Algorithm to find a solution for A xor X = B + X

Question

Given integer A and B, find integer X so that:

• A,B < 2*1e18
• A xor X = B + X

I highly doubt it is possible to solve this equation using maths. This is a coding problem I came across 3 years ago and even now I can't solve this myself.

My code so far: (this is the brute force solution)

#include <iostream>

using namespace std;

int main()
{

    unsigned long long a, b;
    cin >> a >> b;
    for (unsigned long long x = 1; x < max(a, b); x++) {
        unsigned long long c = a ^ x;
        unsigned long long d = b + x;
        if (c == d) {
            cout << x << endl;
            break;
            return 0;
        }
    }

    cout << -1; //if no such integer exists

    return 0;
}

Answer 1

Note that A + X == (A xor X) + ((A and X)<<1). So:

A xor X = A + X - ((A and X)<<1) = B + X
A - B = (A and X)<<1

And we have:

(A - B) and not (A<<1) = 0    (All bits in (A - B) are also set in (A<<1))
(A - B)>>1 = A and X

If the condition is met, for any integer Y that doesn't have bits that are set in A, (((A - B)>>1) or Y) is a solution. If you want just one solution, you could use ((A - B)>>1), where Y = 0. Otherwise there is no solution.

int solve(int a, int b){
    int x = (a - b) >> 1;
    if ((a ^ x) == b + x)
        return x;
    else
        return ERROR;
}

Answer 2

It's not very hard, you just need to think small: suppose we are writing A, B and X in binary and Aᵢ is the value corresponding to the rightmost 2ⁱ bit.

We know that: Aₒ ⊕ Xₒ = Bₒ + Xₒ.

Let's use an example to discover how to evaluate that: A = 15 and B = 6. Converting to binary:

A = 1 1 1 1           B = 0 1 1 0
X = a b c d           X = a b c d

Now we have some possibilities. Let's analyse the rightmost bits of A and B:

1 ⊕ d = 0 + d

We know that d can only be 0 or 1, so:

for d = 0
1 ⊕ d = 0 + d    =>    1 ⊕ 0 = 0 + 0    =>    1 = 0 (not possible)

for d = 1
1 ⊕ d = 0 + d    =>    1 ⊕ 1 = 0 + 1    =>    0 = 1 (not possible)

It's noticeable that XOR behaves just like binary sum (with the difference that XOR doesn't create a carryover for the next bit sum):

    XOR           SUM
0 ⊕ 0 = 0  |   0 + 0 = 0
0 ⊕ 1 = 1  |   0 + 1 = 1
1 ⊕ 0 = 1  |   1 + 0 = 1
1 ⊕ 1 = 0  |   1 + 1 = 0

so it won't be always possible to find a X that satisfies A ⊕ X = B + X, because there isn't a value d that satisfies 1 + d = 0 + d.

Anyway, if X exists, you can just find it out this way, from right to left, finding bit by bit.

---

WORKING FULL EXAMPLE

A = 15, B = 7:

A = 1 1 1 1           B = 0 1 1 1
X = a b c d           X = a b c d

1 ⊕ d = 1 + d 

Here, both d = 0 and d = 1 apply, then what? We need to check the next bit. Suppose d = 1:

A = 1 1 1 1           B = 0 1 1 1
X = a b c d           X = a b c d

1 ⊕ d = 1 + d    =>    1 ⊕ 1 = 1 + 1    =>    0 = 0 (possible)

BUT 1 + 1 = 0 generates a carryover for the next bit sum:

Instead of 1 ⊕ c = 1 + c, we have 1 ⊕ c = 1 + c (+1) =
                                   1 ⊕ c = c  (not possible)

so in this case, d must be 0.

carryover                              0
         A = 1 1 1 1           B = 0 1 1 1
         X = a b 0 0           X = a b 0 0
        -----------------------------------
                   0                     0

we know that c must be 0:

carryover                            0 0
         A = 1 1 1 1           B = 0 1 1 1
         X = a b 0 0           X = a b 0 0
        -----------------------------------
                 1 1                   1 1

but what about b? we need to check the next bit, as always:

if b = 0, there won't be a carryover, so we'll have:

1 ⊕ a = 0 + a  (and this is not possible)

so we try b = 1:

1 ⊕ b = 1 + b    =>    1 ⊕ 1 = 1 + 1    =>    0 = 0 (with carryover)

and now, for a:

carryover                          1 0 0
         A = 1 1 1 1           B = 0 1 1 1
         X = a 1 0 0           X = a 1 0 0
        -----------------------------------
               0 0 0                 0 0 0


1 ⊕ a = 0 + a (+1)    =>    1 ⊕ a = 1 + a

here a can be 0 and 1, but it must be 0, in order to avoid a carryover in the sum B + X.

Then, X = 0 1 0 0, thus X = 4.

---

CODE

#include <iostream>
using namespace std;

inline int bit(int a, int n) {
    if(n > 31) return 0; 
    return (a & ( 1 << n )) >> n; 
}

int main(){
    int A = 19;
    int B = 7;

    int X = 0;
    int carryover = 0;
    int aCurrent, aNext, bCurrent, bNext;

    for(int i = 0; i < 32; i++){
        aCurrent =  bit(A, i);      bCurrent =  bit(B, i);
        aNext =     bit(A, i + 1);  bNext =     bit(B, i + 1);

        if(aCurrent == 0 && bCurrent == 0){
            if(carryover) {X = -1; break;}
            if(aNext != bNext){
                X += 1 << i;
            }
            carryover = 0;
        }
        else if(aCurrent == 0 && bCurrent == 1){
            if(!carryover) {X = -1; break;}
            if(aNext == bNext){
                X += 1 << i;
            }
            carryover = 1;
        }
        else if(aCurrent == 1 && bCurrent == 0){
            if(!carryover) {X = -1; break;}
            if(aNext != bNext){
                X += 1 << i;
                carryover = 1;
            }
            else {
                carryover = 0;
            }
        }
        else if(aCurrent == 1 && bCurrent == 1){
            if(carryover) {X = -1; break;}
            if(aNext != bNext){
                X += 1 << i;
                carryover = 1;
            }
            else {
                carryover = 0;
            }
        }

    }

    if(X != -1) cout<<"X = "<<X<<endl;
    else cout<<"X doesnt exist"<<endl;

    return 0;
}

You can test it here.