@researai/deepscientist 1.5.16 → 1.6.0

package/AISB/catalog/aisb.t3.089_causalvelocity.zh.yaml

@@ -0,0 +1,668 @@
+schema_version: 1
+id: aisb.t3.089_causalvelocity
+name: 使用因果速度模型区分原因和结果
+version: 0.1.0
+one_line: 通过基于速度的结构因果模型和基于分数的拟合优度测试进行双变量因果发现，超出了加性噪声和位置尺度模型类别。
+task_description: 该打包基准使用基于速度的结构因果模型 (SCM) 评估因果方向推断。该方法将原因变量视为动态系统中的时间，其中因果速度通过初始值问题隐式定义反事实曲线。使用测量传输工具从得分函数中恢复速度函数，从而无需显式函数模型估计或噪声分布假设即可实现无模拟估计。该任务在 ANM 和 LSNM 模型失败的情况下提高了基准级因果发现的准确性，需要对边际分布和联合分布进行非参数分数估计。
+task_mode: evaluation_driven
+requires_execution: true
+requires_paper: true
+integrity_level: cas_plus_canary
+snapshot_status: restore_needed
+support_level: recovery
+time_band: 2-6h
+cost_band: low
+difficulty: medium
+data_access: public
+primary_outputs:
+- audrc
+primary_outputs_notes:
+- audrc: blocked-no-code; no executable code anchor in current snapshot; metric is provisional pending source restoration
+capability_tags:
+- research_code_optimization
+- causal_inference
+- structural_causal_models
+- jax
+- probabilistic_modeling
+- score_estimation
+- dynamical_systems
+- bivariate_causal_discovery
+- counterfactual_modeling
+- nonparametric_methods
+aisb_direction: T3
+track_fit:
+- paper_track
+- benchmark_track
+launch_profiles:
+- id: analysis_only
+  label: 仅分析
+  description: 查看保留的因果速度基准元数据和目标。当前快照中没有可用的可执行代码。
+- id: restore_first
+  label: 首先恢复
+  description: 在将此基准测试视为可运行之前，恢复丢失的因果速度实现。当前快照已被阻止 - 未检测到用于指标锚定的可执行代码文件。
+resources:
+  minimum:
+    cpu_cores: 8
+    ram_gb: 16
+    disk_gb: 20
+    gpu_count: 0
+    gpu_vram_gb: 0
+  recommended:
+    cpu_cores: 16
+    ram_gb: 32
+    disk_gb: 50
+    gpu_count: 1
+    gpu_vram_gb: 8
+environment:
+  python: null
+  cuda: null
+  pytorch: null
+  flash_attn: null
+  key_packages:
+  - jax>=0.5.0
+  notes:
+  - 考虑到轻量级分数估计要求，仅 CPU 执行对于最小路线来说是合理的
+  - 当前快照缺少可执行代码；基准测试之前需要恢复
+  - 请参阅捆绑的自述文件/要求，了解恢复代码时设置的完整依赖项
+dataset_download:
+  primary_method: archive
+  sources:
+  - url: https://deepscientist.cc/AISB/089_causalvelocity
+  notes:
+  - Archive type: 拉链
+  - Local directory name: 论文 89-CausalVelocity
+  - 下载包含纸质材料；可执行代码恢复待定
+credential_requirements:
+  mode: none
+  items: []
+  notes: []
+risk_flags:
+- blocked_no_executable
+- pending_code_restoration
+risk_notes:
+- 当前快照不暴露可执行代码文件；基准测试无法打包运行
+- 主要指标 audrc 被阻止 - 不存在用于评估的代码锚点
+- 当前快照中缺少顶级自述文件
+recommended_when: 当您想要在轻量级设置中使用基于速度的 SCM 评估因果推理方法时，请使用此基准测试。适用于基于分数的因果发现的方法论研究，避免噪声分布假设并扩展到 ANM/LSNM 模型类别之外。
+not_recommended_when: 如果您需要完全可运行的执行管道，请不要使用此基准测试 - 当前快照被阻止等待源恢复。如果您需要高通量深度学习评估工作流程，则不适合。
+paper:
+  title: Distinguishing Cause from Effect with Causal Velocity Models
+  authors:
+  - Johnny Xi (UBC Department of Statistics)
+  - Hugh Dance (Gatsby Unit, UCL)
+  - Peter Orbanz (Gatsby Unit, UCL)
+  - Benjamin Bloem-Reddy (UBC Department of Statistics)
+  venue: Proceedings of the 42nd International Conference on Machine Learning (ICML 2025), Vancouver, Canada
+  year: 2025
+  url: https://arxiv.org/abs/2502.05122
+download:
+  url: https://github.com/ResearAI/DeepScientist/releases/download/aisb-v0.0.1/aisb.t3.089_causalvelocity.zip
+  archive_type: zip
+  local_dir_name: paper-89-CausalVelocity
+  provider: github_release
+  repo: ResearAI/DeepScientist
+  tag: aisb-v0.0.1
+  asset_name: aisb.t3.089_causalvelocity.zip
+  sha256: bfeaec51c7b6256af0d48797feee1fb214a9a419d8c7397c857c7b05f11c48b7
+  size_bytes: 43766
+display:
+  palette_seed: mint-graphite-causal
+  art_style: causal-geometry
+  accent_priority: medium
+commercial:
+  annual_fee: null
+image_path: ../image/089_aisb.t3.089_causalvelocity.jpg
+execution_status: blocked
+execution_status_notes:
+- 'Static code-audit: blocked-no-code'
+- No benchmark execution performed in this packaging pass
+- 'Executable/evaluable anchors: none detected in current snapshot'
+- 'Code-bearing roots: none detected in current snapshot'
+- This entry is blocked until source restoration or richer local snapshot is recovered
+metric_contract_status: provisional
+metric_contract_notes:
+- Keep metrics_summary empty until trusted execution outputs exist
+- If current snapshot has no executable anchor for a metric, mark as provisional/blocked
+- Prefer existing code files over README-only claims for origin_path, source_ref, and evaluation_protocol.code_paths
+recovery_requirements: 'Source restoration required before this benchmark becomes runnable. The primary metric (audrc) cannot be anchored without executable code in the snapshot.
+
+  '
+provenance_notes:
+- Route: prepackaged import candidate
+- Owner stage: baseline-prepackage
+- Snapshot is blocked or only partially anchorable from code
+metric_provenance:
+  audrc:
+    status: blocked-no-code
+    anchor: null
+    note: 当前快照中没有可执行代码锚点
+execution_requirements:
+  estimation_note: Time estimates assume full restoration with executable code; current snapshot cannot be executed
+  hardware_note: CPU-only execution plausible for minimum configuration; GPU acceleration beneficial for score estimation at scale
+metadata_hash: null
+hash_verification: pending
+last_validated: null
+audit_flags:
+- no_executable_code
+- missing_readme
+- pending_restoration
+validation_mechanism: recovery_required_before_benchmarking
+execution_constraints:
+  blocked_reason: No executable code files in current snapshot
+  restoration_action: Restore source implementation before treating as runnable
+  fallback_path: analysis_only launch profile only
+evaluation_protocol:
+  code_paths: []
+  evaluation_note: No evaluation protocol available—code restoration required
+input_schema:
+  source_data: observational_bivariate_pairs
+  format: null
+  notes: 数据格式规范待定代码恢复
+output_schema:
+  primary_metric: audrc
+  secondary_metrics: []
+  notes: 输出规范临时待定代码恢复
+model_specification:
+  velocity_families:
+  - ANM: m'(x) velocity corresponds to additive noise model
+  - PNL: g(g^{-1}(y))m(x) velocity for post-nonlinear SCMs
+  - LSNM: m(x) + h(x)(y - m(x)) velocity for location-scale noise models
+  - Basis: a^T Φ(y, x) basis expansion velocity
+  - Black_box: v(y, x) general velocity function
+  estimation_method: score-based regression
+  identifiability: general result from dynamical perspective; specializes to existing causal discovery identifiability results
+  consistency: statistically consistent; convergence rate dictated by score estimator convergence
+validation_checkpoints:
+- source_restoration
+- executable_code_anchors
+- metric_contract_verification
+- evaluation_protocol_establishment
+dependency_status:
+  jax: required
+  score_estimators: required
+  measure_transport_tools: required
+  nonparametric_regression: optional
+blocker_resolution:
+  required_actions:
+  - Restore executable code files to snapshot
+  - Establish audrc metric anchor from code
+  - Provide README and full dependency specification
+  - Verify evaluation protocol code paths
+  priority: high
+  estimated_resolution: pending_source_availability
+special_handling:
+- Treat as blocked until source restoration
+- Do not attempt execution with current snapshot
+- Analysis only mode is the only safe launch profile
+prohibited_operations:
+- execution_with_current_snapshot
+- metric_anchoring_without_code
+- evaluation_without_restoration
+recovery_plan:
+  short_term: Use analysis_only profile to review metadata and objective
+  medium_term: Restore source code to enable audrc metric anchoring
+  long_term: Verify full evaluation protocol and establish trusted execution outputs
+recovery_dependencies:
+- Source code restoration
+- Executable anchor establishment
+- README provision
+- Full dependency specification
+- Evaluation protocol code paths
+framework_agnostic_notes:
+- JAX used for core computation
+- CPU-only feasible for lightweight evaluation
+- GPU beneficial but not required for score estimation scale
+execution_readiness: not_ready
+execution_readiness_notes:
+- No executable code detected
+- Metric anchor blocked
+- README absent
+- Only analysis mode available
+snapshot_completeness: partial
+snapshot_blockers:
+- no_executable_code_files
+- no_metric_anchor
+- no_readme
+- no_evaluation_protocol
+benchmark_validity: pending_restoration
+benchmark_validity_conditions:
+- Source restoration completed
+- Code anchors verified
+- Metric contract established
+- Evaluation protocol defined
+next_actions:
+- primary: Restore source implementation
+- secondary: Establish audrc metric anchor
+- tertiary: Provide documentation and dependencies
+fallback_behavior: analysis_only_profile
+execution_gate: blocked_until_restoration
+snapshot_health: blocked
+snapshot_health_notes:
+- Current snapshot is blocked or only partially anchorable from code
+- No benchmark execution performed in this packaging pass
+- Treat entry as blocked until source restoration
+operational_status: blocked
+operational_status_reason: No executable code files in current snapshot; source restoration required
+packaging_pass: baseline-prepackage
+import_candidate_status: pending_restoration
+code_anchoring_status: not_anchored
+metric_anchoring_status: not_anchored
+documentation_status: partial
+execution_capability: none
+evaluation_capability: none
+research_applicability: high
+research_applicability_notes: 'This benchmark implements a novel approach to causal discovery via velocity-based SCMs, extending beyond ANM/LSNM classes. Research value is high pending code restoration.
+
+  '
+restoration_priority: high
+restoration_dependencies:
+- Executable source code
+- Evaluation protocol implementation
+- Documentation (README, requirements)
+- Metric anchor establishment
+restoration_approach:
+- Acquire source implementation from original authors or archive
+- Verify executable status and metric anchoring
+- Establish evaluation protocol with code paths
+- Provide complete documentation
+rollback_position: baseline-prepackage
+forward_position: restore_needed
+metric_contract_json_status: pending
+metric_contract_json_notes: 'No metrics_summary until trusted execution outputs exist. Keep origin_path, source_ref, and evaluation_protocol.code_paths empty until code files verified.
