Adam DeJans Jr.

An exhaustive Pareto-frontier solution to the Advanced Website Design challenge showing that five nondominated cost-value tradeoffs are missed by a simple weighted-sum objective.

A pair-swap simulated annealing solution to the 210-person Balanced Assignment challenge that reaches a mathematically derived lower bound of 3,370, turning a heuristic result into a proven global optimum.

A regret-aware assignment heuristic for the April 2025 Case Assignments challenge, validated against HiGHS and exhaustive enumeration. A locally attractive greedy move consumes the only future option and makes the sequence infeasible.

A compact integer linear programming formulation for the generic change-making problem, with the 123-cent example solved to a six-coin optimum by HiGHS.

A binary shift-scheduling model for the April 2020 Doctor Planning challenge, covering all 21 weekly shifts while respecting availability, night-shift recovery, weekend, and night-limit rules.

A validated 280-move constraint-guided depth-first search for the 24×24 Grid of Bulbs instance, contrasted with a greedy next-best-action policy that fails to finish.

A Lagrangian-relaxation solution to the Holiday Gift Assignment challenge that decomposes the budgeted assignment by person, produces a tight upper bound, and certifies the optimal happiness of 24.

A binary MILP for the June 2019 Map Coloring with Violations challenge, using soft-edge penalty variables to prove a minimum violation cost of $257.

Differential evolution finds a box with 9,000 cubic inches of volume and only 99.058 inches of strap measure; an independent analytical derivation then proves the continuous optimum.

A four-variable integer programming solution to the September 2024 Rental Boats challenge, solved to proven optimality with HiGHS.

An integer portfolio-allocation model for the July 2024 Smart Investment challenge, with a $4,650 maximum projected one-year gain proven by HiGHS.

In large organizations, the decision that appears on a slide is often not the decision anyone can actually make. Real decision sets are shaped by timing, ownership, infrastructure, and operational constraints long before an optimizer sees them.

A practical example of using simulation data to discover exploitable structure in a noisy tuning problem, replacing repeated global searches with a much faster local search while preserving most of the value.

A compact time-indexed mixed-integer linear programming solution to the July 2026 Brick Factory challenge, solved and verified with HiGHS.

When a business outcome is generated by a policy operating under uncertainty, the most useful optimization target is often downstream economic performance rather than upstream forecast accuracy. Simulation makes it practical to tune the policy directly.

A probabilistic forecast creates value only when the decision layer can consume the distribution. Collapsing uncertainty to a single number often throws away the asymmetry that drives the economics of the decision.

Solution to A Brick Factory · BitBros

Solution to Advanced Website Design · BitBros

Solution to Case Assignments · BitBros

Solution to Mr Bates vs The Post Office · BitBros

Solution to Rental Boats · BitBros

Solution to Smart Investment · BitBros

Solution to Grid of Bulbs · BitBros

Solution to Holiday Gift Assignment · BitBros

Solution to Doctor Planning · BitBros

Solution to Map Coloring with Violations · BitBros

Solution to Balanced Assignment · BitBros

Solution to Change-Making Decision · BitBros