Previous baselines were from old V15 study, not from M1_Tensor best design. Updated all 9 design variables with correct values from model introspection. Baseline Corrections (from expression export): - lateral_inner_angle: 26.79° → 30.18° (at upper bound) - lateral_outer_angle: 14.64° → 15.09° - lateral_outer_pivot: 5.5mm → 6.036mm (0.4 × 15.09°) - lateral_inner_pivot: 10.07mm → 12.072mm (0.4 × 30.18°) - lateral_middle_pivot: 20.73mm → 14.0mm (lower than expected) - lateral_closeness: 11.02mm → 7.89mm - whiffle_min: 40.55mm → 56.7mm - inner_circular_rib_dia: 534.00mm → 537.86mm (fixed parameter) Bound Adjustments: - lateral_inner_pivot max: 11.0 → 13.0mm (to accommodate baseline 12.072) - lateral_closeness min: 9.5 → 5.0mm (to accommodate baseline 7.89) Root Cause: - NX introspection failed (NX not running) - Config was created with V15 study baselines as placeholders - Actual model values now applied from user-provided expression export Files Updated: - optimization_config.json: All baselines corrected - README.md: Design variable table updated - STUDY_REPORT.md: Baseline values corrected Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
4.1 KiB
4.1 KiB
SAT3_Trajectory - Study Report
Status: pending optimization Optimization Started: pending Optimization Completed: pending Total Trials: pending
Optimization Summary
| Metric | Value |
|---|---|
| Algorithm | TPE (Tree-Parzen Estimator) |
| Design Variables | 9 (enabled) |
| Total Trials | pending |
| Successful FEA | pending |
| Failed FEA | pending |
| Best Trial Number | pending |
| Best Weighted Sum | pending nm |
Best Design
Objectives
| Objective | Best Value | Baseline | Improvement |
|---|---|---|---|
| total_filtered_rms_nm (PRIMARY) | pending nm | 4.30 nm | pending % |
| coma_rms_nm (logged) | pending nm | 9.16 nm | pending % |
| astigmatism_rms_nm (logged) | pending nm | 6.55 nm | pending % |
| trefoil_rms_nm (logged) | pending nm | 6.44 nm | pending % |
| spherical_rms_nm (logged) | pending nm | 10.51 nm | pending % |
| linear_fit_r2 | pending | 1.0000 | pending |
| mass_kg | pending kg | pending kg | pending % |
Design Parameters
| Parameter | Best Value | Baseline | Delta |
|---|---|---|---|
| lateral_inner_angle | pending deg | 30.18 deg | pending |
| lateral_outer_angle | pending deg | 15.09 deg | pending |
| lateral_outer_pivot | pending mm | 6.036 mm | pending |
| lateral_inner_pivot | pending mm | 12.072 mm | pending |
| lateral_middle_pivot | pending mm | 14.0 mm | pending |
| lateral_closeness | pending mm | 7.89 mm | pending |
| whiffle_min | pending mm | 56.7 mm | pending |
| whiffle_outer_to_vertical | pending deg | 75.67 deg | pending |
| whiffle_triangle_closeness | pending mm | 60.00 mm | pending |
Fixed parameters (not optimized):
- blank_backface_angle = 4.00 deg
- inner_circular_rib_dia = 537.86 mm
Mode-Specific Analysis
Which Modes Improved Most?
Analysis pending - shows which aberration types benefited from optimization
Sensitivity Matrix
Analysis pending - shows which modes respond to axial vs lateral loads
R² Validation
Analysis pending - confirms physics model held throughout optimization
Convergence Analysis
Total Filtered RMS vs Trial
Plot pending
Mode-Specific RMS vs Trial
Plot pending - overlay coma, astigmatism, trefoil, spherical
Parameter Evolution
Plot pending - shows how design variables evolved
Key Findings
1. Trajectory Method Validation
Analysis pending
- Did R² stay > 0.95 throughout?
- Were any designs nonlinear?
- Did the physics model hold?
2. Mode-Specific Insights
Analysis pending
- Which modes improved most?
- Which modes are dominant now?
- Does coma reduction correlate with lateral support changes?
3. Comparison with V15 NSGA-II
Analysis pending
- How does best trajectory result compare to V15 Pareto front?
- Is trajectory method competitive?
- What insights does trajectory provide that discrete WFE doesn't?
4. Lateral vs Axial Sensitivity
Analysis pending
- Which parameters affect which modes?
- Are lateral supports primarily controlling coma (as predicted)?
- Are axial supports (whiffle tree) controlling spherical?
Recommendations
Next Steps
Analysis pending - based on results, suggest:
- Whether to proceed with SAT trajectory optimization
- Which modes need further attention
- Whether to refine parameter bounds
- Additional angles to include in trajectory
Parameter Bounds Refinement
Analysis pending
- Did any parameters hit bounds?
- Should ranges be expanded or narrowed?
Future Studies
Analysis pending
- SAT3_Trajectory_SAT (100 FEA + 10K surrogate)
- Multi-objective trajectory (optimize modes separately)
- Trajectory + mass trade-off
Files Generated
3_results/study.db- Optuna database with all trials3_results/optimization.log- Full execution log2_iterations/iter_XXXX/- FEA results for each trial- (Trajectory analysis plots - TBD)
Lessons Learned
To be filled after optimization completes
Report template created: 2026-01-29 To be updated after optimization completes