The problem that finally defeated Claude | Project Euler 107

Описание: Three episodes in and the results keep flipping. This time Grok takes first place, Claude doesn't even finish, and Gemini submits a confident wrong answer.

Project Euler 107 — Minimal Network. You're given a fully connected network of 40 nodes with a total cable weight of 261,832. The goal is to find the minimum spanning tree — the cheapest set of connections that keeps every node reachable — and report how much weight you cut from the original.

Here's what each model did:

🔴 Gemini → scipy's MST function (wrong answer — forgot to halve the symmetric matrix total)
🟠 Claude → enumerated every possible spanning tree (never finished)
🟢 ChatGPT → Prim's algorithm (correct, clean)
🟣 Grok → Kruskal's + union-find (correct, and arguably the better fit for this problem)

We break down why Kruskal's is the more natural choice for a sparse edge list, what Gemini's off-by-two error actually was and how it happened, and why spanning tree enumeration is the kind of approach that looks reasonable until you realise the search space is 40 to the power of 38.

This is Project Euler Problem 107 — Minimal Network.

——
0:00 Intro
0:35 The Problem
1:19 Gemini's Approach
2:27 Claude's Approach
3:23 ChatGPT's Approach
4:30 Grok's Approach
5:37 Head to Head
6:18 The Verdict

——
#ai #chatgpt #claude #shorts #artificialintelligence #machinelearning #coding #programming #algorithms #maths #openai #anthropic #computerscience #projecteuler #aicomparison #techshorts #python #datascience #cs #aivsai #elonmusk #spacex #grok