What is the meaning of COE & COK in Fortran?

by yabi
Tags: fortran, meaning
 P: 21 In an old (1974) FORTRAN program, I have following two lines IF(FX.GT.COE(KOP,1)) GOTO 1 F=X*COK(KOP,1) I can't understand the meaning of COE and COK commands. Are they standard FORTRAN commands? PS) If you could kindly guide me to a site whee I can find list of fortran commands, I will be very happy to have it. BR Khoshravan
 Emeritus Sci Advisor HW Helper Thanks PF Gold P: 6,474 COE and COK are the names of two array variables. If you check elsewhere in the program, you should find them declared in a DIMENSION statement, giving the number of rows and columns in each array. Inside the parentheses, KOP and 1 reference a specific location in each array.
 P: 21 Dear SteamKing Thanks for your reply. You are correct. I am absolutely absent minded regarding this foolish question.
 P: 875 What is the meaning of COE & COK in Fortran? A dimension statement is not a must to declare an array; so, don't count on that, simply look for the variable name somewhere else...a simple search or a grep from the command line would do.
 Emeritus Sci Advisor HW Helper Thanks PF Gold P: 6,474 In a FORTRAN program from 1974, I'm pretty sure a DIMENSION statement is lurking somewhere in the program. FORTRAN IV or FORTRAN 66 was pretty particular about declaring array variables.
 P: 21 Really? I never have seen an array without dimension statement in the beginning. Could you please explain in detail how it could be possible to have arrays without using dimension statement?
 P: 875 program nodim integer a(2,2) a = 0 a(1,1) = 4 a(2,2) = 8 write(*,*) a end program nodim
 P: 21 Is the command integer a(2,2) compulsary? If yes, then it is sort of substitute for Dimension command!
 Emeritus Sci Advisor HW Helper Thanks PF Gold P: 6,474 The references a(1,1) and a(2,2) are setting one particular value within the array to the indicated constants. These references are not substitutes for a DIMENSION statement.
Mentor
P: 11,745
 Quote by yabi Is the command integer a(2,2) compulsary? If yes, then it is sort of substitute for Dimension command!
There are (at least) two ways to declare 'a' as a 2x2 array of integers:

integer a
dimension a(2,2)

which "declares as integer" and "declares as array" in separate statements; and

integer a(2,2)

which combines the two declarations.
 P: 21 I am not talking about a(1,1) = 4 a(2,2) = 8 But this one: integer a(2,2) Will your program working without integer command?
 Mentor P: 11,745 If you don't declare the 'a' as integer, the compiler will assume it's real (floating point), according to Fortran's default rule for implicit data types: names beginning with 'i' through 'n' are integer, others are real.
 P: 875 No, it will not work without a type declaration, must be declared something...character, integer or real; implicit typing does not with arrays.
 Mentor P: 11,745 The exact rules probably depend on which version of Fortran is being used. My own experience is mainly with Fortran 77, and I'm pretty sure that the default typing rules apply to arrays in that version. In 1974, of course Fortran 77 hadn't been officially established yet, but some compilers had "extensions" from Fortran 66 which allowed some features that were taken into Fortran 77.
 P: 21 Dear jtbell Thanks for your comments. So Integer will do what dimension do and without integer and dimension it is impossible to declare and array. I hope gsal also reads this comment.
Mentor
P: 21,258
 Quote by gsal No, it will not work without a type declaration, must be declared something...character, integer or real; implicit typing does not with arrays.
 Quote by yabi Dear jtbell Thanks for your comments. So Integer will do what dimension do and without integer and dimension it is impossible to declare and array. I hope gsal also reads this comment.
I don't think you understood what gsal wrote, which was that you need to declare the type of the array.
P: 875
 Quote by jtbell The exact rules probably depend on which version of Fortran is being used. My own experience is mainly with Fortran 77, and I'm pretty sure that the default typing rules apply to arrays in that version. In 1974, of course Fortran 77 hadn't been officially established yet, but some compilers had "extensions" from Fortran 66 which allowed some features that were taken into Fortran 77.
What I meant to say about "something" not applying to arrays was that arrays cannot benefit from implicit typing the way scalars do.

For example, while typing a fortran program and needing a new scalar in the middle of it, I can simply start using it right there and then without having to go back to the top and declare such variable; and, yes, if the name of the scalar variable starts with any of the letters from I to N, it will be an integer variable...I-Nteger, get it?

BUT, if I need an array variable in the middle of the program, I DO NEED to go back and do some kind of declaration, either a type declaration or a dimension declaration. If I start using an array variable without declaring, the compiler immediately complains about "Unexpected array reference".