yeh, it not only "doesn't work", i'm sure it won't even compile either!

 

UnivariateRealFunction is an interface - you must define a class that implements it, then pass an instance of that class to the solver.  for instance (this is typed free-hand, just as an example, not tested):

 

...however..., it appears you're using one of the 3.0 pre-releases of commons math, so good luck with that!, as it changes rather frequently and the docs don't seem to keep up.  you may be better off getting the stable 2.2 release and using one of the simpler solvers available there (BisectionSolver may be adequate)

 

philho is correct, you must catch the exception - whether you DO anything with it or not is up to you.  you'll get a compile error if you don't handle the exception with a try/catch for a method that declares it can throw it.

 

moving on tho, your REAL problem will be that (assuming you're using v2.2) you've likely got the parameters scrambled.  As written, you're looking for roots on [100,-0.1] starting at 0.1.  Bisection is probably giving you a convergence exception because there are MANY roots to sin() on that interval.  (and it may even be an invalid interval, not sure, but suspect it needs to be strictly ordered [lo,hi])  that's what's CAUSING the exception, that's the runtime error (assuming you've got the try/catch block to resolve the compile error)

 

i THINK what you want is "solve(100,function,-0.1,0.1)", so that the 100 is num evaluations, correct?  so, something like:

 

try {

  double c = solver.solve(100,function,-0.1,0.1);

  println("c = " + c); // approx 0, to whatever N you've called solver.setAbsolute/RelativeAccuracy(N)

} catch(Exception e) { println("oops, whatever"); }

 

also, check the docs, and try not to use any deprecated forms (there are several).