Eliminating the last one from binary number

Provided the Number is not zero Then

// representing number in some format  
//A contains the bits before the last one & B is after last one
//small c after variable represent Compliments of that number like N & Nc
Num=A1B
//Some Property -N=Nc+1  Nc is compliment of N
(-Num) =(A1B)c+1=Ac0Bc+1=Ac1B //{ 0Bc since Bc is all bit 1 adding 1 will give 1B}

Num= A1B
-Num=Ac1B
//if we Ac&A will be zero and B&B will be zero too.. so , so remaining will be the last one
(Num&(-Num))   //leave last single '1' 

Try this on you PC..