Condor takes flight
CONDOR, or COnstrained, Non-linear, Direct, parallel Optimization using trust Region method for high-computing load function, is a new algorithm designed to find the minimum of an objective function in the least number of function evaluations. An objective function is the function associated with an optimization problem which determines how good a solution is. It is based on the concept of trust regions. The basic idea is to approximate the function in question with a simpler function, which reasonably reflects the behaviour of the initial function in a specific neighbourhood. This neighbourhood is the trust region. The algorithm is time saving since it reduces the time needed to evaluate the objective function. With other algorithms this process can take days to complete. It has been designed to work in noisy functions, and it uses direct objective functions. That means derivatives of the function are not needed. All that is needed is a simple method or program, which can evaluate the function at a given point. Thus, the tool can be used very broadly in many different situations. In addition to this the efficiency of CONDOR can be increased even further since it can run on several CPUs simultaneously, greatly increasing computing power. CONDOR realises its full potential when combined with big software simulators of industrial processes such as in the petrochemical and other industries. It outperforms other optimization tools based on genetic algorithm (GA) and Artificial Neural Networks (ANN) by factors of up to 100. The possibilities for time and cost saving in the field of noisy and high-computing-load objective function optimization using CONDOR await exploitation.
