Add three unified benchmark tasks

[zqxie77]

No description provided.

[github-actions]

🤖 AI Code Review (gemini-3-flash-preview)

🇬🇧 English Analysis

1. Executive Summary

• Core Purpose: This PR introduces three new engineering benchmark tasks to the frontier_eval framework: AdaptiveLinkScheduling (Communication Engineering), AGVWarehouseRouting (Robotics), and MicrogridBatteryDispatch (Power Systems). It includes task definitions, seed scripts, and deterministic evaluators for these domains.
• Modified File Structure & Modifications:
  • TASK_DETAILS.md & TASK_DETAILS_zh-CN.md: Updated the master task tables to include the three new subtasks across both English and Chinese versions.
  • benchmarks/CommunicationEngineering/AdaptiveLinkScheduling/: Added a new subtask folder containing README.md (usage), Task.md (specifications), scripts/init.py (baseline heuristic), and verification/evaluator.py (SNR-based link simulator).
  • benchmarks/PowerSystems/MicrogridBatteryDispatch/: Added a new subtask folder (content partially truncated in diff) including README.md and Task.md defining battery dispatch optimization under demand charges and degradation.
  • benchmarks/CommunicationEngineering/README.md & README_zh-CN.md: Updated the domain index to include the new scheduling task.

2. AI Content Analysis

• Estimated AI Component: 40%
• Reasoning & Evidence:
  • Documentation Structure: The Task.md and README.md files follow a highly consistent, templated structure seen in other tasks within this repo, likely generated or refined by AI to maintain uniformity.
  • Boilerplate Code: The verification/evaluator.py uses very standard Python patterns for loading modules dynamically (importlib.util.spec_from_file_location) and writing JSON, which are common AI-generated boilerplate snippets.
  • Domain Logic: The core logic in scripts/init.py (e.g., the score = snr + 3.0 * latency + ... heuristic) and the SNR thresholding in the evaluator show domain-specific knowledge (Jain Fairness, MCS tables) that suggests human guidance or high-quality prompt engineering, rather than raw generic AI output.

3. Engineering & Economic Assessment

• Engineering Reality Check:
  • High Realism: The AdaptiveLinkScheduling task correctly models 5G/LTE-like constraints, including MCS (Modulation and Coding Scheme) thresholds, power budgets in mW, and Jain's Fairness Index. It avoids being a "toy" by incorporating implementation margins and outage penalties.
  • Edge Case Handling: The evaluator explicitly handles non-finite power values, clips power to min/max bounds, and penalizes invalid user/MCS IDs, demonstrating production-grade robustness.
• Economic Value: High. These tasks represent multi-billion dollar optimization problems. Efficient link scheduling increases telecom spectral efficiency, and optimal battery dispatch directly reduces commercial electricity bills (demand charge management) and extends hardware lifespan (degradation modeling).

4. Quality Assurance

• Verification & Testing:
  • frontier_eval Integration: Yes.
  • task_name: CommunicationEngineering/AdaptiveLinkScheduling, PowerSystems/MicrogridBatteryDispatch, and Robotics/AGVWarehouseRouting (via task=unified task.benchmark=...).
  • Execution & Dependencies: The .md files provide exact python -m frontier_eval commands. requirements.txt correctly identifies that only the Python standard library is needed for the provided evaluators.
• Documentation Quality: High. The documentation is bilingual and provides clear API contracts (input/output dictionaries). No significant grammatical errors were detected in the visible diff.
• Organizational Structure: Logical and consistent. It adheres to the existing repository hierarchy: benchmarks/{Domain}/{TaskName}/{scripts|verification|frontier_eval}.

5. Security & Privacy Check

• Sensitive Files: Clean. No .env, API keys, or IDE-specific metadata (.vscode) were found.
• Absolute Paths: None detected. The evaluator uses Path(candidate_path).resolve() and relative pathing for file operations.

---

🇨🇳 中文分析

1. 摘要

• 核心目的: 此 PR 为 frontier_eval 框架引入了三个新的工程基准任务：AdaptiveLinkScheduling（通信工程）、AGVWarehouseRouting（机器人）和 MicrogridBatteryDispatch（电力系统）。包含了任务定义、种子脚本以及针对这些领域的确定性评估器。
• 修改的文件结构与变更摘要:
  • TASK_DETAILS.md & TASK_DETAILS_zh-CN.md: 更新了主任务表，在中英文版本中均添加了这三个子任务。
  • benchmarks/CommunicationEngineering/AdaptiveLinkScheduling/: 新增子任务目录，包含 README.md（用法）、Task.md（规范）、scripts/init.py（基准启发式算法）和 verification/evaluator.py（基于 SNR 的链路仿真器）。
  • benchmarks/PowerSystems/MicrogridBatteryDispatch/: 新增子任务目录（部分内容在 diff 中被截断），包含定义需量电费和退化成本下电池调度优化的文档。
  • benchmarks/CommunicationEngineering/README.md & README_zh-CN.md: 更新了领域索引以包含新的调度任务。

