Today I needed a simple algorithm for checking if a number is a power of 2.
The algorithm needs to be:
ulong
value.I came up with this simple algorithm:
private bool IsPowerOfTwo(ulong number)
{
if (number == 0)
return false;
for (ulong power = 1; power > 0; power = power << 1)
{
// This for loop used shifting for powers of 2, meaning
// that the value will become 0 after the last shift
// (from binary 1000...0000 to 0000...0000) then, the 'for'
// loop will break out.
if (power == number)
return true;
if (power > number)
return false;
}
return false;
}
But then I thought, how about checking if log2 x
is an exactly round number? But when I checked for 2^63+1, Math.Log
returned exactly 63 because of rounding. So I checked if 2 to the power 63 is equal to the original number - and it is, because the calculation is done in double
s and not in exact numbers:
private bool IsPowerOfTwo_2(ulong number)
{
double log = Math.Log(number, 2);
double pow = Math.Pow(2, Math.Round(log));
return pow == number;
}
This returned true
for the given wrong value: 9223372036854775809
.
Is there a better algorithm?
This is another method to do it as well
package javacore;
import java.util.Scanner;
public class Main_exercise5 {
public static void main(String[] args) {
// Local Declaration
boolean ispoweroftwo = false;
int n;
Scanner input = new Scanner (System.in);
System.out.println("Enter a number");
n = input.nextInt();
ispoweroftwo = checkNumber(n);
System.out.println(ispoweroftwo);
}
public static boolean checkNumber(int n) {
// Function declaration
boolean ispoweroftwo= false;
// if not divisible by 2, means isnotpoweroftwo
if(n%2!=0){
ispoweroftwo=false;
return ispoweroftwo;
}
else {
for(int power=1; power>0; power=power<<1) {
if (power==n) {
return true;
}
else if (power>n) {
return false;
}
}
}
return ispoweroftwo;
}
}
A number is a power of 2 if it contains only 1 set bit. We can use this property and the generic function countSetBits
to find if a number is power of 2 or not.
This is a C++ program:
int countSetBits(int n)
{
int c = 0;
while(n)
{
c += 1;
n = n & (n-1);
}
return c;
}
bool isPowerOfTwo(int n)
{
return (countSetBits(n)==1);
}
int main()
{
int i, val[] = {0,1,2,3,4,5,15,16,22,32,38,64,70};
for(i=0; i<sizeof(val)/sizeof(val[0]); i++)
printf("Num:%d\tSet Bits:%d\t is power of two: %d\n",val[i], countSetBits(val[i]), isPowerOfTwo(val[i]));
return 0;
}
We dont need to check explicitly for 0 being a Power of 2, as it returns False for 0 as well.
OUTPUT
Num:0 Set Bits:0 is power of two: 0
Num:1 Set Bits:1 is power of two: 1
Num:2 Set Bits:1 is power of two: 1
Num:3 Set Bits:2 is power of two: 0
Num:4 Set Bits:1 is power of two: 1
Num:5 Set Bits:2 is power of two: 0
Num:15 Set Bits:4 is power of two: 0
Num:16 Set Bits:1 is power of two: 1
Num:22 Set Bits:3 is power of two: 0
Num:32 Set Bits:1 is power of two: 1
Num:38 Set Bits:3 is power of two: 0
Num:64 Set Bits:1 is power of two: 1
Num:70 Set Bits:3 is power of two: 0
Find if the given number is a power of 2.
#include <math.h>
int main(void)
{
int n,logval,powval;
printf("Enter a number to find whether it is s power of 2\n");
scanf("%d",&n);
logval=log(n)/log(2);
powval=pow(2,logval);
if(powval==n)
printf("The number is a power of 2");
else
printf("The number is not a power of 2");
getch();
return 0;
}
return ((x != 0) && !(x & (x - 1)));
If x
is a power of two, its lone 1 bit is in position n
. This means x – 1
has a 0 in position n
. To see why, recall how a binary subtraction works. When subtracting 1 from x
, the borrow propagates all the way to position n
; bit n
becomes 0 and all lower bits become 1. Now, since x
has no 1 bits in common with x – 1
, x & (x – 1)
is 0, and !(x & (x – 1))
is true.
bool IsPowerOfTwo(int n)
{
if (n > 1)
{
while (n%2 == 0)
{
n >>= 1;
}
}
return n == 1;
}
And here's a general algorithm for finding out if a number is a power of another number.
bool IsPowerOf(int n,int b)
{
if (n > 1)
{
while (n % b == 0)
{
n /= b;
}
}
return n == 1;
}
I see many answers are suggesting to return n && !(n & (n - 1)) but to my experience if the input values are negative it returns false values. I will share another simple approach here since we know a power of two number have only one set bit so simply we will count number of set bit this will take O(log N) time.
