b' . . . toautomatically predict the atomic rearrangementsinvolved as materials make these transitions. 27PALLAS predicted this complex transition pathway network for silicon. It is a road map that shows possible transition pathways between different structures. PALLAS correctly identifies the lowest-energy transition pathway for silicon upon decompression from high pressure (thick black line). The structures in the four boxes show the different configurations of atoms along the transition path. The background colors indicate contours on the energy landscape. The points indicate stable and metastable structures of silicon with their size and color proportional to volume and energy, respectively. The composite of circles is similar to methods used in social networking analysis. Image courtesy Tim Strobel, Carnegie Institution for ScienceAlthough there are existing approaches tosilicon. PALLAS identified previously known, low-determining these pathways, there are majorenergy phase transition pathways for the wurtzite limitations analyzing transitions between complexto rock-salt structures in cadmium selenide and, crystal structures. importantly, revealed a new lower-energy pathway not yet observed. The team used swarm intelligence and graph theory in PALLAS. Swarm intelligence considersFor silicon, PALLAS explained the labyrinthine the collective intelligence of a system, such as thephase transition steps observed when it is effective performance of a colony of ants overdecompressed from high pressure, revealing the one ant alone, while graph theory studies thecomplex sequence that deviates from the most relationship among data points within complexstable structures. By understanding detailed phase networks. PALLAS predicts the stable andtransition mechanisms, PALLAS can be used to metastable crystalline structures and the transitionguide synthesis of future materials, including pathways that connect them without preexistingmaterials found in deep planetary interiors, information.including so-called super-Earths, and for new technology, and to possibly address biological They compared results from PALLAS againstproblems such as protein folding. known phase transitions in cadmium selenide and'