Stateful strategic MILP
The stateful strategic model adds a risk state to each segment and carries it across periods. It is a linearized mixed-integer program solved by HiGHS through SciPy.
State model
Each segment starts with an observed or derived risk index. At the start of every period, risk grows by a declared annual amount. Selecting a treatment method applies that method’s risk reduction with a lower bound of zero. One method may be selected per segment and period. Annual budgets and crew-hour capacities are hard constraints.
The formulation is exact for this declared linear state. It is not a full nonlinear shrub/tree growth model, and it does not claim BC Hydro clearance compliance.
Objectives
The default objective minimizes length-weighted mean risk, with annual budget and crew-hour constraints. A cost weight, maximum mean risk, and solver time limit are explicit settings. Budget sweeps produce a non-dominated cost/risk frontier.
Replay
After the solve, the selected actions are replayed through the same linear equations. The plan records
the solver, status, objective, annual outcomes, actions, and replay_ok. A replay mismatch is
visible in the artifact rather than hidden.
Command-less package API
The demo builder calls:
from brushline.strategic.stateful import (
StatefulConfig,
StatefulMethod,
StatefulPeriod,
StatefulSegment,
build_stateful_plan,
)
Limits
The risk state is linear and generic.
Treatment productivity and cost are caller inputs.
Access, outages, environmental windows, and asset-owner standards are separate constraints.
The model is a planning prototype, not a safety or regulatory determination.