Differentiable Programming of Chemical Reaction Networks
This work addresses the challenge of programming chemical computing systems for researchers in synthetic biology and computational chemistry, representing an incremental advance by applying differentiable methods to an existing substrate.
The authors tackled the problem of designing chemical reaction networks (CRNs) for computational tasks by introducing a differentiable formulation that enables training via optimization and regularization, resulting in the discovery of non-trivial sparse networks capable of implementing oscillators and other devices.
We present a differentiable formulation of abstract chemical reaction networks (CRNs) that can be trained to solve a variety of computational tasks. Chemical reaction networks are one of the most fundamental computational substrates used by nature. We study well-mixed single-chamber systems, as well as systems with multiple chambers separated by membranes, under mass-action kinetics. We demonstrate that differentiable optimisation, combined with proper regularisation, can discover non-trivial sparse reaction networks that can implement various sorts of oscillators and other chemical computing devices.