[A83] Re: floating point errors?


[Prev][Next][Index][Thread]

[A83] Re: floating point errors?




At 04:06 PM 6/1/02 +0000, you wrote:


> >Only pixels are calculated so, unless some pixel has 1 as value, that
> >special case never occures. Also, you'll have to set the graphing mode
> >to dots, otherwise, the calc connects two points. By the way, in dot
> >mode you can see that it calculates pixel per pixel. There'll be only
> >one point per pixel collumn.
>
>Yes, I knew that, but why does it miscalculate the immediate
>pixels before and after the 'missing point'? Normally it
>should draw a straight line with that one pixel missing.
>Like with zoom 4:decimal

The way you draw lines for the function is to store the last point, then 
calc the new one, then connect the dots.  If the calculation for the new 
point is invalid, then it simply moves to the next one, but the previous 
point is still stored, so a line gets drawn between the last valid point 
and the next valid point.  This is why teachers hate the graphing calcs for 
looking at graphs involving tan(), the asymptotes (almost vertical lines) 
are drawn solid, so the graph looks wrong.





References: