You may want to have an iterative solution. Start from both ends and work your way inward toward 0.
#include <iostream>
#include <algorithm>
#include <vector>
#include <cmath>
size_t solution( const std::vector< int > & A )
{
std::vector< int >::const_iterator f( A.begin() );
std::vector< int >::const_reverse_iterator b( A.rbegin() );
size_t result = 0;
if( A.size() )
for( ; ( f != A.end() ) && ( b != A.rend() ); )
{
if( *f >= 0 )
return result + ( ( A.end() - f ) - ( b - A.rbegin() ) );
else if( *b <= 0 )
return result + ( ( A.rend() - b ) - ( f - A.begin() ) );
else if( *f == -*b )
++result, ++f, ++b;
else if( *f > -*b )
++result, ++b;
else
++result, ++f;
}
return result;
}
int main( int, char ** )
{
std::cout << solution( std::vector< int >{ -5, -3, -1, 0, 3, 6} ) << std::endl;
std::cout << solution( std::vector< int >{ -5, -3, -1, 0, 1, 3, 6} ) << std::endl;
std::cout << solution( std::vector< int >{ -5, -3, -1, 0, 2, 3, 6} ) << std::endl;
std::cout << solution( std::vector< int >{ -5, -3, -1, 3, 6} ) << std::endl;
std::cout << solution( std::vector< int >{ -5, -3, -1, 0, 3, 4, 5} ) << std::endl;
return 0;
}