Abstract
Machine-learning-based practical circuit synthesis requires data sets that consist of massive amounts of samples for the training and testing of the machine learning models. In order to generate such a large amount of samples, fast and automatic data generation methods are in demand. In this paper, we present an application programming interface (API) which, with a single Python script, can launch a massive amount of simulations in Keysight ADS, collect them, and post-process them in a machine learning environment. The proposed API can also be used to consider practical issues (e.g., parasitic effects, nonlinearity) during circuit synthesis and provide the part numbers of commercial components needed for a circuit with desired performance. To demonstrate the proposed API, it was employed in a synthesis tool of a practical 4 th -order impedance matching circuit.