Title: Computing properties of stable configurations of thermodynamic binding networks

Abstract: The promise of chemical computation lies in controlling systems incompatible with traditional electronic micro-controllers, with applications in synthetic biology and nano-scale manufacturing. Computation is typically embedded in kinetics---the specific time evolution of a chemical system. However, if the desired output is not thermodynamically stable, basic physical chemistry dictates that thermodynamic forces will drive the system toward error throughout the computation. The thermodynamic binding network (TBN) model was introduced to formally study how the thermodynamic equilibrium can be made consistent with the desired computation, and it idealizes tradeoffs between configurational entropy and binding. Here we prove the computational hardness of natural questions about TBNs and develop a practical algorithm for verifying the correctness of constructions by translating the problem into propositional logic and solving the resulting formula. The TBN model together with automated verification tools will help inform strategies for error reduction in molecular computing, including the extensively studied models of strand displacement cascades and algorithmic tile assembly.