while (n > 0) {
int count = 0;
n = n & (n - 1);
count++;
}
return count == 1;
Check this article to count no. of set bits
Example
0000 0001 Yes
0001 0001 No
Algorithm
Using a bit mask, divide NUM
the variable in binary
IF R > 0 AND L > 0: Return FALSE
Otherwise, NUM
becomes the one that is non-zero
IF NUM = 1: Return TRUE
Otherwise, go to Step 1
Complexity
Time ~ O(log(d))
where d
is number of binary digits
Some sites that document and explain this and other bit twiddling hacks are:
And the grandaddy of them, the book "Hacker's Delight" by Henry Warren, Jr.:
As Sean Anderson's page explains, the expression ((x & (x - 1)) == 0)
incorrectly indicates that 0 is a power of 2. He suggests to use:
(!(x & (x - 1)) && x)
to correct that problem.
bool isPowerOfTwo(int x_)
{
register int bitpos, bitpos2;
asm ("bsrl %1,%0": "+r" (bitpos):"rm" (x_));
asm ("bsfl %1,%0": "+r" (bitpos2):"rm" (x_));
return bitpos > 0 && bitpos == bitpos2;
}
Improving the answer of @user134548, without bits arithmetic:
public static bool IsPowerOfTwo(ulong n)
{
if (n % 2 != 0) return false; // is odd (can't be power of 2)
double exp = Math.Log(n, 2);
if (exp != Math.Floor(exp)) return false; // if exp is not integer, n can't be power
return Math.Pow(2, exp) == n;
}
This works fine for:
IsPowerOfTwo(9223372036854775809)
int isPowerOfTwo(unsigned int x)
{
return ((x != 0) && ((x & (~x + 1)) == x));
}
This is really fast. It takes about 6 minutes and 43 seconds to check all 2^32 integers.
for any power of 2, the following also holds.
NOTE: fails for n=0 , so need to check for it
Reason why this works is:
-n is the 2s complement of n. -n will have every bit to the left of rightmost set bit of n flipped compared to n. For powers of 2 there is only one set bit.
bool IsPowerOfTwo(ulong x)
{
return x > 0 && (x & (x - 1)) == 0;
}
In C, I tested the i && !(i & (i - 1)
trick and compared it with __builtin_popcount(i)
, using gcc on Linux, with the -mpopcnt flag to be sure to use the CPU's POPCNT instruction. My test program counted the # of integers between 0 and 2^31 that were a power of two.
At first I thought that i && !(i & (i - 1)
was 10% faster, even though I verified that POPCNT was used in the disassembly where I used__builtin_popcount
.
However, I realized that I had included an if statement, and branch prediction was probably doing better on the bit twiddling version. I removed the if and POPCNT ended up faster, as expected.
Results:
Intel(R) Core(TM) i7-4771 CPU max 3.90GHz
Timing (i & !(i & (i - 1))) trick
30
real 0m13.804s
user 0m13.799s
sys 0m0.000s
Timing POPCNT
30
real 0m11.916s
user 0m11.916s
sys 0m0.000s
AMD Ryzen Threadripper 2950X 16-Core Processor max 3.50GHz
Timing (i && !(i & (i - 1))) trick
30
real 0m13.675s
user 0m13.673s
sys 0m0.000s
Timing POPCNT
30
real 0m13.156s
user 0m13.153s
sys 0m0.000s
Note that here the Intel CPU seems slightly slower than AMD with the bit twiddling, but has a much faster POPCNT; the AMD POPCNT doesn't provide as much of a boost.
popcnt_test.c:
#include "stdio.h"
// Count # of integers that are powers of 2 up to 2^31;
int main() {
int n;
for (int z = 0; z < 20; z++){
n = 0;
for (unsigned long i = 0; i < 1<<30; i++) {
#ifdef USE_POPCNT
n += (__builtin_popcount(i)==1); // Was: if (__builtin_popcount(i) == 1) n++;
#else
n += (i && !(i & (i - 1))); // Was: if (i && !(i & (i - 1))) n++;
#endif
}
}
printf("%d\n", n);
return 0;
}
Run tests:
gcc popcnt_test.c -O3 -o test.exe
gcc popcnt_test.c -O3 -DUSE_POPCNT -mpopcnt -o test-popcnt.exe
echo "Timing (i && !(i & (i - 1))) trick"
time ./test.exe
echo
echo "Timing POPCNT"
time ./test-opt.exe
in this approach , you can check if there is only 1 set bit in the integer and the integer is > 0 (c++).
