Introduction To Scientific Computing Part II: C and C++

Introduction to Scientic Computing
Part II: C and C++

C. David Sherrill
School of Chemistry and Biochemistry
Georgia Institute of Technology
The C Programming Language:
Low-level operators
Created by Dennis Ritchie for the DEC PDP-11 UNIX
operating system, 1970s
ANSI standard 1983
A superset called C++ for object-oriented programming
(Stroustrup, 1980+).
C/C++ currently dominant programming language (used
to write, e.g., operating systems)
C and Scientic Computing
Scientic computing was traditionally done with Fortran. C
was slow to catch on during the 1980s. C/C++ taken more
seriously as scientic programs became more complex.
Ab initio programs:
Gaussian-9x: Maybe 60% Fortran, 40% C??
Q-Chem 2.0: 10% C++, 50% C, 40% Fortran
PSI 3.0: 20% C++, 60% C, 20% Fortran
NWChem: C and Fortran
MPQC: 100% C++ (and Curt++ !)
A Minimal C Program
#include <stdio.h>
main()
{
printf("Hello, world!\n");
}
N.b. This is not good ANSI C! Strictly speaking, should have
types.
C is a Typed Language
All variables (and functions) must have a given type. Some
allowed types:
int integer
oat oating point number, single-precision (bad)
double oating point number, double-precision (good)
char character value
FILE le structure
void A weird catch-all meaning nothing or anything
In C++, you make up your own data types!
An ANSI Approved Hello World
#include <stdio.h>
int main(void)
{
printf("Hello, world!\n");
return(0); /* exit w/ success status */
}
Compiling a Program
To compile a simple C program in UNIX, you invoke the C
compiler (usually named cc) like this:
cc hello-world.c -o hello-world
This would compile a C le called hello-world.c (containing
the previous example, perhaps) and make an executable
program called hello-world. The executable need not have
the same name as the program le. If no name is given by the
-o switch, the program will be named a.out by default.
To compile a C++ program, one would use the C++ compiler
[often named cpp, g++ (GNU), or xlC (IBM)].
To compile two C les into one executable, one rst compiles
the C source into object les
cc -c hello-world.c other-file.c
creating hello-world.o and other-file.o. The object les
are linked into the nal executable:
cc -o hello-world hello-world.o other-file.o
Often this can be done all in one step as a shortcut like this:
cc -o hello-world hello-world.c other-file.c
Linking Libraries
Frequently one wishes to call standard library functions, such
as the square root function sqrt() from the math library, etc.
These libraries are special les with names ending in a .a sux
(the a stands for archive). Names of libraries usually start
with the previx lib, as in libm.a, the C math library.
To link against a library one uses the -l ag. The math library
can be included by a command like:
cc hello-world.c -o hello-world -lm
The -l ag automatically adds a lib prex and .a sux to
determine the library name.
Makeles for Large Programs
Programs containing more than a few source code les are
best compiled using a special program called make. The make
command reads a le called Makefile to determine how to
compile the program, what libraries to link, etc. An example
follows:
ROOT = /home/users/sherrill/C
LIBS = -L$(ROOT)/lib -lm -lds_io -lds_str
CFLAGS = -I$(ROOT)/include -O
CC = cc
NOBJ = biblio.o cparse.o format.o
SRC = $(NOBJ:%.o=%.c)
biblio: $(NOBJ)
$(CC) $(CFLAGS) $(NOBJ) $(LIBS) -o biblio
clean:
/bin/rm -f $(NOBJ)
# DO NOT DELETE THIS LINE -- make depend depends on it
biblio.o: biblio.c
cparse.o: cparse.c
format.o: format.c
A More Complex Program Example
#include <stdio.h>
main()
{
double x, y;
double crazy_function(double x);
x = 4.0;
y = crazy_function(x);
printf("The result is %lf\n", y);
}
double crazy_function(double x)
{
double z;
x = x * x;
z = x + 1.0;
return(z);
}
When run, the program prints
The result is 17.000000
Pass-by-Value
Suppose we modied the previous example as such:
x = 4.0;
printf("The value of x is %lf\n", x);
Whats x ? You might think 16.0, but its 4.0. How could you
modify crazy function so x would really be changed by it?
Solution I: C Pointers
y = crazy_function(&x);
double crazy_function(double *x)
{
double z;
*x = *x * *x;
z = *x + 1.0;
return(z);
}
Solution I: C++ References
double crazy_function(double &x)
{
double z;
x = x * x;
z = x + 1.0;
return(z);
}
Exactly same as original except for declaration of crazy function.
#include <stdio.h>
#include <math.h>
main()
{
double x, y;
x = 4.0;
y = sqrt(x);
}
The result is 2.000000
Why didnt we need to declare sqrt?
Subroutines Also Called Functions
main()
{
double x;
void dumb_subroutine(double x);
x = 4.0;
dumb_subroutine(x);
}
void dumb_subroutine(double y)
{
}
Whats So Great About C++ ?
Retains C as a subset
Has nice new featues like references, constants, and
especially user-dened datatypes or objects
Object-oriented language: seen as big advantage for very
large codes
Not very ecient; may need to do computationally intensive
subroutines in C or Fortran (or call optimized math library
like BLAS)
Object-Oriented Programming and C++
Contrasts to procedural programming
The program consists of objects which know how to relate
to each other
Objects hide their own data and can only be accessed
through their interfaces keeps others from messing up
your beautiful code
Separation of interface from implementation makes up-
grades easier
C++ programmers tend to write insanely complex code;
natural result of taking object ideas to their limit
Suggested Reading
The C Programming Language, 2nd ed., Brian W. Kernighan
and Dennis M. Ritchie (Prentice Hall, Englewood Clis, NJ,
1988).
The C++ Programming Language, 3rd ed., Bjarne Strous-
trup (Addison-Wesley, Reading, MA, 1997).

Introduction To Scientific Computing Part II: C and C++

Uploaded by

Copyright:

Available Formats

Introduction To Scientific Computing Part II: C and C++

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Introduction To Scientific Computing Part II: C and C++

Uploaded by

Copyright:

Available Formats

Introduction to Scientic Computing

Part II: C and C++

You might also like