Every placement
changes what fits.

I compare a fast greedy choice, a bounded genetic search, and an exhaustive reference. Change the constraints and inspect what each method gives up.

Loading recorded results. The local Python app also supports fresh experiments.

measurement tables and raw data

01 / Change the problem

Model inputs, not production data

Random seedGenerations

Objective = admitted value − capacity cost

Select a method above to inspect its assignment.

Rejected requests

Request labels show demand / maximum delay. Bars show used / available capacity.

Best feasible objective

Genetic searchGreedyExhaustive optimum

A flat trace means this search stopped improving. It does not establish optimality; the exhaustive reference does, for this small model.

Inspect every request and its constraints

Request	Demand	Value	Max delay	Placement