bool is_pow_of_2(int n){
int count = 0;
for(int i = 0; i < 32; i++){
count += (n>>i & 1);
}
return count == 1 && n > 0;
}
Here is another method I devised, in this case using |
instead of &
:
bool is_power_of_2(ulong x) {
if(x == (1 << (sizeof(ulong)*8 -1) ) return true;
return (x > 0) && (x<<1 == (x|(x-1)) +1));
}
Mark gravell suggested this if you have .NET Core 3, System.Runtime.Intrinsics.X86.Popcnt.PopCount
public bool IsPowerOfTwo(uint i)
{
return Popcnt.PopCount(i) == 1
}
Single instruction, faster than (x != 0) && ((x & (x - 1)) == 0)
but less portable.
This program in java returns "true" if number is a power of 2 and returns "false" if its not a power of 2
// To check if the given number is power of 2
import java.util.Scanner;
public class PowerOfTwo {
int n;
void solve() {
while(true) {
// To eleminate the odd numbers
if((n%2)!= 0){
System.out.println("false");
break;
}
// Tracing the number back till 2
n = n/2;
// 2/2 gives one so condition should be 1
if(n == 1) {
System.out.println("true");
break;
}
}
}
public static void main(String[] args) {
// TODO Auto-generated method stub
Scanner in = new Scanner(System.in);
PowerOfTwo obj = new PowerOfTwo();
obj.n = in.nextInt();
obj.solve();
}
}
OUTPUT :
34
false
16
true
Here's a simple C++ solution:
bool IsPowerOfTwo( unsigned int i )
{
return std::bitset<32>(i).count() == 1;
}
The following addendum to the accepted answer may be useful for some people:
A power of two, when expressed in binary, will always look like 1 followed by n zeroes where n is greater than or equal to 0. Ex:
Decimal Binary
1 1 (1 followed by 0 zero)
2 10 (1 followed by 1 zero)
4 100 (1 followed by 2 zeroes)
8 1000 (1 followed by 3 zeroes)
. .
. .
. .
and so on.
When we subtract 1
from these kind of numbers, they become 0 followed by n ones and again n is same as above. Ex:
Decimal Binary
1 - 1 = 0 0 (0 followed by 0 one)
2 - 1 = 1 01 (0 followed by 1 one)
4 - 1 = 3 011 (0 followed by 2 ones)
8 - 1 = 7 0111 (0 followed by 3 ones)
. .
. .
. .
and so on.
Coming to the crux
What happens when we do a bitwise AND of a number
x
, which is a power of 2, andx - 1
?
The one of x
gets aligned with the zero of x - 1
and all the zeroes of x
get aligned with ones of x - 1
, causing the bitwise AND to result in 0. And that is how we have the single line answer mentioned above being right.
So, we have a property at our disposal now:
When we subtract 1 from any number, then in the binary representation the rightmost 1 will become 0 and all the zeroes to the left of that rightmost 1 will now become 1.
One awesome use of this property is in finding out - How many 1s are present in the binary representation of a given number? The short and sweet code to do that for a given integer x
is:
byte count = 0;
for ( ; x != 0; x &= (x - 1)) count++;
Console.Write("Total ones in the binary representation of x = {0}", count);
Another aspect of numbers that can be proved from the concept explained above is "Can every positive number be represented as the sum of powers of 2?".
Yes, every positive number can be represented as the sum of powers of 2. For any number, take its binary representation. Ex: Take number 117
.
The binary representation of 117 is 1110101
Because 1110101 = 1000000 + 100000 + 10000 + 0000 + 100 + 00 + 1
we have 117 = 64 + 32 + 16 + 0 + 4 + 0 + 1
return (i & -i) == i
bool isPow2 = ((x & ~(x-1))==x)? !!x : 0;
private static bool IsPowerOfTwo(ulong x)
{
var l = Math.Log(x, 2);
return (l == Math.Floor(l));
}
After posting the question I thought of the following solution:
We need to check if exactly one of the binary digits is one. So we simply shift the number right one digit at a time, and return true
if it equals 1. If at any point we come by an odd number ((number & 1) == 1
), we know the result is false
. This proved (using a benchmark) slightly faster than the original method for (large) true values and much faster for false or small values.
private static bool IsPowerOfTwo(ulong number)
{
while (number != 0)
{
if (number == 1)
return true;
if ((number & 1) == 1)
// number is an odd number and not 1 - so it's not a power of two.
return false;
number = number >> 1;
}
return false;
}
Of course, Greg's solution is much better.
Source: Stackoverflow.com