If you put all this together, you get the source code for a small application. The application defines a parser variables ("a
") and adds
a user defined functions named "MyFunc
". When using the parser make sure that you don't forget to catch the Parser::exception_type
in your application. It contains detailed information helping you to find syntax errors in your formula.
#include <iostream> #include "muParser.h" // Function callback double MySqr(double a_fVal) { return a_fVal*a_fVal; } // main program int main(int argc, char* argv[]) { using namespace mu; try { double fVal = 1; Parser p; p.DefineVar("a", &fVal); p.DefineFun("MySqr", MySqr); p.SetExpr("MySqr(a)*_pi+min(10,a)"); for (std::size_t a=0; a<100; ++a) { fVal = a; // Change value of variable a std::cout << p.Eval() << std::endl; } } catch (Parser::exception_type &e) { std::cout << e.GetMsg() << std::endl; } return 0; }
Finally, I'd like to give you some benchmarks. The benchmarking was done on an Intel Pentium P-4 with 2.6 GHz, with a version compiled by using MSVC++ 7.1 (Standard edition). The diagram shows number of evaluations per seconds vs. expression length. I compared both the static lib and the dll version with two other parsers that are freely available on the net, very fast and have a similar set of features. One of them is a commercial product.
A higher curve means better performance. Expressions were created randomly. They used only sin
and cos
functions and contained
multiple variables and constants. In order to smoothen the curves each point represents the value of a
running average over 10 sample expressions.