- formatting
- images
- links
- math
- code
- blockquotes
- external-services
•
•
•
•
•
•
-
Counting the Realisable States of an N-Tuple
Most entries of an n-tuple table can never be addressed on a Connect-4 board, because gravity forbids the patterns they describe. A column-by-column count, a closed form, a check against ten trained agents, and a ranking that stores a trained network without any indices.
-
Crossing the Frozen Lake: Monte Carlo, TD(0) and Q-Learning
Monte Carlo, TD(0) and Q-learning on an 8×8 Frozen Lake that is small enough to inspect every learned value: how experience turns into state values, why slippery ice changes them, and how action values lead to a policy that crosses the lake.
-
Almost-Equal Isosceles Triangles: When Height Nearly Matches Base
An exploration of a curious class of integer-sided isosceles triangles whose height differs from the base by exactly one. What begins as a simple geometric question quickly leads to Pell equations and elegant recurrences.
-
Online Estimation: Updating the Inverse Covariance Matrix
Deriving Woodbury and Sherman-Morrison updates for the scatter matrix step by step, handling the singular startup phase explicitly, and verifying the reported covariance and precision with Python and R.
-
Building Intelligent Agents for Connect-4: Final Considerations
The remaining techniques, an honest accounting of what each optimization was actually worth, the ideas which did not work, and why a perfect-playing Connect-4 agent has learned nothing at all.