From: anon@nowhere.com   
      
   Not sure if this is the right group, so here goes...   I am attempting to   
   use 'simulated annealing' to tackle the problem of 'polynomial   
   factorization'.  I know deterministic methods exist to do this but I want to   
   try this out of interest.   
      
   My method works by firstly generating a possible 'template solution' . For   
   example, if we have a polynomial of degree 5, then there exist 4 possible   
   types of solution of 'number of factors' 2,3,4 or 5.   Each type of template   
   solution is tried in turn looking for a possiible situation whereby its   
   evaluation is equal to the evaluation of the original polynomial, there the   
   algorithm terminates as a possible factorization has been found.   
      
   The way I am writing this is that the coefficients for each template   
   solution are randomly selected and then a small random change is made to   
   them. The test solution is then evaluated and compared to the evaluation of   
   the original polynomial. (Actually several evaluations are made and their   
   average taken) Obviously the closer the test solution gets to the 'true   
   evaluation' the better the solution.   
      
   The search space then becomes the space encompassing all possible integer   
   values of the coefficients of the test polynomials.  I am then using the SA   
   technique of deciding on whether to keep the new solution according to a   
   probability function.   
      
   What I want to know really, is is this a good application for 'simulated   
   annealing'...?   Maybe not the best of times to ask this as I've already   
   started creating the routine. But do you think that this is a feasible way   
   of factorizing relatively small degree polynomials?   
      
      
   Thanks   
      
   [ comp.ai is moderated ... your article may take a while to appear. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)