Which makes Knuth not suitable as a reference to support   
   R. Fatemans conjecture about GCD and GB.   
   Also there is a problem GB is a moving target, take F4,   
   its quite a mixture of different approaches.   
      
   So lets say R. Fatemans conjecture is still undecided   
   in my opinion, what evidence do we have so far? Are there   
   some theoretical results that would support R. Fatemans   
   conjecture, not only some lousy maxima benchmarks?   
      
   and where do we draw a line? What if the next big GB   
   algorithm uses subresultants? Its all very thin ice,   
   not clearly specified what the R. Fateman conjecture   
   should be at all. There is also doubt that the R. Fateman   
      
   conjecture makes any sense at all. Given that any   
   GCD algorithm will progress through some states S1,..,Sn   
   and any GB algorithm also progresses through some states   
   T1,..,Tn, and that for example in the realm of GCD   
      
   there are already theoretical results that for example   
   Euclid GCD based on polynomial division and GCD based on   
   subresultant, are essentially the same, except for one   
   lego brick that was replaced.   
      
   It might be that there are also such holes in the GB   
   algorithm, where we can plug-in different lego bricks.   
   This is already seen in GB algorithms for some symmetric   
   cases, don't ask me more, but it seems the variation point   
      
   there are the S-polynomials. Etc.. Etc..   
      
   Am Freitag, 7. Juli 2017 00:30:00 UTC+2 schrieb burs...@gmail.com:   
   > Knuth is kind of a methusalem reference for GCD, his 2nd   
   > vol is from 1969, he even doesn't mention GCD via GB.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)