feat: Expand underpopulated optimization categories (social_inspired, probabilistic, constrained) #36
Description
Summary
Several optimization categories have significantly fewer algorithms compared to well-populated categories like swarm_intelligence (50+ algorithms). This issue tracks potential algorithms to add to balance the library.
Current State
| Category | Count | Algorithms |
|---|---|---|
social_inspired |
4 | TeachingLearningOptimizer, PoliticalOptimizer, SoccerLeagueOptimizer, SocialGroupOptimizer |
probabilistic |
5 | LDAnalysis, ParzenTreeEstimator, BayesianOptimizer, SequentialMonteCarloOptimizer, AdaptiveMetropolisOptimizer |
constrained |
5 | AugmentedLagrangian, SuccessiveLinearProgramming, BarrierMethodOptimizer, PenaltyMethodOptimizer, SequentialQuadraticProgramming |
Proposed Additions
Social-Inspired Algorithms
- Election Algorithm (EOA) - Meta-heuristic based on election processes (MDPI 2024)
- Greedy Politics Optimization - Political strategies during state elections
- Social Learning Optimizer (SLO) - Human social learning behaviors
- Anarchic Society Optimization (ASO) - Anarchic social behaviors
- Brain Storm Optimization (BSO) - Human brainstorming process
Probabilistic Algorithms
- Hamiltonian Monte Carlo (HMC) - MCMC using Hamiltonian dynamics
- Slice Sampling - Adaptive MCMC method
- Gibbs Sampling - Component-wise MCMC sampling
- Parallel Tempering - Multiple temperature MCMC chains
- Nested Sampling - Bayesian evidence computation
Constrained Algorithms
- Interior Point Method - Primal-dual interior point optimization
- ADMM - Alternating Direction Method of Multipliers
- Filter Method - Constraint violation filter approach
- Exact Penalty Method - L1 exact penalty approach
- Projection Method - Gradient projection for constraints
References
- Election Algorithm: https://www.mdpi.com/2227-7390/12/10/1513
- Hamiltonian Monte Carlo: Neal, R. M. (2011) "MCMC using Hamiltonian dynamics"
- Interior Point Methods: Nocedal & Wright, "Numerical Optimization"
- ADMM: Boyd et al. (2011) "Distributed Optimization and Statistical Learning"
Acceptance Criteria
- Each new algorithm follows
AbstractOptimizerpattern - Includes proper docstrings with references
- Added to respective
__init__.pyexports - Included in test suite
- All tests pass