a few bugs while executing the program - c

I want to write a program to see if an integer is the power of another integer(true return 1 and false return 0). And the code is as follows:
#include <stdio.h>
#include <math.h>
int cal_base_power(int);
int main()
{
int x,r;
printf("Please input an integer\n");
scanf("%d\n",&x);
r=cal_base_power(x);
printf("result is %d\n",r);
}
int cal_base_power(int input)
{
int i=2;
double a=pow(input,(double)1/i);
while (a>=2)
{
if ((double)a==(int)a)
return 1;
i++;
a=pow(input,(double)1/i);
}
return 0;
}
It is ok with 4,8,125 some cases like these. But failed when input 216 and 343. Also it will not automatically output 0 and 1. I must input some random characteristics before the result 0 or 1 comes out.
Can anyone help me? I know it is quite easy. But I really need your help

You can't do equality comparisons on floating-point.
(double)a==(int)a
Due to round-off error, a might not be exactly an integer even though it's supposed to be.
EDIT:
There's two ways to do this:
Allow tolerance on the comparison: fabs(a - (int)a) < 0.0001 (or something like that, you'll need to tweak the threshold.)
Round a to the nearest integer and power it back up (using integers only) to see if it matches the input.

Related

why does it show segmentation fault when I tried to use recursion here?

I tried to write a code to calculate how many 1 are there in a number's binary form. This is my code:
#include <stdio.h>
#include <math.h>
static int num = 0;
void binary(int target){
int n = 0;
int a = 1;
if(target != 0){
while(target >= a ){
n++;
a = pow(2, n);
}
a = pow(2, n-1);
num++;
binary(target - a);
}
}
int main() {
int target = 0;
scanf("%d", &target);
binary(target);
printf("%d",num);
return 0;
}
However, this shows segmentation fault when I run it. I don't know where has the code tried to access memories that are not allowed. I figured it might have something to do with the recursion in the binary function. Can anyone tell me what have caused the segmentation fault here? Thank you so much. I really can't understand segfaults :(
As mentioned in the comments, pow is not a good candiate here due to its signature:
double pow(double x, double y);
so any place that you are using pow, you are implicitly using floating point numbers. I was able to cause a segfault with the input 1<<31 which is the value -2147483648. This will cause your loop to terminate with n=0, a=1. You then set a = pow(2, -1), but since a is an integer, this gets floored down to just 0. You then recurse with binary(target - 0) which might as well just be binary(target) again, hence you have an infinite call with no termination.
I'll also leave as a note that recursion for this type of problem is probably not the right tool, unless your goal is to learn about recursion. There is a much more concise and reliable method via a loop and the & operator. I would also suggest using unsigned values to avoid issues like this with negative terms.

Appropriate data type to be used in a C program

I'm trying to run a program that does certain operations on factorials of large numbers (say 50!; viz 3.041e+64 - huge!) and therefore doesn't fit in the normal int data types that I'm aware of(unsigned long long int etc)
Which data type do I use to store these values?
P.S I was trying to find the trailing zeroes in a factorial. The following was my approach:
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
int FactorialFinder(int a)
{
if (a>1)
a= a* (FactorialFinder(a-1)) ;
return a;
}
int main()
{
printf("Enter number \n");
int num ;
scanf("%d",&num) ;
printf("number is %d\n",num);
printf("Factorial is %d",(num = FactorialFinder(num))) ;
int x=0, count = 0 ;
while(num>0)
{
x = (num%10) ;
if (x == 0)
count++ ;
else
break;
num= num/10 ;
}
printf("\nNumber of trailing zeroes is %d",count) ;
getchar() ;
return 0;
}
Works fine upto 12! beyond which the results are erroneous (from 17! it starts returning negative factorial values(?), from 34! it gives 0) I'm guessing due to the datatype problem. Can someone help me out?
Well these kind of numbers cannot be stored as single number rightly as you've reasoned there are no data types to hold them. The best way to work with large numbers is to store them as arrays of int type or char type.
for example you can store 1234567898765 as an array int big[14] where,
big[0]=1
big[1]=2
.
.
.
big[13]=5 //last element
big[14]=-1 //to mark the end of number...
or in the reverse order with -1 as last element (select which is is convenient for your implementation)
and now comes the challenging part where you have to create functions for addition, subtraction, multiplication and other operations which you require.. There are many ways of implementing these function..give a try.. or you can just look up how to do them here's a source : click
this provides a implementation of arithmetic numbers for up to 100 digit numbers however you can try to build one which can deal with even larger numbers :)

