Giuseppe Nicosia, Title: Robust Circuit Design by Pareto Optimality

We present a simulation-based multi-scenario multi-objective optimization algorithm for the circuit design problem (e.g., analog/mixed-signal circuits and biochemical circuits). We define the circuit design problem as a large-scale multi-scenario constrained-multi-objective optimization problem.

The algorithm, ParetoDesignAutomation (PareDA), has been tested on six different operational amplifiers and compared with commercial (MunEDA WiCkedTM, CADENCE ADE GXLTM and academic  (Geometric Programming, IPOPT, NSGA-II, YdIRCO) methods, demonstrating its effectiveness: PareDA obtains a significant average performance improvement (from +35% to  +49%) over the state-of-the-art techniques. It satisfies the constraints, optimizes the performances while minimizes silicon area and power consumption maximizing the overall yield. Moreover it scales gracefully with circuit size and type and can find solutions working at different operative conditions. The PareDA algorithm thanks to the parallel computations gains a speedup of 5.62x  with  70% efficiency. Moreover the algorithm has been integrated with sensitivity, epsilon-dominance and robustness analysis in order to select the effective circuit design.

Finally, we use the framework to design metabolic pathways in which many objectives are optimized simultaneously. This allows to characterize the energy signature in models of algal and mitochondrial metabolism and to produce biofuel, 1,4 Butanediol, using E. coli.