Capstone Thread: A Defensible Algorithm Choice

Produce a short decision record for a capstone component that searches, schedules, aggregates, or allocates resources. Use the lab scheduler first if the project is not yet defined.

State input/output, endpoint and duplicate conventions, constraints, and the objective. Present a simple oracle, the chosen algorithm, a correctness argument, a termination measure, and a cost model including setup and returned output. Attach finite test evidence and a minimized counterexample to a rejected approach.

For the scheduler, explain why count and value require different proofs, and why a valid value without a feasible original-index witness is an incomplete result. For a latency-sensitive component, distinguish amortized work from a per-operation deadline.

Finish with a changed assumption (closed endpoints, signed values, huge integers, or a new objective) and identify which part of the reasoning must be redone. Use AD4–AD7 in the assessment rubric. A plausible design narrative without executable evidence and a defensible model is not enough.