Arithmetic problem

[Schizoid Man]

Hi,

I have a very strange arithmetic problem in C:

double t = 0.1;
int steps = 10;
double time_step = t / (double)steps;

I would expect the output of time_step to be 0.01000 (my output is of
the form %5.5f), but instead it a very large (and incorrect) negative
number.

Any ideas?

Thanks,
Schiz

[Richard Tobin]

In article <e5**********@g eraldo.cc.utexa s.edu>,
Schizoid Man <sc***@sf.com > wrote:

I have a very strange arithmetic problem in C:

Show us a complete program that has the problem.

-- Richard

[Schizoid Man]

Richard Tobin wrote:

Show us a complete program that has the problem.

Hi Richard,

Here's the complete program. Thank you.

(This is the problematic module - any of the values passed to this
function apart from steps, returns a very odd number)

#include "stdafx.h"
#include "binomialtree.h "

double treeNPV(double s, double k, double r, double v, double t, int steps)
{
double time_step, up, dn, rt, p, npv;
time_step = t / (double)(steps) ;
rt = exp(r*time_step );
up = exp(v*sqrt(time _step));
dn = 1/up;
p = (rt-dn)/(up-dn);

npv = 0.0;

//for (int i=0; i<=steps; i++)
npv = v;

return npv;
}

Main modules:

#include "stdafx.h"
#include "binomialtree.h "

int main(void)
{
double s, k, r, v, t;
double call_npv;
int steps;
printf("Enter asset price S: ");
scanf_s("%f", &s);
printf("Enter strike price K: ");
scanf_s("%f", &k);
printf("Enter riskfree rate R: ");
scanf_s("%f", &r);
printf("Enter volatility V: ");
scanf_s("%f", &v);
printf("Enter time to expiry T: ");
scanf_s("%f", &t);
printf("Enter number of steps: ");
scanf_s("%d", &steps);
call_npv = treeNPV(s,k,r,v ,t,steps);
printf("Call option NPV is: %5.5f", call_npv);
scanf_s("%d", &steps);
return 0;
}

Header files:

// stdafx.h : include file for standard system include files,

#pragma once
#define WIN32_LEAN_AND_ MEAN // Exclude rarely-used stuff from Windows
headers
#include <stdio.h>
#include <tchar.h>

// binomialtree.h
#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#define pi 3.1415926535897 93238

double treeNPV(double s, double k, double r, double v, double t, int steps);

[Richard Tobin]

In article <e5**********@g eraldo.cc.utexa s.edu>,
Schizoid Man <sc***@sf.com > wrote:

Here's the complete program. Thank you. scanf_s("%f", &s);
I don't know what scanf_s is. It's not a standard C function.
#include <tchar.h>

And I don't know what tchar.h is.

But if scanf_s is like scanf, then you should be using format %lf
to read into a double.

-- Richard

[Xicheng Jia]

Schizoid Man wrote:

Richard Tobin wrote:

Show us a complete program that has the problem.

Hi Richard,

Here's the complete program. Thank you.

(This is the problematic module - any of the values passed to this
function apart from steps, returns a very odd number)

#include "stdafx.h"
#include "binomialtree.h "

double treeNPV(double s, double k, double r, double v, double t, int steps)
{
double time_step, up, dn, rt, p, npv;
time_step = t / (double)(steps) ;
rt = exp(r*time_step );
up = exp(v*sqrt(time _step));
dn = 1/up;
p = (rt-dn)/(up-dn);

npv = 0.0;

//for (int i=0; i<=steps; i++)
npv = v;

return npv;
}

Main modules:

#include "stdafx.h"
#include "binomialtree.h "

int main(void)
{
double s, k, r, v, t;
double call_npv;
int steps;
printf("Enter asset price S: ");
scanf_s("%f", &s);

for "double", you may need to use %lf

scanf_s("%lf", &s);

Anyway, dont know what's scanf_s() in your code. there might have some
strange behaviers for some implementations by using scanf() statements
to accept inputs. but if scanf_s() is your self-defined function using
something like fget() and sscanf(), that should be OK:)..

BTW, what's the values of "steps" before and after you enter the
subroutine treeNPV(). gdb is always your friend to check this info. :-)

Xicheng

[Schizoid Man]

Richard Tobin wrote:

And I don't know what tchar.h is.

But if scanf_s is like scanf, then you should be using format %lf
to read into a double.

Hi Richard,

Absolutely right. %lf did the trick. I'm looking at C after ages - I
have a job interview coming up! :)

scanf_s is a method that replaces scanf in the new Visual C++, scanf has
been deprecated. Similarly, for the tchar.h file too.

On another note, can you tell me why the following method are not giving
me Combination results over 12? The following program works fine till 12
and then mysteriously blows up.

unsigned long int factorial(int x) {
if (x==0)
return 1;
else {
unsigned int y;
y = x;
for (int i=x-1; i>=1; i--)
y=y*i;
return y;
}
}

unsigned long int combin(int n, int r) {
if (r==0 || r==n)
return 1;
if (r==1 || r==n-1)
return n;
if (r>1 && r<n)
return factorial(n)/(factorial(r)*f actorial(n-r));
}

[Robert Gamble]

Richard Tobin wrote:

In article <e5**********@g eraldo.cc.utexa s.edu>,
Schizoid Man <sc***@sf.com > wrote:

Here's the complete program. Thank you.

scanf_s("%f", &s);

I don't know what scanf_s is.

This is part of a Microsoft initiative to make parts of the C library
"safer". There is now a Technical Report dedicated to this initiative
in the hopes of making it Standard some day. Check out
http://www.open-std.org/jtc1/sc22/wg...docs/n1146.pdf for the ugly
details. It is, IMNSHO, completely ass-backwards.
It's not a standard C function.

Let's hope it stays that way.

Robert Gamble

[Coos Haak]

Op Sat, 03 Jun 2006 18:01:37 -0700 schreef Schizoid Man:

Richard Tobin wrote:

And I don't know what tchar.h is.

But if scanf_s is like scanf, then you should be using format %lf
to read into a double.

Hi Richard,

Absolutely right. %lf did the trick. I'm looking at C after ages - I
have a job interview coming up! :)

scanf_s is a method that replaces scanf in the new Visual C++, scanf has
been deprecated. Similarly, for the tchar.h file too.

On another note, can you tell me why the following method are not giving
me Combination results over 12? The following program works fine till 12
and then mysteriously blows up.

12 factorial can fit in 32 bits. 13 factorial won't.
--
Coos

[Schizoid Man]

Schizoid Man wrote:

On another note, can you tell me why the following method are not giving
me Combination results over 12? The following program works fine till 12
and then mysteriously blows up.

Actually, not so mysterious after all - the unsigned int was the
culprit. Replacing that with a long double does the trick.

Thanks.

[Schizoid Man]

Coos Haak wrote:

On another note, can you tell me why the following method are not giving
me Combination results over 12? The following program works fine till 12
and then mysteriously blows up.

12 factorial can fit in 32 bits. 13 factorial won't.

Hi Coos,

I figured that out, albeit after I posted my message.

The function I'm trying to evaluate is Combin(n,r) = factorial(n) /
(factorial (n-r) * factorial (r)).

For some reason, C is blowing up for numbers like Combin(350,170) ,
whereas Excel is computing these large numbers quite easily. Any
suggestions?

Thanks.