On 04/06/2013 04:58 PM, Michael ODonnell wrote: > >> kmax=1; >> kfactor=(kmax-1.0)/delta/(orderbc+1.0)/delbc^orderbc; >> Unless matlab treats floats and ints differently than most other >> languages, I'd say its a bug because zero over anything is always zero. > > Shame on anybody who'd intentionally code something like > that without also providing a comment about such a seemingly > counterintuitive approach. >-/ Let's hope it's simply a bug... > > Out of morbid curiosity, how does Octave report the value of > kfactor after > > kmax=1 > kfactor=(kmax-1.0) > > ?? > > _______________________________________________ > gnhlug-discuss mailing list > gnhlug-discuss@mail.gnhlug.org > http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/ > > kfactor = 0 is reported by octave in debug mode.
That's a long bit of code kfactor=(kmax-1.0)/delta/(orderbc+1.0)/delbc^orderbc; just to set it to zero. Clearly a bug, or just plain wrong and not thought through. Drives me nuts, since this construct was used for the x, y, and z axes. Guess I'll have to contact one of the authors of the book. The graduate student that wrote the code has since graduated. (Got her PhD int 2010.) Humph, you'd think the code samples for something this complicated would be better. \rant on The level of commenting in the code sample is not up to teaching standards, IMHO. If you are doing something tricky, write a descriptive comment on it! (I can't even remember tricky stuff I did last week without good comments.) Please, everyone, don't write ultra dense code like this, spaces are your friend! dx(:,2:je_tot,2:ke_tot)=C1ex(:,2:je_tot,2:ke_tot).* dx(:,2:je_tot,2:ke_tot)+ C2ex(:,2:je_tot,2:ke_tot).*((hz(:,2:je_tot,2:ke_tot)-hz(:,1:je_tot-1,2:ke_tot))- (hy(:,2:je_tot,2:ke_tot)-hy(:,2:je_tot,1:ke_tot-1)))./delta; Especially if there are 5 more lines just like it, with slightly differing indices, with no spaces in between! \rant off As you were. :) _______________________________________________ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
