熵函数法与几种优化方法的比较

THE COMPARISON BETWEEN ENTROPY FUNCTION METHOD AND SEVERAL OTHER OPTIMIZATION METHODS

熵函数法（或称极大熵方法）是近些年发展起来的求约束优化问题的一种方法,一些数值例子表明了它的实用性,而且这种方法具有一些良好的性质[1-4,6,7,12].这种方法的想法是将原优化问题化为一个含参数p的无约束光滑优化问题去求解,当参数p充分大时,以无约束优化问题的解作为原问题的近似解.这一点与简单罚函数法十分相似,所以将这二种方法进行比较,有助于对嫡函数法进一步的了解.更主要地我们希望了解这种方法的效率.由于熵函数法与简单罚函数法的相似性,对极大极小问题,可以从理论上对二者进行比较,但无法从理论上将熵函数法与其它一些方法进行比较.

The entropy function method is a kind of method developed in the last decade for solving optimization problems. In this paper we make a comparison between this method and several other optimization methods. Our numerical experiment shows that the entropy function method is much better than simple penalty function method and has similar efficiency with SQP algorithm and Vardi's method.