I have been building a neural network using Tensorflow to predict the location of flying object. I have been able to successfully run the data through and get a predicted output, but I want to include physics constraints such as location-based air pressure which will increase the probability that the object is in a certain location. I don't want this to be a part of the training data, just a constraint. How do I do this? Should I just use an optimization function? Could I manually input increasing the probability in the ML script?