PARAMETER attribute for data type

[ya0037]

Dear All,

I have two questions I hope someone could help me.

If I want to use parameter attribute for data type declaration there would be some problems, please look at the below example:

TYPE PEOPLE ! Defining derived type PEOPLE
INTEGER AGE
CHARACTER*20 NAME
END TYPE PEOPLE
TYPE(PEOPLE), Parameter :: SMITH = PEOPLE(25,'John Smith')

TYPE triplet
REAL, DIMENSION(3) :: vertex
END TYPE triplet
TYPE(triplet), PARAMETER :: t = triplet(1.0)

In this example there is no problem but if I want to give different value to the data type variable what should I do??

TYPE triplet
REAL, DIMENSION(3) :: vertex
END TYPE triplet
TYPE(triplet), PARAMETER :: t = triplet(1.0, 2.0, 3.0)

I mean how should I give different value to the data type above :

t%vertex(1)=1.0
t%vertex(2)=2.0
t%vertex(3)=3.0


Second question:::

There are some papers in which there are some comparisons between FORTRAN and c++ speed. In all of them they stated that FORTRAN is faster than C++.

There it comes my question, if FORTRAN is faster why all the big software and open source projects like big MD (Molecular Dynamics) codes are changing toward c++, in these numerical codes speed is very important, but they gradually changing toward c++.

I am wondering should we change toward that!!!!!


Thank you very much.

I am really looking forward to hearing from you.

Regards,

 

[GerritGroot]

What do you mean by different values? If you want to change the values, leave away the word "PARAMETER

 

[ya0037]

I have told that,

I want data type parameter declaration that gives the following result : t%vertex(1)=1.0 and t%vertex(2)=2.0 and t%vertex(3)=3.0

 

[ya0037]

My second example will give error:

TYPE triplet
REAL, DIMENSION(3) :: vertex
END TYPE triplet
TYPE(triplet), PARAMETER :: t = triplet(1.0, 2.0, 3.0)

how should I correct it?????

If I give
TYPE(triplet), PARAMETER :: t = triplet(1.0)
then t%vertex(1)=1.0 and t%vertex(2)=1.0 and t%vertex(3)=1.0

But I want the different value:
t%vertex(1)=1.0 and t%vertex(2)=2.0 and t%vertex(3)=3.0

 

[ArkM]

Only 10% (or less) LOC of large computational program code performs math calculations (traditional Fortran area). There are massive data pre- and post-processing, complex process control, specific DBMSs, data viewing and other modules in these programs. C++ is a most comfortable and effective language in these problem areas. In essence it's system programming, not "a formula translation".

That's why so many serious "number-crunching" programs goes "from Fortran to C++".

That's a good example of a senseless statement: "the language A is faster than the language B". Underlying abstract Fortran and C++ machines are the same. Both languages are strong static typed. Classic CRAY-like processors (remember Fortran 90+ arrays vector-oriented stuff) are historical artifacts now. Modern cluster Unix-driven supercomputers - natural areal of C++ codes.

Good programs (written in C++ or Fortran) are faster than bad programs (written in Fortran or C++) - that's all. There are much more good C++ programmers in the world now...

There are tons of legacy Fortran codes. So Fortran (now borrowed most of C++ OOP features is immortal

 

[GerritGroot]

First of all, I'm happy to hear that there are people who think that fortran is immortal , I don't feel for learning C++

About the original question, it's a type incompatibility that gives the error.

PROGRAM Test

TYPE triplet
  REAL, DIMENSION(3) :: vertex
END TYPE triplet

TYPE(triplet), PARAMETER :: t=triplet(RESHAPE(SOURCE=(/1.0,2.0,3.0/),SHAPE=(/3/)))

WRITE(*,*)t%vertex(:)

END PROGRAM Test

Gives:

   1.0000000       2.0000000       3.0000000

As output

The fact that you can say t=triplet using the typename for assigning values was new to me, thanks for that!

 

[gullipe]

Hi all

I completely agree with GerritGroot, I´m happy to see that ArkM thinks that Fortran in immortal! I learned Fortran years before C was even invented and I have used it extensively in my work as an engineer and still do (although many simple calculations can be done in Excel). I have had to write a few C programs during the years, but still consider C inferior to Fortran in my area of work. Also these endless "{" and "}" are boring...

 

[ya0037]

Dear All,

Thank you all very much, I understand my mistake.

Instead of :

  TYPE(triplet), PARAMETER :: t = triplet(1.0, 2.0, 3.0)

I should simply use:

  TYPE(triplet), PARAMETER :: t = triplet((/1.0, 2.0, 3.0/))

Thanks a lot.

Best Regards,