Measures CPU performance by integrating the extended 3-species competitive Lotka-Volterra system over 3 million timesteps. More arithmetic per step than the 2-species version. Lower time is better.
While the classic two-species Lotka-Volterra model captures predator-prey dynamics, the extended competitive version models n species competing for shared resources. In the three-species formulation each species grows logistically and is suppressed not only by its own density but also by the two competing species through interaction coefficients αᵢⱼ. Depending on these coefficients the system can reach stable coexistence, exhibit oscillations, or result in competitive exclusion where weaker species go extinct.
The governing equations are dxᵢ/dt = rᵢ·xᵢ·(1 − (xᵢ + Σⱼ αᵢⱼ·xⱼ) / Kᵢ) for each species i. With three species this creates a 3×3 interaction matrix with nine off-diagonal terms, each requiring a multiply per step. The competitive model can be fully symmetric — all three species mutually suppress each other — unlike the asymmetric predator-prey system. This richer interaction structure produces a wider variety of long-term outcomes for similar parameter choices.
This benchmark integrates the 3-species competitive system over 3 million Euler steps with dt = 0.001. Each iteration computes nine interaction products plus three logistic growth terms — roughly 2.25× the arithmetic of the basic predator-prey benchmark. The higher instruction count per loop body makes this a more demanding test of the CPU's floating-point pipeline and instruction dispatch, while the same sequential data-dependency prevents parallelization within a single step. Lower time is better.