I load several large multi-dimensional lookup tables in the initialization of a subsystem that has lookup table blocks. The lookup tables are in a library model. When I build a Simulink model from the library, I will have several instances of the library. This means something like ~10-20 large lookup tables. This really seems to slow down the initialization. When I am interactively running the model, I don't care about the extra time. However, I want to call the big model several thousands times in a loop. The biggest cost is the loading the lookup tables and doing a calculation on the lookup tables to reduce them by 1 dimensions that really only needs done the first time the library runs. Is there a clever way for the data to remain persistent from run to run?
- The persistent keyword only works in MATLAB functions. It does not work in Simulink initializations.
- I cannot use the base workspace. I cannot guarantee what will be going on the base workspace because I will share my library.