For overcoming the weakness of the population climbing evolutionary algorithm,we design a new algorithmthat randomly chooses many parents from the population to recombine and the worse individuals to mutate so as to de-crease the size of population, accelerate the convergence rate and improve the performance. The results of numericalexperiments including seven non-linear optimization problems show that the new algorithm is characteristic of robustand high efficiency,and can quickly find the global solutions which are better than those got by MATLAB and othermethods.