teso-black-box-optimization

Apply tabu search enhanced with short-term tabu lists and long-term elite memory for simulation optimization with expensive, noisy evaluations. Balances exploration (avoiding cycling) and exploitation (leveraging best solutions). Use for multimodal landscapes where function evaluations are costly and multiple function calls per solution are impractical.

インストールへ移動

リポジトリ: ADu2021/skillXiv
ソースの最終更新活動: 2026年3月24日 19:42
検出された SKILL.md の言語: 英語
スター: 6
フォーク: 0

インストール方法

デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。

Codex または Claude でインストールこの Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。

直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。

npx skills add https://github.com/ADu2021/skillXiv --skill teso-black-box-optimization

コマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。

ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。

ソースファイルを確認

インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。

SKILL.md

ソースの指示 · 読み取り専用プレビュー

検索に戻るインストールに戻る

meaningful-kebab-case-name

Convert arXiv papers into ready-to-use agent skills using category-aware extraction. First classifies the paper into one or more of 11 research categories, then applies a specialized extraction pipeline for each category — because different types of papers produce different types of usable knowledge. A single paper can yield multiple skills if it spans categories. Use this skill whenever the user wants to turn a paper into a skill, extract practical techniques from research, build a skill library from papers, convert arXiv papers into reusable agent instructions, or batch-process multiple papers into skills. Also trigger when someone asks about extracting actionable knowledge from papers, making research practical for LLM agents, or systematically converting academic contributions into structured agent capabilities.

2026-03-26

action-quantization-behavior-cloning

Establish regret bounds for behavior cloning with discretized actions combining statistical error and quantization error terms. Prove smoothness requirements for safe quantizer design, show that learning-based quantizers fail these requirements, and propose model-based augmentation to reduce error dependence from H² to H.

2026-03-26

adaptive-lora-personalized-ranks

Dynamically allocate LoRA ranks per-layer during fine-tuning instead of using fixed uniform ranks. Learn optimal rank for each layer and subject via variational framework with discretized exponential distribution, reducing memory footprint while maintaining fidelity and text-alignment.

2026-03-26

このリポジトリのすべてを見る

Problem: Design a hospital queueing system Variables: Number of doctors, nurses, waiting areas Evaluation: Run 1000-hour simulation → collect wait times, satisfaction scores Issue: Each simulation takes 2 minutes. Noisy (random arrivals differ each run). Landscape is multimodal (different staffing configs may be optimal).

Iteration 0: Visit solution X Iteration 1: Explore neighbors of X Iteration 2: Find neighbor Y with improvement Iteration 3: Visit Y (move from X → Y) Iteration 4: X is now TABU (forbidden) for T=5 iterations Iteration 5: Neighbors of Y are explored, but can't go back to X Iteration 6: Can't go to X (still tabu) ... Iteration 9: X becomes non-tabu again

Best solutions found: - Solution A: Objective = 85.2 (iteration 5) - Solution B: Objective = 87.1 (iteration 18) - Solution C: Objective = 86.9 (iteration 42) Later, if stuck at objective 80, restart from elite B or C (perturbed slightly to encourage new exploration)

# TESO: Tabu search with elite memory for black-box optimization class TabuEnhancedSimulationOptimization: def __init__(self, objective_fn, tabu_tenure=7, elite_pool_size=5, perturbation_strength=0.1): self.objective_fn = objective_fn # Black-box simulator self.tabu_tenure = tabu_tenure # Steps before solution un-tabued self.elite_pool = [] # Top K solutions found self.elite_pool_size = elite_pool_size self.perturbation_strength = perturbation_strength self.tabu_list = set() # Currently forbidden solutions self.iteration = 0 self.best_solution = None self.best_objective = float('-inf') def optimize(self, initial_solution, max_iterations=1000): """Main TESO loop""" current = initial_solution.copy() current_obj = self.objective_fn(current) for iteration in range(max_iterations): self.iteration = iteration # Step 1: Generate neighborhood around current solution neighbors = self.generate_neighborhood(current) # Step 2: Filter out tabu solutions, apply aspiration criterion candidates = [] for neighbor in neighbors: if not self.is_tabu(neighbor): candidates.append(neighbor) elif self.aspiration_criterion(neighbor): # Aspiration: override tabu if solution is very good candidates.append(neighbor) if not candidates: # If all neighbors are tabu, pick best tabu (with penalty) candidates = neighbors # Force at least one move # Step 3: Evaluate candidates candidate_objs = [self.objective_fn(c) for c in candidates] best_idx = np.argmax(candidate_objs) best_neighbor = candidates[best_idx] best_neighbor_obj = candidate_objs[best_idx] # Step 4: Move to best neighbor (even if worse: key to escaping local optima) current = best_neighbor current_obj = best_neighbor_obj # Step 5: Update global best and elite memory if current_obj > self.best_objective: self.best_objective = current_obj self.best_solution = current.copy() print(f"Iteration {iteration}: New best = {self.best_objective:.4f}") # Step 6: Update elite pool self.update_elite_pool(current, current_obj) # Step 7: Add current to tabu list (will be un-tabued after tenure) self.tabu_list.add(tuple(current)) self.remove_expired_tabu() # Step 8: Periodically restart from elite solutions if iteration % 50 == 0 and self.elite_pool: restart_solution = self.elite_pool[0]['solution'].copy() # Perturb for novelty current = self.perturb(restart_solution, self.perturbation_strength) return self.best_solution, self.best_objective def generate_neighborhood(self, solution, neighborhood_size=20): """Create neighborhood by local perturbations""" neighbors = [] for _ in range(neighborhood_size): neighbor = solution.copy() # Random local change (e.g., variable adjustment) variable_idx = np.random.randint(len(solution)) delta = np.random.normal(0, 0.1) neighbor[variable_idx] += delta neighbors.append(neighbor) return neighbors def is_tabu(self, solution): """Check if solution is currently forbidden""" return tuple(solution) in self.tabu_list def aspiration_criterion(self, solution): """Override tabu restriction if solution is exceptionally good""" # Test evaluation (noisy, so approximate) obj_estimate = self.objective_fn(solution) # Aspiration: override if better than current best by margin return obj_estimate > self.best_objective + 0.1 * abs(self.best_objective) def update_elite_pool(self, solution, objective): """Maintain elite pool of best K solutions""" self.elite_pool.append({ 'solution': solution.copy(), 'objective': objective, 'iteration': self.iteration }) # Keep only top K self.elite_pool.sort(key=lambda x: x['objective'], reverse=True) self.elite_pool = self.elite_pool[:self.elite_pool_size] def remove_expired_tabu(self): """Un-tabu solutions whose tenure has expired""" # Simplified: in practice, track iteration number for each tabu if self.iteration % self.tabu_tenure == 0: self.tabu_list.clear() # Reset tabu list periodically def perturb(self, solution, strength): """Perturb solution for restart exploration""" perturbed = solution.copy() for i in range(len(perturbed)): perturbed[i] += np.random.normal(0, strength) return perturbed

