about precision

[keremusu1]

Hello!

I need to set a variable "n" which is equal to 10E21. When I declare it as a "real" number, it appears to be 9.9999998E+021, and when I declare it as a "double precision" number, it appears to be 9.999999778196308E+021.

What should I do to make n=10E21 exactly.

[Steven_L_Intel1]

Double precision is good to only about 15 decimal digits. You could try quad precision (REAL(16)) which is good to 33 digits, but that will be a lot slower.

While integer values are exactly representable, assuming there are enough bits in the datatype, many decimal fractions are not.

[jimdempseyatthecove]

Keremusu,

You might find this insightful

program literals  real(8) :: a, b
  a = 10E21
  b = 10E21_8  write(*,*) a,b
end program literals
Jim Dempsey

[Steven_L_Intel1]

I would also recommend the three-part series The Perils of Real Numbers.in the Newsletter thread.