Abstract

Active matter systems are composed of non-equilibrium units that consume energy to perform directed motion [1,2,3]. Examples of acive particles at the mesoscopic scale are living, such as bacteria, or artificial, such as synthetic active colloids [4,5,6,7]. The theoretical framework describing these systems has shown tremendous success at finding universal phenomenology. When dealing with active colloids, on the one side, one might think of studying the features of a suspension of particles whose interactions are inspired on Soft Matter (passive) systems, such as isotropic (strongly repulsive [8,9,10], attractive [11, 12,13], micelle-inducing [14]) or anisotropic (Janus-like) interactions¹⁵, unravelling the relevance of hydrodynamics [12, 16]. On the other side, one might consider determining the forces that control the dynamics of the individual colloids directly from experiments. Accessing this local information would be the key for understanding the physics governing these systems and for creating models that explain the observed collective phenomena. For this purpose, we propose a machine-learning tool that uses the collective movement of the particles to learn the active and two-body forces, controlling particles’ individual dynamics. The method has been successfully tested not only on numerical simulations of Active Brownian Particles, considering different interaction potentials and levels of activity, but also on experiments of electrophoretic Janus particles, extracting the active and two-body forces that control the dynamics of the colloids [17]. We foresee that this methodology can open a new avenue for the study and modelling of experimental systems of active particles. REFERENCES [1] C. Bechinger, R. Di Leonardo, H. Lowen, C. Reichhardt, G. Volpe, and G.Volpe, Reviews of Modern Physics 88, 045006 (2016) [2] M.E. Cates, J. Tailleur. Annu. Rev. of Condens. Matt. Phys. 6, pp. 219-244 (2015). [3] S.Mallory,C.ValerianiandA.CacciutoAnnualreviewofPhysicalChemistry,6959(2018) [4] S. Thutupalli, R. Seemann, S. Herminghaus New J. Phys. 13, 073021 (2011). [5] W.F. Paxton et al. Chem. Commun. 441, 3 (2005). [6] S. Fournier-Bidoz et al. J. Am. Chem. Soc. 126, 13424 (2004). [7] I. Buttinoni, J. Bialke, F. Kummel, H. Lowen, C. Bechinger, T. Speck. Phys.Rev. Lett. 110, 238301 (2013). [8] J.Martin Roca, R Martinez, A Luis Diez, L Alexander, D Aartz, F Alarcon, J Ramirez and C Valeriani J chem Phys154, 164901 (2021) [9] DR Rodriguez, F Alarcon, R Martinez, J Ramírez, C Valeriani, Soft matter 16 (5), 1162 (2020) [10] J. Martin Roca, R. Martinez, F.Martinez Pedrero, J.Ramirez and C Valeriani, J. Chem. Phys. 156, 164502 (2022) [11] B. Mognetti, A. Saric, S. Angioletti-Uberti, A. Cacciuto, C. Valeriani and D. Frenkel Phys.Rev.Lett., 111 245702 (2013) [12] F.Alarcon, C.Valeriani and I.Pagonabarraga Soft Matter 10.1039/ C6SM01752E (2017) [13] J.Harder, S.Mallory, C.Tung, C.Valeriani and A.Cacciuto, J.Chem.Phys. 141 194901 (2014) [14] C.Tung, J.Harder, C.Valeriani and A.Cacciuto, Soft Matter 12 555 (2016) [15] S.Mallory, F.Alarcon, A.Cacciuto and C.Valeriani New Journal of Physics (2017) [16] F.Alarcon, E.Navarro, C.Valeriani and I.Pagonabarraga, PRE submitted (2018) [17] “Discovering dynamic laws from observations with Graph Neural Networks: the case of self-propelled, interacting colloids” M. Ruiz-Garcia, C.M.Barriuso, L.Alexander, D.Aarts, L.Ghiringhelli and C.Valeriani, submitted (2022)