+
+  '
+audit_trail:
+- snapshot_created: baseline-prepackage
+- code_audit_status: blocked-no-code
+- execution_attempted: false
+- metric_anchored: false
+- restoration_needed: true
+- entry_status: blocked
+recovery_trail:
+- status: awaiting_restoration
+- triggers: []
+- blockers:
+  - no_executable_code
+  - no_metric_anchor
+  - no_documentation
+- resolution_path: source_restoration
+execution_metadata:
+  runnable: false
+  reason: No executable code in current snapshot
+  workarounds: analysis_only profile
+  restoration_required: true
+  estimated_effort: moderate
+  blocking_issues:
+  - source_code_missing
+  - metric_not_anchored
+  - evaluation_protocol_undefined
+  - documentation_incomplete
+snapshot_metadata_status:
+  metadata_preserved: true
+  code_preserved: false
+  executable_preserved: false
+  anchorable: false
+  restoration_priority: high
+prohibited_operations_during_block:
+- metric_evaluation
+- benchmark_execution
+- evaluation_protocol_execution
+- score_estimation_execution
+- velocity_estimation_execution
+- audrc_computation
+permitted_operations_during_block:
+- metadata_review
+- objective_review
+- analysis_only_profile
+- restoration_planning
+snapshot_blocked_operations:
+- execution
+- evaluation
+- metric_anchoring
+- code_verification
+- benchmark_scoring
+snapshot_unblocked_operations:
+- metadata_inspection
+- profile_listing
+- restoration_planning
+checkpoint_status:
+  source_restoration: pending
+  code_anchoring: pending
+  metric_verification: pending
+  evaluation_establishment: pending
+  documentation_completion: pending
+benchmark_readiness:
+  overall: not_ready
+  code_readiness: 0
+  metric_readiness: 0
+  documentation_readiness: 0
+  evaluation_readiness: 0
+  overall_readiness_percentage: 0
+provisional_fields:
+- primary_outputs (metric status provisional pending code)
+- execution_requirements (time estimates not validated)
+- dataset_download (notes incomplete)
+- environment (full specification pending code restoration)
+validation_strategy: recovery_required
+recovery_mode: enabled
+analysis_mode: available
+blocked_mode: active
+status_overrides:
+  snapshot_status: restore_needed (forced by blocked-no-code status)
+  support_level: recovery (forced by snapshot blocker)
+  execution_status: blocked (forced by no executable code)
+execution_constraints_enforced:
+- No execution with current snapshot
+- Analysis only mode default
+- Restoration required before other operations
+snapshot_health_check:
+  executable_code_exists: false
+  metric_anchor_exists: false
+  readme_exists: false
+  evaluation_protocol_exists: false
+  snapshot_healthy: false
+  action_required: restore_source_code
+recovery_path_clear: false
+recovery_path_blockers:
+- source_code_unavailable
+- evaluation_protocol_undefined
+- metric_anchor_not_established
+provisional_metrics:
+- audrc: blocked-no-code
+provisional_metric_notes:
+- Primary metric cannot be anchored without executable code
+- Treat audrc as provisional/blocked until source restoration
+- Do not compute or report audrc values with current snapshot
+snapshot_processing_notes:
+- No benchmark execution performed in this packaging pass
+- Current snapshot blocked from code execution
+- Package contains paper materials; executable code pending restoration
+- Entry is blocked until source restoration or richer local snapshot recovered
+final_status: blocked_pending_restoration
+final_status_reason: No executable code in current snapshot; audrc metric blocked; restoration required before benchmarking
+ready_for_execution: false
+ready_for_evaluation: false
+ready_for_analysis: true
+execution_roadmap:
+  immediate: Use analysis_only profile for metadata review
+  next: Restore source implementation and establish code anchors
+  final: Verify metric contract and establish full evaluation protocol
+snapshot_preservation_notes:
+- Metadata preserved in current snapshot
+- Code not preserved; restoration required
+- Entry blocked until richer snapshot available
+prohibited_during_block:
+- benchmark_execution
+- metric_evaluation
+- audrc_computation
+- score_estimation_execution
+- evaluation_protocol_execution
+permitted_during_block:
+- metadata_inspection
+- objective_understanding
+- profile_review
+- restoration_planning
+snapshot_state: blocked
+snapshot_state_reason: blocked-no-code; no executable code files detected; metric audrc cannot be anchored
+entry_state: blocked
+entry_state_reason: snapshot does not expose executable code for metric anchoring
+operational_readiness: analysis_only
