Sometimes dual_axial_max will "lose" a peak. This may happen when the search range includes several local maxima and the initial guess happens to hit a local maxima instead of a global maxima. In order to exclude the local maxima from the search range, the pressure field in that area can be calculated and displayed:
MAKE_VOLUME,system_file,output_file CALC_LENS,output_file KPLOT,outputfile,/contour,/db
Then the search range in the maximum-finding parameter file can be refined appropriately. It takes about 2 minutes to generate the pressure field for the default sampling of MAKE_VOLUME.
Another way to avoid the lost peak problem is to change the initial guess in the y-direction. The situation which usually causes lost peaks is when the pressure field looks like:
x X x
where the small x's are local maxima and the large X is the global maxima. If the y guess is in line with one of the little x's instead of the large X, the routine will get stuck on the little x's and jump back and forth between the x's until it gives up and "loses" the peak. Changing the y guess to a position in between the two x's will solve the problem.
A third way to avoid this situation is to search in the y direction first instead of second (by specifying yx instead of xy as the search type in the parameter file.) GENERATE_SYSTEM uses this approach. In addition, it sets the initial guess in the x direction to a point 2/3 of the way from the left side of the search range in order to avoid the local maxima shown above.
Go back to the Utility Programs
Kevin Fink's Home Page (http://www.fink.com/Kevin.html)email@example.com