[Gnuplot-bugs] non-integer roots of negative numbers

[Paul V. <pau...@gm...>]

Hello,

I am writing in to report a (possible) bug.

Consider the function : f(x) = 1/((3.43923-1.6*x*2)3/2)

For all x > 1.466123716, 3.43923-1.6*x*2 is a negative number. The exponent
of this value is not an integer (1.5) and, therefore, the function f(x)
should be undefined for all x > 1.466123716.

The corresponding graph of this function can be seen here :

http://wims.unice.fr/wims/wims.cgi?session=H8A290B772.1&+lang=en&+module=tool%2Fanalysis%2Ffunction.en

You have entered: *f* (x) = 1/((3.43923-1.6*x*2)3/2).



It can also be seen here:

http://newgraph.seriesmathstudy.com/


Now, contrast this with the output produced by gnuplot ( try the command :
"plot 1/((3.43923-1.6*x**2)**(3/2))" (without quotes, ofcourse) ).


Is this something that needs to be fixed?


Sincerely,

Paul
AI LAB, UIC