name	teso-black-box-optimization
title	TESO Tabu Enhanced Simulation Optimization for Noisy Black Box Problems
version	0.0.2
engine	skillxiv-v0.0.2-claude-opus-4.6
license	MIT
url	https://arxiv.org/abs/2512.24007
keywords	["black-box optimization","metaheuristic","tabu search","noisy evaluation","memory-guided search","simulation optimization"]
description	Apply tabu search enhanced with short-term tabu lists and long-term elite memory for simulation optimization with expensive, noisy evaluations. Balances exploration (avoiding cycling) and exploitation (leveraging best solutions). Use for multimodal landscapes where function evaluations are costly and multiple function calls per solution are impractical.

name	teso-black-box-optimization
title	TESO Tabu Enhanced Simulation Optimization for Noisy Black Box Problems
version	0.0.2
engine	skillxiv-v0.0.2-claude-opus-4.6
license	MIT
url	https://arxiv.org/abs/2512.24007
keywords	["black-box optimization","metaheuristic","tabu search","noisy evaluation","memory-guided search","simulation optimization"]
description	Apply tabu search enhanced with short-term tabu lists and long-term elite memory for simulation optimization with expensive, noisy evaluations. Balances exploration (avoiding cycling) and exploitation (leveraging best solutions). Use for multimodal landscapes where function evaluations are costly and multiple function calls per solution are impractical.

Parameter	Role	Typical Range
Tabu tenure	How long to forbid solutions	5-15 (proportion of neighborhood)
Elite pool size	Number of best solutions to track	3-10
Neighborhood size	Candidates per iteration	10-50
Perturbation strength	How much to perturb restarts	0.05-0.3x variable range
Restart frequency	When to restart from elite	Every 30-100 iterations

Baseline	When TESO Wins
Random search	Always (tabu + memory vs. pure randomness)
Simulated Annealing	Complex multimodal landscapes (better memory)
Genetic Algorithms	Few variables, expensive evals (GA needs population)
Bayesian Optimization	Limited evaluation budget (BO assumes smooth)

teso-black-box-optimization

インストール方法

ソースファイルを確認

When to Use This Skill

When NOT to Use This Skill

The Simulation Optimization Challenge

Core Algorithm: Tabu Search with Memory

1. Short-Term Tabu List

2. Long-Term Elite Memory

When to Use This Skill

When NOT to Use This Skill

The Simulation Optimization Challenge

Core Algorithm: Tabu Search with Memory

1. Short-Term Tabu List

2. Long-Term Elite Memory

Architecture Pattern

Handling Noisy Evaluations

Multimodal Landscape Strategy

Empirical Example: Queue Optimization

Tuning Hyperparameters

When TESO Outperforms Alternatives

Implementation Checklist

References

teso-black-box-optimization

ソース情報

インストール方法

ソースファイルを確認

このリポジトリの他の Skills

このリポジトリの他の Skills

When to Use This Skill

When NOT to Use This Skill

The Simulation Optimization Challenge

Core Algorithm: Tabu Search with Memory

1. Short-Term Tabu List

2. Long-Term Elite Memory

When to Use This Skill

When NOT to Use This Skill

The Simulation Optimization Challenge

Core Algorithm: Tabu Search with Memory

1. Short-Term Tabu List

2. Long-Term Elite Memory

Architecture Pattern

Handling Noisy Evaluations

Multimodal Landscape Strategy

Empirical Example: Queue Optimization

Tuning Hyperparameters

When TESO Outperforms Alternatives

Implementation Checklist

References