Continue Work in Progress

[skyelaird]

…idation pass rate

Summary

Investigated Phase 5 reliability calculations showing 0% values. Determined the reliability formula is correct but reveals systematic signal power deficit of ~6 dB.

Changes

• Added optional debug logging to _calc_reliability() for troubleshooting
• Added optional debug logging to _compute_signal() for loss analysis
• Created RELIABILITY_INVESTIGATION.md with detailed findings
• Updated validation_results.json with latest test outcomes

Key Findings

✅ Reliability calculation formula is correct and matches FORTRAN RELBIL.FOR ✅ 72.2% pass rate on reference VOACAP validation (test_voacap_reference.py) ✅ 81.2% pass rate on functional validation (validate_predictions.py) ⚠️ Signal power ~6 dB too low (within documented tolerance for nighttime) ⚠️ Reliability values systematically 25-30% lower than reference

Root Cause

Per ABSORPTION_BUG_ANALYSIS.md, nighttime signal errors of 2-7 dB are expected. Our 6 dB deficit is within this range. Remaining issues require investigation of:

• Deviation term calculations
• XNSQ calculation for D-E modes
• Mode selection logic

Validation Results

• test_voacap_reference.py: 156/216 pass (72.2%)
• validate_predictions.py: 13/16 pass (81.2%)
• Most predictions show non-zero reliability (30-50%)
• Edge cases with extreme SNR values identified

Next Steps

Per NEXT_STEPS.md:
✅ Week 1 Milestone: Predictions show >0% reliability - ACHIEVED ✅ Week 1 Milestone: One path validates - EXCEEDED (72-81%) → Week 2: Continue signal power investigation (deviation term, XNSQ) → Week 3-4: Expand validation test suite, target >80% pass rate

Refs: NEXT_STEPS.md, ABSORPTION_BUG_ANALYSIS.md, FORTRAN_ANALYSIS_AND_RECOMMENDATIONS.md