Closed-form and alternating models

PCA and RP-PCA use eigendecomposition. IPCA uses alternating least squares.

These models expose no neural epoch grid; structural output is reported at epoch = 0.

Neural checkpoints

CAE, SAE, and neural portfolio learners train for configured epochs or optimization steps.

Emit interval checkpoints or explicit checkpoint lists. Set default_checkpoint for prediction.

Phased SDF training

1. Unconditional SDF warm-up
2. Moment-network fitting
3. Conditional SDF refinement

For long SDF runs, use 4-8 explicit structural checkpoints. Beta-head checkpoints have a separate, usually smaller grid.

Reporting and evaluation

Models emit configured checkpoints and metadata. Diagnostics and case studies select, score, and compare them.

Use interval checkpoints for short neural runs and explicit checkpoints for long phased runs. Keep prediction-time defaults explicit.