1: \begin{abstract} 

2: Biologically-inspired network models provide a powerful framework for optimization algorithms. One prominent algorithm simulates the networks of filaments built by the unicellular organism Physarum polycephalum. Inspired by nature, we analyse the networks predicted by this dynamics in the presence of stochastic forces simulating a thermal environment. The resulting networks exhibit different topologies as a function of the temperature, which are generally multistable. Remarkably, the system converges to the most robust configuration at non-zero temperatures exhibiting a resonant-like behavior. This configuration is reached with the maximal convergence rate and is not found by the deterministic algorithm. Our results show that stochastic dynamics could qualitatively increase the computational power of optimization algorithms by actively modifying the landscape of the minima corresponding to the sought solutions.

3: \end{abstract}

4: