home bbs files messages ]

Forums before death by AOL, social media and spammers... "We can't have nice things"

   sci.optics      Discussion relating to the science of op      12,750 messages   

[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]

   Message 11,699 of 12,750   
   Gash M to haitic...@gmail.com   
   Re: algorithm for piecewise defined mult   
   10 Feb 14 02:04:37   
   
   From: mgashaw@gmail.com   
      
   On Sunday, February 9, 2014 1:30:17 PM UTC+1, haitic...@gmail.com wrote:   
   > On Saturday, February 8, 2014 10:32:16 AM UTC-5, Gash M wrote:   
   >    
   > > Dear all,   
   >    
   > >    
   >    
   > >    
   >    
   > >    
   >    
   > > I want to make an optimization with a multivariable nonlinear merit   
   function defined in such a way that it is well-behaved(continuous function)   
   within a given interval but generally it is a piecewise function over the   
   large domain. Derivative-based    
   optimization algorithm is valid only in a subset of the whole domain.What kind   
   of algorithm do you suggest me to use which can accommodate the whole   
   interval? Or how do you tackle such problems?   
   >    
   > >    
   >    
   > >    
   >    
   > >    
   >    
   > > Thanks,   
   >    
   >    
   >    
   > Hi   
   >    
   > What are you optimizing?    
   >    
   > I can suggest neural networks as an easy way to get a solution if you have a   
   set of input and output data (or desired output.) But neural nets are not an   
   analytical solution, as you will have to dissect them afterwords to tell why   
   they did what.   
   >    
   > These kind of problems (if I understand in general what you are doing),   
   usually require specialized parts to solve specific areas.    
   >    
   > If you look at the blood circulation, it is handled differently according to   
   where it is. The capillary micro-environment and systems are quite different   
   from the cardio area.   
   >    
   > SO, in trying to make general systems, try to establish overall umbrellas of   
   generalizations first in order to get rules for interlocking optimized   
   systems. The blood flow to the periphery interacts with the heart, but it   
   mostly is autonomous.   
   >    
   > Interlocking systems are mostly "Libertarian," in that they operate by   
   themselves.    
   >    
   > It would be helpful to know what your applications are, or if you are   
   looking to develop a general theory.   
      
   Thanks for the reply.It is an optical design problem where the merit   
   function(MF) I defined is function of the surface profile of an optical system   
   which is a multivariable piecewise function.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)   

[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]


(c) 1994,  bbs@darkrealms.ca