Software verification may yield spurious failures when environment\nassumptions are not accounted for. Environment assumptions are the expectations\nthat a system or a component makes about its operational environment and are\noften specified in terms of conditions over the inputs of that system or\ncomponent. In this article, we propose an approach to automatically infer\nenvironment assumptions for Cyber-Physical Systems (CPS). Our approach improves\nthe state-of-the-art in three different ways: First, we learn assumptions for\ncomplex CPS models involving signal and numeric variables; second, the learned\nassumptions include arithmetic expressions defined over multiple variables;\nthird, we identify the trade-off between soundness and informativeness of\nenvironment assumptions and demonstrate the flexibility of our approach in\nprioritizing either of these criteria.\n We evaluate our approach using a public domain benchmark of CPS models from\nLockheed Martin and a component of a satellite control system from LuxSpace, a\nsatellite system provider. The results show that our approach outperforms\nstate-of-the-art techniques on learning assumptions for CPS models, and\nfurther, when applied to our industrial CPS model, our approach is able to\nlearn assumptions that are sufficiently close to the assumptions manually\ndeveloped by engineers to be of practical value.\n