Many calculation tasks can be handled either with a variable or a function. Are the calculations handled differently in the code? Are there advantages of one over the other, or are they just different ways of organization?
For example, a model might have a source term R that depends on many things. One could define R as R: myVariable and in a variables node: myVariable = thing1 * thing2^2 / sqrt(thing3)
Alternatively, one could use a function: R: myFunction(thing1, thing2, thing3) and then define the analytical function myFunction(a,b,c) = a * b^2 / sqrt(c)
Once the model is run, are these approaches different in efficiency or stability?