In order to solve linear programs of realistic size ,it is necessary to take advantage of the sparsity of the constraint matix. So in this papers instead of the QR factorization we study the algorithm which used in large sparse problem0 We use the LU factorization combine with the Linear Least Squares Problem to solve The gradient projection .This method can retain the sparsity of the constraint matix, at the same time , this method can reduce the storage Next , we apply this method to the Least Squares Problem, analyse the a algorithms and give the storage and we obtain a substantial reduction of the calculation cost .At last, we give the numerical experiments 0...
|