Instructor Guide and Review Status
Authored candidate, September 24, 2026. Independent mathematical/code review and learner pilots remain pending. Executable and rendered-site checks belong in the repository authoring review; they do not establish learner outcomes.
Workload
Plan 100–145 hours: lessons/diagnostic/targeted reading 32–45; labs 20–28; independent practice/revision 20–28; numerical experiments 6–10; assessment and remediation 10–16; project 12–18. At ten hours per week this is about 10–15 weeks, not seven days. These are unpiloted estimates; prerequisite repair and optional full-book or library work add time.
Teaching and review priorities
Teach exact models before tolerances, and spaces before numerical solver output. Review empty-set closure, row versus column information, rectangular coordinates, least-squares coefficient uniqueness, partial-pivot growth, the scalar field of diagonalization, power-iteration hypotheses, SPD versus PSD and SVD cutoff semantics. General SVD algorithms, Jordan computation and advanced sparse methods remain later depth.
Have a code reviewer inspect independent rank oracles and semantic certificates, then inject faults. Compare orthogonality and reconstruction separately. Tests must accept legitimate signs and eigenbasis choices. The kit's exact small inverse is a diagnostic exercise, not a production solve recommendation.
Pilot record
| Evidence | Status |
|---|---|
| Reviewer names, dates, disagreements and repairs | Pending |
| Learner arithmetic, proof and programming prerequisites | Pending |
| Actual hours and first blocked instruction per activity | Pending |
| Assistance and solution exposure | Pending |
| C1–C10 initial, remedial and changed-input scores | Pending |
| Delayed structural/numerical transfer after 7–14 days | Pending |
| Revisions justified by observed learner evidence | Pending |
Authored, executed, independently reviewed and learner-piloted are distinct statuses. No elite-percentile outcome follows from this packet or a self-score.