Optioneer's Algorithms

optioneer uses a variety of proprietary and industry standard algorithms, depending on the required use case there are two core algorithms in use for the majority of cases docid\ skr80mrm8eua2s mznxx4 (ea) nature inspired optimization that evolves solutions over generations through selection, mutation, and recombination, effectively tackling complex problems with user editable rules docid\ mc iqzkhf9iyf6rabdo25 (lcp) a simplified algorithm for initial "seeding" of results in the earliest stages of analysis we try to use the best algorithm for the best situation ultimately we believe the best results will be achieved by combining the strenghts of both methods our docid\ syz9arnkk8ipehnexea6c functionality is based on an lcp solution docid\ u m3xc 1mi7ywqjjuukwq is based on lcp initially, and then enhances solutions using ea the docid 4enojcrblrnfpqjxvxu4 use case is based on eas core algorithms comparison table ✅ lcp is ideal for fast, early stage screening and feasibility analysis where problem structure is simpler ✅ ea is preferred for final route development , incorporating full engineering and siting considerations that are infeasible to encode directly into an lcp framework aspect least cost path (lcp) evolutionary algorithms (ea) primary use in workflow early stage, rapid analysis and screening complex, nuanced infrastructure siting and detailed routing optimisation approach deterministic optimisation over simplified graph representations stochastic global search optimising over complex evaluation functions speed very fast for “regular” sized problems; limited complexity in routing logic slower due to extensive evaluations needed to converge, especially for detailed siting handling of complex design constraints requires constraints to be encoded into the graph structure, often as approximations (e g penalising high elevations to approximate pressure constraints) evaluation function is fully separate from algorithm → complex constraints (e g pipeline totex, spanning, installation methods) can be modelled directly and accurately ease of adding new design considerations can be difficult; constraints must fit the lcp problem structure relatively easy; design considerations are implemented within the evaluation function without changing the algorithm itself optimizing multiple objectives not straightforward; limited true multi objective capability naturally multi objective due to population based search evolving diverse solutions explainability high; solutions are deterministic and follow clear avoidance of high cost areas lower; stochastic nature makes explanations more complex we include a "decision log" in optioneer for this reason optimality returns the mathematically optimal solution for the simplified problem it has been given solutions often near optimal but not guaranteed globally, with a preference for providing diverse options for deeper consideration scalability & flexibility scales well within its problem formulation, but struggles with highly nuanced siting decisions scales better for complex, large, irregular search spaces with multiple design constraints typical use case in infrastructure routing rapid initial route generation for screening, flow equilibrium, network feasibility detailed siting optimisations considering min/max spans, angle constraints, pressure compliance, installation method selection