12K
3
Replies
105
Views
14K
4
Replies
175
Views
25K
Replies
28
Views
6K
Replies
13
Views
3K
Replies
10
Views
5K
Back
Top