For the macroscopic world, classical thermodynamics formulates the laws governing the transformation of various forms of energy into each other. Stochastic thermodynamics extends these concepts to micro- and nano-systems embedded or coupled to a heat bath where fluctuations play a dominant role. Examples are colloidal particles in time-dependent laser traps, single biomolecules manipulated by optical tweezers or AFM tips, and transport through quantum dots. For these systems, exact non-equilibrium relations like the Jarzynski relation, fluctuation theorems and, most recently, a thermodynamic uncertainty relation have been discovered. First, I will introduce the main principles and show a few representative experimental applications. In the second part, I will discuss the universal trade-off between the thermodynamic cost and the precision of any biomolecular, or, more generally, of any stationary non-equilibrium process. By applying this thermodynamic uncertainty relation to molecular motors, I will introduce the emerging field of "thermodynamic inference" where relations from stochastic thermodynamics are used to infer otherwise yet inaccessible properties of (bio)physical and (bio)chemical systems.
The formation of self-organized patterns of cell differentiation is key to the morphogenesis of multicellular organisms, and while a general theory of biological pattern formation is still lacking, we introduce a generalisation of Turings work on pattern formation in monophasic systems to biphasic multicellular tissues. Our model incorporate morphogen production and transport, cell differentiation and tissue mechanics in a single framework, where a first tissue phase consists in a three-dimensional viscoelastic network made of cells, and a second phase composed of extracellular fluid, both compartments being separated by cell membranes, actively regulating interfacial fluid and morphogens exchanges. Coupling reaction-diffusion, active membrane transport and tissue poroelasticity, we show that tissue spatial organisation and mechanics can control developmental pattern formation. Overcoming crucial limitations of conventional reaction-diffusion models, we demonstrate the possibility of generating robust spatial patterns of cell differentiation using biochemical signalling pathways and gene regulatory networks involved in cell fate decisions, with either a single morphogen or multiple equally diffusing molecular signals.
Tensor networks are an efficient representation of interesting many-body wavefunctions and underpin powerful algorithms for strongly correlated systems. But tensor networks could be applied much more broadly than just for representing wavefunctions. Large tensors similar to wavefunctions appear naturally in certain families of models studied extensively in machine learning. Decomposing the model parameters as a tensor network leads to interesting algorithms for training models on real-world data which scale better than existing approaches. In addition to training models directly for recognizing labeled data, tensor network real-space renormalization approaches can be used to extract statistically significant "features" for subsequent learning tasks. I will also highlight other benefits of the tensor network approach such as the flexibility to blend different approaches and to interpret trained models.