Implementation and Transfer Studio
The required implementations are the eight APIs in five guided labs. This page organizes repair and depth work; it does not impose a second hidden implementation checklist. A failed concept requires targeted remediation, not an arbitrary quota of additional problems.
Required evidence
For each core API, provide a state card, recurrence proof, source-code explanation, tested boundary case, witness check where applicable and resource analysis. Preserve a meaningful bug and its smallest reproducer when one occurs. Do not invent errors to reach a mistake-log target. Record help and independent checks.
Choose one transfer after the core
| Extension | Deliverable | New proof obligation |
|---|---|---|
| Exact-fill knapsack | Impossible status plus a reconstructed exact-weight selection | Unreachable bases and legal transitions |
| Coin combinations versus sequences | Two counters with enumerated tiny objects | Disjoint decomposition and duplicate denomination policy |
| Grid obstacle-removal budget | Path or count with a token coordinate | Histories distinguished by remaining tokens |
| Tree distance rerooting | All-root distance sums | Child contribution exclusion and linear total work |
| Digit property with started flag | Count canonical numbers containing digit zero | Padding and integer-zero terminal policy |
| Linear-space edit value | Full-table comparison plus memory account | Preserving old diagonal; no claim of free alignment recovery |
Depth options beyond this core include affine alignment, Hirschberg reconstruction, optimal BSTs with unsuccessful queries, profile tilings and advanced split optimizations. Derive prerequisites and assumptions before implementation. Merely naming Knuth optimization or using a contest template is not acceptance evidence.
Completion
Use the competency rubric, then the independent assessment and delayed transfer. The capstone adds an explicit budget constraint to weighted scheduling and requires a new state and witness.