Skip to main content

Tiny Correctness

Purpose​

Validate that federated QC, KING, and LR outputs agree with a centralized PLINK baseline.

Data​

Generate synthetic data if missing:

python pipeline/simulation/simulated_data/generate_synthetic_data.py \
  --scale tiny \
  --partition-strategy even \
  --seed 42 \
  --output-dir experiments/correctness/tiny_even/data

Command​

python experiments/tools/generate_baseline.py \
  experiments/correctness/tiny_even/config.yaml

flwr run . local-simulation --stream --run-config \
  'simulation=true num-server-rounds=100 config_path="experiments/correctness/tiny_even/configs"'

python experiments/tools/evaluation/evaluate_all.py \
  experiments/correctness/tiny_even/results_2 \
  --baseline experiments/correctness/tiny_even/data/tiny/centralized_baseline \
  --king

Expected output​

  • Server and client logs under experiments/correctness/tiny_even/results_2/.
  • QC and LR agreement reports from the evaluation tools.
  • Successful runs should reach the done stage and meet the thresholds documented in experiments/correctness/tiny_even/README.md.