2. AI 成分分析

• 预估 AI 含量: 40%
• 判断依据与证据:
  • 文档结构: Task.md 和 README.md 文件遵循高度一致的模板结构，这与仓库中其他任务一致，很可能是由 AI 生成或润色以保持统一性。
  • 样板代码: verification/evaluator.py 使用了非常标准的 Python 模式来动态加载模块（importlib.util.spec_from_file_location）和写入 JSON，这些是常见的 AI 生成样板代码片段。
  • 领域逻辑: scripts/init.py 中的核心逻辑（例如 score = snr + 3.0 * latency + ... 启发式算法）以及评估器中的 SNR 阈值处理显示了特定领域的知识（Jain 公平指数、MCS 表），这表明有专家指导或高质量的提示词工程，而非纯粹的通用 AI 输出。

3. 工程与经济评估

• 工程现实检验:
  • 高现实性: AdaptiveLinkScheduling 任务正确模拟了类 5G/LTE 约束，包括 MCS（调制与编码策略）阈值、以毫瓦 (mW) 为单位的功率预算以及 Jain 公平指数。通过引入实现余量（implementation margins）和停机惩罚，避免了成为“玩具示例”。
  • 边缘情况处理: 评估器显式处理了非有限功率值，将功率限制在最小/最大范围内，并对无效的用户/MCS ID 进行惩罚，展示了生产级的健壮性。
• 经济价值: 高。这些任务代表了价值数十亿美元的优化问题。高效的链路调度可提高电信频谱效率，而优化的电池调度可直接降低商业电费（需量电费管理）并延长硬件寿命（退化建模）。

4. 质量保证

• 验证与测试:
  • frontier_eval 集成: 是。
  • task_name: CommunicationEngineering/AdaptiveLinkScheduling, PowerSystems/MicrogridBatteryDispatch, 以及 Robotics/AGVWarehouseRouting (通过 task=unified task.benchmark=... 调用)。
  • 运行与依赖: .md 文件提供了确切的 python -m frontier_eval 命令。requirements.txt 正确指出所提供的评估器仅需 Python 标准库。
• 文档质量: 高。文档采用双语编写，并提供了清晰的 API 契约（输入/输出字典）。在可见的 diff 中未检测到明显的语法错误。
• 组织结构: 逻辑清晰且一致。遵循现有的仓库层级结构：benchmarks/{领域}/{任务名}/{scripts|verification|frontier_eval}。

5. 安全与隐私检查

• 敏感文件: 未发现异常。未发现 .env、API 密钥或 IDE 特定元数据（.vscode）。
• 绝对路径: 未检测到。评估器使用 Path(candidate_path).resolve() 和相对路径进行文件操作。

[github-actions]

🤖 AI Code Review (gemini-3-flash-preview)

🇬🇧 English Analysis

1. Executive Summary

• Core Purpose: This PR introduces three new engineering benchmark tasks to the frontier_eval framework: AdaptiveLinkScheduling (Communication Engineering), AGVWarehouseRouting (Robotics), and MicrogridBatteryDispatch (Power Systems). It includes task definitions, seed scripts, and deterministic evaluators for these domains.
• Modified File Structure & Modifications:
  • TASK_DETAILS.md & TASK_DETAILS_zh-CN.md: Updated the master task tables to include the three new subtasks across both English and Chinese versions.
  • benchmarks/CommunicationEngineering/AdaptiveLinkScheduling/: Added a new subtask folder containing README.md (usage), Task.md (specifications), scripts/init.py (baseline heuristic), and verification/evaluator.py (SNR-based link simulator).
  • benchmarks/PowerSystems/MicrogridBatteryDispatch/: Added a new subtask folder (content partially truncated in diff) including README.md and Task.md defining battery dispatch optimization under demand charges and degradation.
  • benchmarks/CommunicationEngineering/README.md & README_zh-CN.md: Updated the domain index to include the new scheduling task.

2. AI Content Analysis

• Estimated AI Component: 40%
• Reasoning & Evidence:
  • Documentation Structure: The Task.md and README.md files follow a highly consistent, templated structure seen in other tasks within this repo, likely generated or refined by AI to maintain uniformity.
  • Boilerplate Code: The verification/evaluator.py uses very standard Python patterns for loading modules dynamically (importlib.util.spec_from_file_location) and writing JSON, which are common AI-generated boilerplate snippets.
  • Domain Logic: The core logic in scripts/init.py (e.g., the score = snr + 3.0 * latency + ... heuristic) and the SNR thresholding in the evaluator show domain-specific knowledge (Jain Fairness, MCS tables) that suggests human guidance or high-quality prompt engineering, rather than raw generic AI output.

3. Engineering & Economic Assessment

• Engineering Reality Check:
  • High Realism: The AdaptiveLinkScheduling task correctly models 5G/LTE-like constraints, including MCS (Modulation and Coding Scheme) thresholds, power budgets in mW, and Jain's Fairness Index. It avoids being a "toy" by incorporating implementation margins and outage penalties.
  • Edge Case Handling: The evaluator explicitly handles non-finite power values, clips power to min/max bounds, and penalizes invalid user/MCS IDs, demonstrating production-grade robustness.
• Economic Value: High. These tasks represent multi-billion dollar optimization problems. Efficient link scheduling increases telecom spectral efficiency, and optimal battery dispatch directly reduces commercial electricity bills (demand charge management) and extends hardware lifespan (degradation modeling).