Why am I getting this strange output on this simple C program? [closed]

Closed. This question needs details or clarity. It is not currently accepting answers.
Want to improve this question? Add details and clarify the problem by editing this post.
Closed 8 years ago.
Improve this question
I am working on a personal project where one part of it deals with counting squares and cubes under a certain bound (in this case 10,000). So, I wrote a simple C program I thought would work to verify my results. Here is the little program I put together to see all of the cubes:
#include <stdlib.h>
#include <stdio.h>
#include <math.h>
int main() {
double i;
int cubes = 0;
for (i = 1; i < 10000; i++) {
if ( i == cbrt(i) * cbrt(i) * cbrt(i) ) {
printf("%f --- %f\n",i, cbrt(i));
cubes++;
}
}
printf("%i\n", cubes);
return 0;
}
I got the (incorrect) output : 24. If you want to look at this see the problem look at numbers 15 and 20 on the output. Why I am getting the wrong answer (the correct answer is 21) is an entirely different matter. My question arose when I was messing around with my code to try and fix this and I temporarily changed it to this:
int main() {
double i;
int cubes = 0;
for (i = 1; i < 10000; i++) {
double temp = (cbrt(i) * cbrt(i) * cbrt(i));
if ( i == temp ) {
printf("%f -> %f\n", i, temp);
cubes++;
}
}
printf("%i\n", cubes);
return 0;
}
Now, the program is printing every number between 1 and 9999. So, am I missing something ridiculously easy or what is going on? All I did was instead of having cbrt(i)*cbrt(i)*cbrt(i) in the if conditional I set a double variable equal to result and placed that in the conditional. Why is my program doing this?
I am not sure why this got down voted. I feel like this is a legitimate question. Sorry S.O. community...
double cbrt(double x) returns the closest representable cubic root of x.
The inexactness of the result, then cubed, may not exactly equal 'x' again.
Why 2nd program differs:
C is not obliged to perform double math only to double precision. It may use wider (long double). Depending on many things, the 2nd code appears to have done more in long double than the first. With the extra precision, its easy to see that the results, rounded to double appear exact.
C11dr ยง5.2.4.2.2 9 Except for assignment and cast (which remove all extra range and precision), the values yielded by operators with floating operands and values subject to the usual arithmetic conversions and of floating constants are evaluated to a format whose range and precision may be greater than required by the type.
Why a typical program run (of either code) produces a result of about 3333.
Consider the double numbers from 2 to 4 and 8 to 64. double numbers are logarithmically distributed. There are as many different double from 2 to 4 as 8 to 16 as 16 to 32 as 32 to 64.
So now all 3 sets from 8 to 64 have a cube root of some answer in the 1 set of 2 to 4. Now if we cube the numbers 2 to 4, we get answers in the range 8 to 64. 1 set of numbers mapping into 3 sets. The round trip is not exact. See Pigeonhole principle. IOW: On average, 3 numbers in the range 8 to 64 have the same cubic root. Then the cube of that root will be 1 of the 3 original.
To find the count of the perfect integer cubes 0 to N
unsigned Perfect_Cube_Count(unsigned n) {
if (n == 0)
return 1;
unsigned i;
// overflow not possible
for (i = 0; i*i < n/i; i++);
return i;
}
Or
