pow() does not handle fractional exponent correctly

[Double_D]

The pow() function does not appear to handle fractional exponents correctly.  The correct value is returned by pow() for the following code:

hole_position[B5_B].neutronCount.presentCount = pow ( (hole_position[10].detectorCurrent  *hole_position[12].detectorCurrent* hole_position[13].detectorCurrent), 0.333); 

For current values of 1.1E-7, 1.3E-7, and 1.4E-7, a value of 1.28E-7 is returned.

I had originally used the following code:

hole_position[B5_B].neutronCount.presentCount =
                pow ( (hole_position[10].detectorCurrent * hole_position[12].detectorCurrent
                        * hole_position[13].detectorCurrent), 1/3);   

For this case, pow() returned a value of 1 for the same current values.

Using 0.333 gives the needed precision for my application, but the behavior when passing a fractional exponent seems to be a minor bug in this library function. 

[Mert_A. · Accepted solution]

This is not a problem with pow, but rather a gotcha of the C language. In your code, 1 and 3 are being treated as integers, and the division of two integers yields an integer result. Thus 1/3 becomes the integer 0. This is why pow returns 1.

The solution is to do something to this effect:

((double)1) / 3

or better,

1.0 / 3.0

As long as you ensure that one of the operands is a double, the result will be too.

Mert A.

National Instruments

[ebalci]

If you ever need to replace the numeric constants 1 and 3 with variables, you can mutiply them with 1.0 to convert them to double.

Sometimes I prefer this to type casting.