4. Quality Assurance

• Verification & Testing:
  • frontier_eval Integration: Yes.
  • task_name: CommunicationEngineering/AdaptiveLinkScheduling, PowerSystems/MicrogridBatteryDispatch, and Robotics/AGVWarehouseRouting (via task=unified task.benchmark=...).
  • Execution & Dependencies: The .md files provide exact python -m frontier_eval commands. requirements.txt correctly identifies that only the Python standard library is needed for the provided evaluators.
• Documentation Quality: High. The documentation is bilingual and provides clear API contracts (input/output dictionaries). No significant grammatical errors were detected in the visible diff.
• Organizational Structure: Logical and consistent. It adheres to the existing repository hierarchy: benchmarks/{Domain}/{TaskName}/{scripts|verification|frontier_eval}.

5. Security & Privacy Check

• Sensitive Files: Clean. No .env, API keys, or IDE-specific metadata (.vscode) were found.
• Absolute Paths: None detected. The evaluator uses Path(candidate_path).resolve() and relative pathing for file operations.

---

🇨🇳 中文分析

1. 摘要

• 核心目的: 此 PR 为 frontier_eval 框架引入了三个新的工程基准任务：AdaptiveLinkScheduling（通信工程）、AGVWarehouseRouting（机器人）和 MicrogridBatteryDispatch（电力系统）。包含了任务定义、种子脚本以及针对这些领域的确定性评估器。
• 修改的文件结构与变更摘要:
  • TASK_DETAILS.md & TASK_DETAILS_zh-CN.md: 更新了主任务表，在中英文版本中均添加了这三个子任务。
  • benchmarks/CommunicationEngineering/AdaptiveLinkScheduling/: 新增子任务目录，包含 README.md（用法）、Task.md（规范）、scripts/init.py（基准启发式算法）和 verification/evaluator.py（基于 SNR 的链路仿真器）。
  • benchmarks/PowerSystems/MicrogridBatteryDispatch/: 新增子任务目录（部分内容在 diff 中被截断），包含定义需量电费和退化成本下电池调度优化的文档。
  • benchmarks/CommunicationEngineering/README.md & README_zh-CN.md: 更新了领域索引以包含新的调度任务。

2. AI 成分分析

• 预估 AI 含量: 40%
• 判断依据与证据:
  • 文档结构: Task.md 和 README.md 文件遵循高度一致的模板结构，这与仓库中其他任务一致，很可能是由 AI 生成或润色以保持统一性。
  • 样板代码: verification/evaluator.py 使用了非常标准的 Python 模式来动态加载模块（importlib.util.spec_from_file_location）和写入 JSON，这些是常见的 AI 生成样板代码片段。
  • 领域逻辑: scripts/init.py 中的核心逻辑（例如 score = snr + 3.0 * latency + ... 启发式算法）以及评估器中的 SNR 阈值处理显示了特定领域的知识（Jain 公平指数、MCS 表），这表明有专家指导或高质量的提示词工程，而非纯粹的通用 AI 输出。

3. 工程与经济评估

• 工程现实检验:
  • 高现实性: AdaptiveLinkScheduling 任务正确模拟了类 5G/LTE 约束，包括 MCS（调制与编码策略）阈值、以毫瓦 (mW) 为单位的功率预算以及 Jain 公平指数。通过引入实现余量（implementation margins）和停机惩罚，避免了成为“玩具示例”。
  • 边缘情况处理: 评估器显式处理了非有限功率值，将功率限制在最小/最大范围内，并对无效的用户/MCS ID 进行惩罚，展示了生产级的健壮性。
• 经济价值: 高。这些任务代表了价值数十亿美元的优化问题。高效的链路调度可提高电信频谱效率，而优化的电池调度可直接降低商业电费（需量电费管理）并延长硬件寿命（退化建模）。

4. 质量保证

• 验证与测试:
  • frontier_eval 集成: 是。
  • task_name: CommunicationEngineering/AdaptiveLinkScheduling, PowerSystems/MicrogridBatteryDispatch, 以及 Robotics/AGVWarehouseRouting (通过 task=unified task.benchmark=... 调用)。
  • 运行与依赖: .md 文件提供了确切的 python -m frontier_eval 命令。requirements.txt 正确指出所提供的评估器仅需 Python 标准库。
• 文档质量: 高。文档采用双语编写，并提供了清晰的 API 契约（输入/输出字典）。在可见的 diff 中未检测到明显的语法错误。
• 组织结构: 逻辑清晰且一致。遵循现有的仓库层级结构：benchmarks/{领域}/{任务名}/{scripts|verification|frontier_eval}。

5. 安全与隐私检查

• 敏感文件: 未发现异常。未发现 .env、API 密钥或 IDE 特定元数据（.vscode）。
• 绝对路径: 未检测到。评估器使用 Path(candidate_path).resolve() 和相对路径进行文件操作。