// valid for 0 <= x <= something_well_over_1e9
double Perfect_Cube_Count_d(double x) {
double y = cbrt(x);
return floor(y) + 1;
}
You probably want, as Andrew guessed, whole-number cube roots. Float math is quite tricky because of rounding errors. Generally you cannot rely on equality but must compare with an error margin.
To solve your problem though I'd construct the 21 cubes beforehand and then iterate over integers, comparing against the pre-constructed cubes. Or is that cheating? ;-)
In Samuel Becket's novel Watt there is a chapter about a Scottish "Math genius" who could in his head compute all integer third roots of integer cubes up to 10000 or so, too!
My uess, is your compiler does an optimization in the second case, eli inating cbrt calls. It just says the result of cbrt is strictly defined by the standard, so it might as well be always thte case that (i == temp)
You can twak this by some command line arguments, and force it to do exactly what is written in the code. As I recall, this should thhe default thing to do for C compilers regarding float arthimetic, but your compiler may think it is smarter than you or something.
EDIT
And yes, this code has nothing to do with finding perfect cubes...
EDIT
Totally not an answer to the question, but as a quick exercise, this I wrote this:
#include <stdlib.h>
#include <stdio.h>
#include <limits.h>
static unsigned long count_cubes(unsigned long max_n)
{
unsigned long n = 1;
while (n*n*n <= max_n) {
++n;
}
return n-1;
}
int main(int argc, char **argv)
{
unsigned long max_n;
char *p;
if (argc < 2) {
return EXIT_FAILURE;
}
max_n = strtoul(argv[1], &p, 10);
if (max_n < 1 || max_n == ULONG_MAX) {
return EXIT_FAILURE;
}
printf("%lu\n", count_cubes(max_n));
return EXIT_SUCCESS;
}
Note: no need for floating point arithmetic
EDIT
Sorry, I really got into this...
This one can be a bit faster:
#include <stdlib.h>
#include <stdio.h>
#include <limits.h>
#include <math.h>
static unsigned long count_cubes(unsigned long max_n)
{
unsigned long n;
if (max_n < 256) {
n = 1;
}
else {
n = cbrtl(max_n) - 1;
}
while (n*n*n <= max_n) {
++n;
}
return n-1;
}
int main(int argc, char **argv)
{
unsigned long max_n;
char *p;
if (argc < 2) {
return EXIT_FAILURE;
}
max_n = strtoul(argv[1], &p, 10);
if (max_n < 1 || max_n == ULONG_MAX) {
return EXIT_FAILURE;
}
printf("%lu\n", count_cubes(max_n));
return EXIT_SUCCESS;
}
EDIT ( last time, I promise... )
To show an explanation of my little loop above, starting at cbrt(max_n)-1, I tried the one suggested by #chux , here are some results with slightly larger numbers:
PerfectCubes(18446724184312856125) == 2642246
which is fine but also
PerfectCubes(18446724184312856125-10) == 2642246
which is totally not fine, since 18446724184312856125 == 2642245^3 , meaning there are 2642245 perfect cubes <= 18446724184312856125-10 .
This also results from inaccuracies in floating point representation. You can try it for yourself, if your computer is somewhat similar to mine:
printf("%f\n", cbrt( 2642245UL * 2642245UL * 2642245UL));
/* prints 2642245.000000 */
printf("%f\n", cbrt( 2642245UL * 2642245UL * 2642245UL - 10UL));
/* prints 2642245.000000 */
These two numbers clearly don't have the same cubic root, yet cbrt returns the same results. In this case, floor doesn't help either. Anyways, one always needs to be very careful using floating point arithmetics. And now I really should go to sleep.