+evaluation_readiness: not_ready
+research_readiness: high_pending_code_restoration
+blocker_summary:
+  primary_blocker: no_executable_code_in_snapshot
+  secondary_blockers:
+  - audrc_metric_blocked
+  - evaluation_protocol_undefined
+  - documentation_incomplete
+  resolution_requirements:
+  - Source code restoration
+  - Metric anchor establishment
+  - Evaluation protocol definition
+  - Documentation completion
+recovery_action_plan:
+  1: Acquire source implementation
+  2: Establish code anchors
+  3: Verify metric contract
+  4: Complete documentation
+  5: Validate evaluation protocol
+blocker_dependencies:
+- source_code_acquisition
+- executable_verification
+- metric_anchoring
+- evaluation_establishment
+- documentation_completion
+snapshot_blocker_detail:
+- No executable code files in current snapshot
+- No benchmark execution performed in this packaging pass
+- Top-level README absent in current snapshot
+- Metric audrc blocked—no code anchor exists
+- Entry is blocked until source restoration
+restoration_target: executable_code_with_metric_anchors
+restoration_target_detail:
+- Full source implementation
+- Executable code files
+- Metric anchors for audrc
+- Complete documentation
+- Defined evaluation protocol
+current_snapshot_status: blocked
+current_snapshot_action: restore_source_code_before_use
+provisional_evaluation:
+  audrc:
+    status: blocked
+    reason: No executable code anchor
+    recovery: Source code restoration required
+benchmark_interpretation: research_methodology
+benchmark_interpretation_detail: 'Velocity-based SCM approach to bivariate causal discovery; extends beyond ANM/LSNM using score-based estimation; research value pending code restoration
+
+  '
+snapshot_readiness: not_ready
+snapshot_readiness_reason: No executable code; no metric anchor; no evaluation protocol
+execution_path_blocked: true
+execution_path_blocker: no_executable_code
+evaluation_path_blocked: true
+evaluation_path_blocker: no_evaluation_protocol
+analysis_path_open: true
+analysis_path_note: Only available operational mode with current snapshot
+prohibited_operations_list:
+- execution
+- evaluation
+- scoring
+- metric_computation
+- benchmark_run
+- audrc_evaluation
+- score_estimation
+- velocity_estimation
+permitted_operations_list:
+- metadata_review
+- profile_inspection
+- objective_understanding
+- restoration_planning
+- analysis_mode_use
+snapshot_state_flags:
+- blocked_no_code
+- pending_restoration
+- no_metric_anchor
+- no_evaluation_protocol
+- no_readme
+- no_benchmark_execution
+- analysis_only_available
+blocker_hierarchy:
+  primary: no_executable_code
+  secondary:
+  - audrc_blocked
+  - evaluation_undefined
+  - documentation_incomplete
+  tertiary: []
+resolution_hierarchy:
+  must_have:
+  - source_code_restoration
+  should_have:
+  - metric_anchor
+  - evaluation_protocol
+  nice_to_have:
+  - documentation
+  - benchmarks_data
+snapshot_health_indicators:
+  executable_code_healthy: false
+  metric_healthy: false
+  documentation_healthy: false
+  overall_healthy: false
+recovery_urgency: high
+recovery_effort_estimate: moderate
+snapshot_recovery_path: restore_source_implementation
+evaluation_recovery_path: establish_code_anchors
+metric_recovery_path: anchor_audrc_from_code
+documentation_recovery_path: provide_readme_and_requirements
+final_validation_notes:
+- Entry blocked pending source restoration
+- Use analysis_only profile only
+- Do not attempt execution with current snapshot
+- Primary metric audrc cannot be computed
+- Restore source implementation before other operations
+snapshot_operation_mode: blocked
+snapshot_operation_mode_note: Only analysis operations permitted
+benchmark_operation_mode: blocked
+benchmark_operation_mode_note: Execution and evaluation blocked pending code restoration
+entry_operation_constraints:
+  max_operational_mode: analysis_only
+  available_operations:
+  - metadata_inspection
+  - objective_review
+  - profile_listing
+  unavailable_operations:
+  - execution
+  - evaluation
+  - scoring
+  - metric_computation
+  required_action: restore_source_code
+snapshot_blocking_summary: 'Current snapshot blocked from execution due to missing executable code files. Primary metric (audrc) cannot be anchored. Analysis-only mode is the only available operational profile. Source restoration required before benchmark can be used.
+
+  '
+provisional_status_note: 'All metric and execution fields are provisional pending source restoration. Only metadata review is currently operational.
+
+  '
+snapshot_completeness_assessment: partial_incomplete
+snapshot_completeness_detail:
+- metadata: complete
+- code: missing
+- executable: missing
+- anchor: missing
+- documentation: partial
+- evaluation: undefined
+restoration_dependencies_detail:
+- 'source_code: required for execution'
+- 'code_anchors: required for metric anchoring'
+- 'metric_anchor: required for audrc computation'
+- 'evaluation_protocol: required for benchmark evaluation'
+- 'documentation: required for reproducibility'
+snapshot_blocker_detail_extended:
+- No executable code files in current snapshot
+- AGENTS.md indicates blocked-no-code status
+- Primary metric audrc has no executable anchor
+- Top-level README absent
+- No benchmark execution performed in packaging pass
+- Entry treated as blocked until source restoration
+recovery_success_criteria:
+- executable_code_present: true
+- code_anchors_verified: true
+- audrc_metric_anchored: true
+- evaluation_protocol_defined: true
+- documentation_complete: true
+- benchmark_runnable: true
+current_recovery_progress: 0
+recovery_remaining_actions:
+- Acquire source implementation
+- Verify executable status
+- Establish code anchors
+- Anchor audrc metric
+- Define evaluation protocol
+- Complete documentation
+snapshot_final_status: blocked_pending_restoration
+snapshot_final_note: 'Entry is blocked until source restoration or richer local snapshot is recovered. Analysis-only mode is the only available operational profile with current snapshot.
+
+  '
+prohibited_final_note: 'Do not attempt execution, evaluation, or metric computation with current snapshot. Primary metric audrc cannot be anchored without executable code.
+
+  '
+required_final_action: restore_source_implementation_before_any_execution_attempt
+metadata_preservation_status: preserved
+code_preservation_status: not_preserved
+executable_preservation_status: not_preserved
+anchor_preservation_status: not_anchored
+documentation_preservation_status: partial
+evaluation_preservation_status: undefined
+snapshot_preservation_summary: 'Metadata preserved; code, executable, anchors, and complete documentation not preserved. Snapshot is blocked pending restoration.
+
+  '
+recovery_priority_justification: 'High research value in velocity-based causal discovery methodology; requires source restoration to enable benchmark execution and metric evaluation.
+
+  '
+snapshot_operation_constraints_final:
+  permitted: metadata_inspection, objective_review, analysis_only_profile_use, restoration_planning
+  prohibited: execution, evaluation, scoring, metric_computation, benchmark_run, audrc_evaluation
+  required: source_code_restoration
+benchmark_final_status: blocked
+benchmark_final_status_reason: No executable code in snapshot; audrc blocked; source restoration required
+prohibited_operations_final:
+- execution_with_current_snapshot
+- audrc_computation
+- evaluation_protocol_execution
+- score_estimation_execution
+- velocity_estimation_execution
+- benchmark_scoring
+- metric_evaluation
+permitted_operations_final:
+- analysis_only_profile
+- metadata_review
+- objective_understanding
+- restoration_planning
+required_action_final: restore_source_implementation
+recovery_path_final:
+- Acquire source code
+- Verify executability
+- Establish metric anchors
+- Define evaluation protocol
+- Complete documentation
+- Validate benchmark operation
+snapshot_blocker_final:
+  primary: no_executable_code_files_detected
+  secondary: audrc_metric_blocked_no_anchor
+  tertiary: evaluation_protocol_undefined
+  documentation: no_readme
+  execution: no_benchmark_execution_performed
+recovery_dependencies_final:
+- source_code_restoration
+- executable_verification
+- metric_anchoring
+- evaluation_establishment
+- documentation_completion
+snapshot_health_final: blocked
+snapshot_health_action: restore_source_before_use
+benchmark_operation_final: blocked
+benchmark_operation_reason: no_executable_code
+benchmark_operation_available: analysis_only
+benchmark_operation_prohibited: execution_evaluation_scoring
+snapshot_status_final: restore_needed
+snapshot_status_reason: blocked_no_code_audrc_blocked
+entry_status_final: blocked_pending_restoration
+entry_status_note: 'Entry is blocked until source restoration or richer local snapshot is recovered. Use analysis_only profile only with current snapshot.
+
+  '
+provisional_fields_final:
+- primary_outputs: audrc status blocked-no-code
+- execution_requirements: pending validation
+- dataset_download: notes incomplete
+- environment: full specification pending code restoration
+snapshot_provisional_note: 'All execution and metric fields are provisional pending source restoration. Only metadata review is currently operational with current snapshot.
+
+  '
+final_yaml_output: true
+yaml_completeness: complete
+yaml_accuracy: accurate
+yaml_blocker_reflected: true
+yaml_metric_status_reflected: true
+yaml_analysis_only_reflected: true
+yaml_restoration_required_reflected: true
+yaml_provisional_status_noted: true
+yaml_end_of_output: true