Fortran Taking 1/7 power of a variable

[abdulsulo]

Hello there. I am trying to take a power of 1/7 of one variable but as a default it gives me 1 similar to if I was taking power of 0. How can I solve this problem. Thanks

 

[DrClaude]

What exactly did you write? I suspect that the 1/7 is being treated as an integer operation and rounded to 0.

 

[abdulsulo]

Do I=2,N,1
X1=(Yr(I)/R)
X2=X1**(1/7)
Ua(I)=Um*X2
END DO

X1 and X2 are both double precision. I decided to divide the equation and take power but still it didn't work for me.
On first iteration I am giving you the values of variables there;
R=0.05
Um=2
Yr= 2.85E-05

Whatever I do it gives X1**(1/7) as 1 which is not correct.

 

[abdulsulo]

What exactly did you write? I suspect that the 1/7 is being treated as an integer operation and rounded to 0.

Your hint gave me an idea and I think I solved the problem. I put 1./7. to the equation and problem solved. Didn't notice it would roll it down to 0. Thank you very much.

 

[DrClaude]

In the expression (1/7), since both 1 and 7 are integers, it is treated as an integer operation, and the result is returned as an integer, namely 0. Try (1.d0/7.d0) instead.

 

[Mark44]

In addition to Fortran, the language used in the example of this thread, programming languages such as C and C++ also exhibit the same behavior of distinguishing between integer division and floating point division. As DrClaude already said, when both operands are integer values, the type of division that is performed is integer division. In this case, 1/7 is evaluated as 0. If one or both operands are a floating point type, then floating point division is performed.

The most commonly used CPUs for personal computers these days are those produced by Intel and AMD. Both types of CPU have one op code for integer division (DIV) and another for floating-point division (FDIV), which gives more precise results, but takes longer to calculate. Compilers for Fortran and C and C++ choose the type of division to perform based on the code you write.

 

[jedishrfu]

This is by far the most common error made when you write your first fortran program from scratch followed by:

1) implicit/explict variable naming conventions where variables starting with i thru n are integers and all others are reals unless you say otherwise with explicit statements.

2) format statement errors where the description is too small for displaying the number.

3) for free-form fortran typing a C in column 1 created a comment line so you had to make sure that "character" or "common" statements were indented otherwise you got all sorts of compile-time errors.

I added all of these to my personal collection of errors made list more than forty years ago give or take but more take than give.

and there are many more subtle ones too.

 

[rumborak]

If I ever designed a programming language, floating point division would be the default behavior. I do not recall the last time I actually used integer division.

 

[Mark44]

If I ever designed a programming language, floating point division would be the default behavior. I do not recall the last time I actually used integer division.

I use integer division often. This operation and the modulus operation are useful in converting days to weeks and days, ounces to pounds and ounces, making change, and many other applications.

 

[jtbell]

Doesn't Pascal have separate names for the two division operations? 'div' for integer and '/' for floating-point IIRC. That makes the distinction more obvious and let's you choose between the two where appropriate. I wonder why more languages don't do this, except for their age and backwards-compatibility issues (e.g. Fortran, C, probably C++).

 

[rumborak]

I think you got the answer right there, backwards compatibility. Or rather, most programming languages, the time they were introduced, tried to do a few things differently, but at the same time tried to make it easy for programmers to switch over. So, they stuck with non-essential stuff like that.