Functions must be Y=f(X)
When you enter an equation into Prism, the independent variable must be 'X' and the dependent variable must be 'Y'. So if you measure a voltage as a function of time, you cannot enter an equation that defines V as a function of t. It must define Y as a function of X.
•Prism can fit an implicit equation where Y is on both sides of the equation.
•Prism can fit a differential equation that defines dY/dX.
•Prism cannot fit a model defined by a set of differential equations. For this reason, it cannot fit many compartmental models.
No models with more than one X variable
Prism does not calculate multiple nonlinear regression, so cannot fit models with two or more independent (X) variables. But note that you can define a parameter to be a column constant, in which case its value comes from the column titles. In some cases, you can think of these column constants as being a second independent variable.
Ways to evade this limitation.
Model complexity
Prism compiles your equation into an internal format it uses to calculate the math efficiently. If the compiled version of your equation won't fit in the space Prism sets aside for this purpose, it reports that the equation is "too complex" .
If you see this message, don't give up. You can usually rewrite an equation to make it less complex. Do this by defining an intermediate variable that defines combinations of variables. For example if your equation uses the term "K1+K2" four times, you reduce complexity (but keep exactly the same mathematical meaning) by defining an intermediate variable at the top of your equation (say, K1P2=K1+K2) and then using that intermediate later in the equation. That way Prism has fewer steps to store.
No loops or sums
Prism does not have any syntax to allow for summations or loops in user-defined equations.
If the loop or sum occurs a small and consistent number of times, a workaround would be to write the equation several times, once for n=1, then for n=2 ... and add up the results to compute the final value of Y.
