Question

I’m porting a C# Library to Java (for nonlinear regression, original code here). The library uses the Func<> class, which doesn’t exist in Java. I.E. (A, B, C, and D, and time are parameters used for regression, don’t need fixed.)

Func<double>[] regressionFunctions = new Func<double>[]
{() => A * Math.Exp(time) * Math.Sin(B * time),
() => C * Math.Exp(time) * Math.Cos(D * time)}; 

What I would like to do is convert this to Java code. I saw something about creating an anonymous inner class, but I’m not certain of the correct usage for this particular situation. I would like the equation to be evaluable at a later time (for a particular value of t). Do I need a new class, an interface, or what would be the best method?
Any help would be appreciated! Thanks in advance.

Answer 1

You can define your own interface, and use anonymous implementations to port the code:

// Declaring the interface
interface FuncDouble {
    double calculate();
}

public LevenbergMarquardt(FuncDouble[] regressionFunctions, ...) {
    // Using the functor:
    double functionValue = _regressionFunctions[i].calculate()
}

// Declaring an array of functors:
FuncDouble[] regressionFunctions = new FuncDouble[] {
    new FuncDouble() {
        public double calculate() {
            return A * Math.Exp(time) * Math.Sin(B * time);
        }
    }
,   new FuncDouble() {
        public double calculate() {
            return C * Math.Exp(time) * Math.Cos(D * time);
        }
    }
};

In order for the implementation to work, A, B, C, D, and time variables must be either instance/class variables, or be final local variables.