The Nonlinear Programming Solver

Example 9.2 Solving Unconstrained and Bound-Constrained Optimization Problems

Although the NLP techniques are suited for solving generally constrained nonlinear optimization problems, these techniques can also be used to solve unconstrained and bound-constrained problems efficiently. This example considers the relatively large nonlinear optimization problems

\[  \displaystyle \mathop \textrm{minimize}f(x) = \sum _{i=1}^{n-1} ( - 4 x_{i} + 3.0) + \sum _{i=1}^{n-1} (x_{i}^{2} + x_{n}^{2})^{2}  \]

and

\[  \begin{array}{ll} \displaystyle \mathop \textrm{minimize}&  f(x) = \sum _{i=1}^{n-1} \cos (-.5x_{i+1} - x_ i^2) \\ \textrm{subject\  to}&  1 \le x_ i \le 2, \;  i = 1,\dots , n \end{array}  \]

with $n=100,000$. These problems are unconstrained and bound-constrained, respectively.

For large-scale problems, the default memory limit might be too small, which can lead to out-of-memory status. To prevent this occurrence, it is recommended that you set a larger memory size. See the section Memory Limit for more information.

To solve the first problem, you can write the following statements:

proc optmodel;
   number N=100000;
   var x{1..N} init 1.0;

   minimize f = sum {i in 1..N - 1} (-4 * x[i] + 3.0)  +
                sum {i in 1..N - 1} (x[i]^2 + x[N]^2)^2;

   solve with nlp;
quit;

The problem and solution summaries are shown in Output 9.2.1.

Output 9.2.1: Problem Summary and Solution Summary

The OPTMODEL Procedure

Problem Summary
Objective Sense Minimization
Objective Function f
Objective Type Nonlinear
   
Number of Variables 100000
Bounded Above 0
Bounded Below 0
Bounded Below and Above 0
Free 100000
Fixed 0
   
Number of Constraints 0

Performance Information
Execution Mode Single-Machine
Number of Threads 4

Solution Summary
Solver NLP
Algorithm Interior Point
Objective Function f
Solution Status Optimal
Objective Value 0
   
Optimality Error 1.001127E-14
Infeasibility 0
   
Iterations 16
Presolve Time 0.00
Solution Time 3.51


To solve the second problem, you can write the following statements (here the active-set method is specifically selected):

proc optmodel;
   number N=100000;
   var x{1..N} >= 1 <= 2;

   minimize f = sum {i in 1..N - 1} cos(-0.5*x[i+1] - x[i]^2);

   solve with nlp / algorithm=activeset;
quit;

The problem and solution summaries are shown in Output 9.2.2.

Output 9.2.2: Problem Summary and Solution Summary

The OPTMODEL Procedure

Problem Summary
Objective Sense Minimization
Objective Function f
Objective Type Nonlinear
   
Number of Variables 100000
Bounded Above 0
Bounded Below 0
Bounded Below and Above 100000
Free 0
Fixed 0
   
Number of Constraints 0

Performance Information
Execution Mode Single-Machine
Number of Threads 4

Solution Summary
Solver NLP
Algorithm Active Set
Objective Function f
Solution Status Optimal
Objective Value -99999
   
Optimality Error 1.423935E-12
Infeasibility 0
   
Iterations 8
Presolve Time 0.00
Solution Time 3.78