Determining if a float has a fractional part?

Here is the problem: The game Totals can be played by any number of people. It starts with a total of 100 and each player in turn makes an integer displacement between -20 and 20 to that total. The winner is the player whose adjustment makes the total equal to 5. Using only the three variables given:
total
adjustment
counter
Here is what I have so far:
#include <stdio.h>
int main (void)
{
int counter=0;
float adj;
int ttl=100;
printf("You all know the rules now lets begin!!!\n\n\nWe start with 100. What is\n");
while (ttl!=5)
{
printf("YOUR ADJUSTMENT?");
scanf("%f",&adj);
counter++;
if (adj<=20 && adj>=-20)
{
ttl=ttl+adj;
printf("The total is %d\n",ttl);
}
else
{
printf ("I'm sorry. Do you not know the rules?\n");
}
}
printf("The game is won in %d steps!",counter);
}
What I need:
When a decimal number is entered it goes to the else. How do I determine if a float has a fractional part.
You can cast the float to an int and then compare it to your original variable. If they are the same there was no fractional part.
By using this method, there is no need for a temporary variable or a function call.
float adj;
....
if (adj == (int)adj)
printf ("no fractional!\n");
else
printf ("fractional!\n");
Explanation
Since an int cannot handle fractions the value of your float will be truncated into an int (as an example (float)14.3 will be truncated into (int)14).
When comparing 14 to 14.3 it's obvious that they are not the same value, and therefore "fractional!" will be printed.
#include <stdio.h>
#include <math.h>
int main ()
{
float param, fractpart, intpart;
param = 3.14159265;
fractpart = modff (param , &intpart);
return 0;
}
http://www.cplusplus.com/reference/clibrary/cmath/modf/
modff finds the fractional part, so I guess testing whether it's equal to 0 or null will answer your question.
if you want to know whether a real number x has no fractional part, try x==floor(x).
I am only learning C so tell me if I am wrong, please.
But if instead of using
scanf("%f",&adj);
if you use:
scanf("%d%d", &adj, &IsUndef);
Therefore if the user typed anything other than a whole integer &IsUndef would not equal NULL and must have a fractional part sending the user to else.
maybe.
Using scanf() is problematic. If the user typed -5 +10 -15 -15 on the first line of input, then hit return, you'd process the 4 numbers in turn with scanf(). This is likely not what you wanted. Also, of course, if the user types +3 or more, then the first conversion stops once the space is read, and all subsequent conversions fail on the o or or, and the code goes into a loop. You must check the return value from scanf() to know whether it was able to convert anything.
The read-ahead problems are sufficiently severe that I'd go for the quasi-standard alternative of using fgets() to read a line of data, and then using sscanf() (that extra s is all important) to parse a number.
To determine whether a floating point number has a fractional part as well as an integer part, you could use the modf() or modff() function - the latter since your adj is a float:
#include <math.h>
double modf(double x, double *iptr);
float modff(float value, float *iptr);
The return value is the signed fractional part of x; the value in iptr is the integer part. Note that modff() may not be available in compilers (runtime libraries) that do not support C99. In that case, you may have to use double and modf(). However, it is probably as simple to restrict the user to entering integers with %d format and an integer type for adj; that's what I'd have done from the start.
Another point of detail: do you really want to count invalid numbers in the total number of attempts?
#include <stdio.h>
#include <math.h>
int main(void)
{
int counter=0;
int ttl=100;
printf("You all know the rules now lets begin!!!\n"
"\n\nWe start with 100. What is\n");
while (ttl != 5)
{
char buffer[4096];
float a_int;
float adj;
printf("YOUR ADJUSTMENT?");
if (fgets(buffer, sizeof(buffer), stdin) == 0)
break;
if (sscanf("%f", &adj) != 1)
break;
if (adj<=20 && adj>=-20 && modff(adj, &a_int) == 0.0)
{
counter++; // Not counting invalid numbers
ttl += adj;
printf("The total is %d\n", ttl);
}
else
{
printf ("I'm sorry. Do you not know the rules?\n");
}
}
if (ttl == 5)
printf("The game is won in %d steps!\n", counter);
else
printf("No-one wins; the total is not 5\n");
return(0);
}
Clearly, I'm studiously ignoring the possibility that someone might type in more than 4095 characters before typing return.

Recursive harmonic function returns NaN

I have written the following sample code to find the harmonic value of N. (1+1/2+1/3+...1/N). Read the comments in the code written in BOLD and help me to find why is this happening.
#include <stdio.h>
float harmonic(float n, float har) {
if(n==0) {
return 0;
}
if(n==1) {
printf("%f\n", har+1.0f);***/* This prints value 1.5000*/***
return har+1.0f;
}else{
harmonic(n-1, (har+(1/n)));
}
}
int main()
{
printf("%f\n", harmonic(2, 0.0f)); **/* But this prints value nan(Not a Number)*/**
return 0;
}
Thanks,
Naga
I think you want to do:
return harmonic(n-1, (har+(1/n)));
My first thought was that you should almost never compare floats with simple equality so "if(n==0)" should be "if(n<=EPSILON)" and "if(n==1)" should be "if(n<= 1.0f + EPSILON)" where EPSILON is a small positive fraction, maybe 1.0e-5. Depends on how much precision you can depend on.
But then I realized that n should be an int. Cast it to a float before the division. As the comparisons with "n" stand you risk infinite recursion.
Consider using a double instead of a float.
Matthew Flaschen's answer gets to real reason you get the NaN message. The original code doesn't return anything from the "else" so the caller is probably reading garbage from the stack. Hence, the NaN.

Resources