Fletcher routine

Figure 22 Root-mean-squared error progression plot for Fletcher nonlinear optimization and back-propagation algorithms during training.-----Fletcher routine —A—Gradient search.

For the molecules (2) MNDO heats of formation are calculated by starting from known structural data and optimizing their respective geometries via the Davidson-Fletcher-Powell program s ib-routine (Figure 2). 

It is not appropriate in this review to discuss the relative efficiency of minimization routines based on steepest descent procedures, but a few of the more important references are given. If the gradient can be evaluated analytically, and as McIver and Komomicki 25) have pointed out this is the case for many surfaces calculated by semi-empirical molecular orbital methods, then this is the best approach. The methods developed by Davidson 26), Fletcher and oweZZ (27), and AfurfagZz and Sargent 28) assume this to be the case. If the gradient has to be obtained numerically then the method developed by Powell 29) has found wide acceptance for its efficiency. 

Some of the above-mentioned methods have limitations. Southern blot analysis requires a large amount of high-quality DNA, and the results are unreliable if the sample has a low percentage of tumor cells (Ross and Fletcher, 1998). For these and other drawbacks, this method is precluded from becoming a routine clinical protocol to determine HER-2 status. 

The program uses several minimizing routines that can easily be chosen, such as the pit-mapping method developed by Sillen, but also Simplex, Davidon-Fletcher-Powell, and Monte Carlo methods of MINUIT package [53]. As for AR, up to third-degree equations can be used ... 

---