Skip to content

Add the LIPO and RankOpt global optimization algorithms - #216

Open
gaetanserre wants to merge 11 commits into
LeanMachineLearning:mainfrom
gaetanserre:decision_optimization
Open

gaetanserre wants to merge 11 commits into
LeanMachineLearning:mainfrom
gaetanserre:decision_optimization

Conversation

@gaetanserre

Copy link
Copy Markdown
Contributor

Add the LIPO and RankOpt global optimization algorithms.

Define the Decision interface for decision-based algorithm i.e., algorithms that accepts a candidate point sampled from a probability measure w.r.t. its membership to a decision set.

Comment thread LeanMachineLearning/Optimization/Algorithms/RankOpt.lean Outdated
Comment thread LeanMachineLearning/Optimization/Algorithms/LIPO.lean Outdated
(htop : ∀ n (data : Iic n → α × β), κs n data (decision n data) ≠ ⊤)

/-- The interface for decision-based optimization algorithms. -/
noncomputable def Decision : Algorithm α β where

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm not a fan of the Decision name. It's too vague. Every algorithm is a decision procedure.
Can we try to describe it more precisely? What you are doing is taking a sequence of kernels (an Algorithm, up to the fact that they might not be Markov), and conditioning them on some sets. It almost feels like it should be Algorithm.cond, but the kernels are not necessarily Markov so it can't be. I'll be thinking about a name.

gaetanserre and others added 3 commits August 24, 2026 15:59
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants