Float Pow Bug

malc · 3 May 2005

Hi unkargherth,

It's not a bug as such, although you could argue that it should indicate what's happening in a better way.

Any negative number that you raise to the power of a value whose absolute is less than 1 has no solution in the real numbers. -1 to the power 0.5 is the classic complex number i for example.

Maybe it should say something more useful than -1#IND.

unkargherth · 3 May 2005

As far as my Casio Hand Calculator knows, -0.5^-0.3=-1.231144413, which, in fact is -1*(0.5^-0.3)

I checked it in Scientific Notebook ( a kind of mupad, mapple engine) and gives same result

Sk · 3 May 2005

unkargherth wrote: As far as my Casio Hand Calculator knows, -0.5^-0.3=-1.231144413, which, in fact is -1*(0.5^-0.3)

I checked it in Scientific Notebook ( a kind of mupad, mapple engine) and gives same result

You're wrong, because of the order math must be executed. These two are different: a) (-0.5)^(-0.3) is imaginary: it equals the inverse of the tenth root of (-1/2)^3, therefore it's like j*(2^(0.3)); b) -(0.5^(-0.3)); this equals the opposite of the inverse of the tenth root of (1/2)^3, which is -(2^(.3)).

malc · 3 May 2005

unkargherth wrote: As far as my Casio Hand Calculator knows, -0.5^-0.3=-1.231144413, which, in fact is -1*(0.5^-0.3)

A bug in your Casio? 😉

Seriously though, Sk is right and it does look like your calculator is interpreting the operation differently. I tried it on the Windows XP Calculator accessory and it made the same mistake. However, if you try punching in (-0.5)^-0.3 you'll see that it gives an invalid result.

All good fun 🙂

Sk · 4 May 2005

I think this happens because in SM the module takes the float you give him as the parameter between parenthesis, so if you put -0.5 it solves (-0.5)^(exp). Am I wrong?