How riadhossain56/cse307-os-term-paper works
This is a research simulator for comparing classical and learned page-replacement policies under a synthetic workload shift. The main workflow generates a trace, runs policies through the paging simulator, and evaluates hit ratios; the learned-policy branch trains on warmup telemetry and scores resident pages. An optional explanation branch checks eviction decisions against Belady’s MIN and reports confidence calibration. The experiment runner coordinates benchmarks and writes results. Relationships are limited to calls shown in excerpts or module references in the SOURCE INDEX; unsampled implementation details are not inferred.
Main entry points: simulator.py Paging Simulator · run_experiments.py Experiment Runner · learned_policy.py Learned Policy · workload.py Trace Generator
Components 9
Workload
- Trace Generator
src/workload.py
Replacement Policies
- Classical Policies
src/classical_policies.py - Learned Policy
src/learned_policy.py - NumPy Classifiers
src/learned_models.py
Simulation
- Paging Simulator
src/simulator.py - Explanation Engine
src/explanation_engine.py
Evaluation
- Experiment Runner
experiments/run_experiments.py - Benchmark Results
results
Other components
- Researcher
Connections 11
- Researcher → Experiment Runner: runs benchmarks
README.md - Experiment Runner → Trace Generator: generates trace
experiments/run_experiments.py - Experiment Runner → Paging Simulator: runs policies
experiments/run_experiments.py - Trace Generator → Paging Simulator: provides trace
experiments/run_experiments.py - Paging Simulator → Classical Policies: dispatches policy
src/simulator.py - Paging Simulator → Learned Policy: dispatches policy
src/simulator.py - Learned Policy → Classical Policies: inherits base
src/learned_policy.py - Learned Policy → NumPy Classifiers: scores candidates
src/learned_policy.py - Paging Simulator → Explanation Engine: records decisions
src/simulator.py - Paging Simulator → Explanation Engine: evaluates calibration
src/simulator.py - Experiment Runner → Benchmark Results: writes metrics
experiments/run_experiments.py