In the scope of Work Package 1, the research team developed several methods to analyze the two distinct types of networks outlined in WP 1.1 and WP 1.2. A central part of this effort involved applying the framework of dynamical mean field theory to explore the thermodynamics of general networks encompassing both categories. To support this analysis, the team drew upon the mathematical foundations of optimal transport theory.
Additionally, a general framework was established to optimize the driving forces within these networks, considering both energy dissipation and other thermodynamic properties such as precision. This optimization relied on linear response theory. Complementing this, the researchers devised a method to infer these thermodynamic quantities by leveraging known thermodynamic bounds.
In Work Package 2, the investigation began with a fundamental network motif: the memory device. A machine learning algorithm was employed to minimize the thermodynamic cost associated with memory erasure. The focus then shifted to sensory systems, particularly proofreading mechanisms. Here, the team extended the thermodynamic uncertainty relation to derive a universal thermodynamic bound. They also examined alternative proofreading strategies, such as energy-relay proofreading.
The work concluded with an analysis of signaling networks, with particular attention to positional information. Using macroscopic fluctuation theory, the researchers derived an upper bound on the achievable positional information, expressed in terms of underlying thermodynamic gradients.
