Skip to main content

Instructor Guide and Review Status

Status: authored reference candidate with executable verification. Independent subject review, grading agreement and learner pilots remain pending. Keep these statuses distinct from successful code execution or website rendering.

Subject-review focus​

Review all 20 concepts and the certificate lesson, especially: edge identity and zero weights; iterative DFS events; Kosaraju pass orientation and Tarjan's active stack; blocked versus cyclic tasks; scoped negative-cycle effects; snapshot versus in-place Bellman–Ford invariants; Floyd–Warshall initialization and path algebra; tied MST exchange rules; second-best definitions; residual pairing; flow/cut certificates; matching integrality and scheduling resource assumptions.

Review reference implementations and independent oracles separately. Validate that parent edges form finite routes, component partitions are maximal, forests span original components, flows conserve, cuts use outgoing original capacities, and matchings respect endpoint uniqueness. A correct scalar is insufficient.

Executable and grading review​

Extract the seven-file package and run reference checks and reports. Confirm the untouched starter fails. Introduce a wrong transpose, global negative-cycle classification, missing reverse residual update or infeasible witness and verify rejection; restore the reference afterward. Review source and proofs for bounds, since finite tests cannot establish asymptotic behavior or independent authorship.

Use the per-competency rubric on two independently scored sample submissions. Record disagreements over tied witnesses, incomplete proofs and resource costs. Revise ambiguous criteria rather than averaging a missing skill into a passing total.

Pilot protocol​

Recruit learners who pass the diagnostic, including someone without prior graph algorithm fluency. Record actual hours by block, setup trouble, misunderstood contracts, assistance, common failures and remediation. The 85–120 hour range is an author estimate, not observed effort. After 7–14 days, give an unseen graph with changed weights, direction or capacities and score transfer independently.

EvidenceRecord
Author revision2026-09-18; findings and executed checks in repository authoring review
Independent subject reviewPending: reviewer, scope, findings and resolutions
Cross-platform executionPending beyond the environment recorded in authoring notes
Scoring agreementPending: independent samples and disagreements
Learner pilotPending: prerequisites, hours, errors and resulting revisions
Delayed transferPending: fresh task, date and per-competency results

Do not promote a semester quality status or promise a professional percentile from this revision alone. Continue authoring while arranging human validation, then revise the candidate pattern using that evidence.