Abstract
Analog circuit design requires substantial human expertise and involvement,which is a significant roadblock to design productivity. Bayesian Optimization(BO), a popular machine learning based optimization strategy, has beenleveraged to automate analog design given its applicability across variouscircuit topologies and technologies. Traditional BO methods employ black boxGaussian Process surrogate models and optimized labeled data queries to findoptimization solutions by trading off between exploration and exploitation.However, the search for the optimal design solution in BO can be expensive fromboth a computational and data usage point of view, particularly for highdimensional optimization problems. This paper presents ADO-LLM, the first workintegrating large language models (LLMs) with Bayesian Optimization for analogdesign optimization. ADO-LLM leverages the LLM's ability to infuse domainknowledge to rapidly generate viable design points to remedy BO's inefficiencyin finding high value design areas specifically under the limited design spacecoverage of the BO's probabilistic surrogate model. In the meantime, samplingof design points evaluated in the iterative BO process provides qualitydemonstrations for the LLM to generate high quality design points whileleveraging infused broad design knowledge. Furthermore, the diversity broughtby BO's exploration enriches the contextual understanding of the LLM and allowsit to more broadly search in the design space and prevent repetitive andredundant suggestions. We evaluate the proposed framework on two differenttypes of analog circuits and demonstrate notable improvements in designefficiency and effectiveness.