Currently, models are transformed to expression trees and then compiled for evaluation.
Derivatives are calculated symbolically.
An interpreter for expression trees (with auto-diff support) would allow to use the library in settings where we have evaluate many different models efficiently.
Currently, models are transformed to expression trees and then compiled for evaluation. Derivatives are calculated symbolically.
An interpreter for expression trees (with auto-diff support) would allow to use the library in settings where we have evaluate many different models efficiently.