What is meant by ones-compliment of a decimal number?

Answer 1

One-complement applies to binary values, not decimal values. Therefore when we say the ones-complement of a decimal value we mean convert the value to binary, invert all the bits (the ones-complement), then convert the result back to decimal.

For example, the decimal value 42 has the following representation in 8-bit binary:

00101010

If we invert all the bits we get 11010101 which is 213 decimal. Thus 213 is the ones-complement of 42, and vice versa.

However, it's not quite as straightforward as that because some (older) systems use ones-complement notation to represent signed values, such that 11010101 represents the decimal value -42. The problem with this notation is that the ones-complement of 00000000 is 11111111 which means the decimal value 0 has two representations, +0 and -0 respectively. In the real-world, zero is neither positive nor negative.

To resolve this problem, modern systems use twos-complement to represent signed values. The twos-complement of any value is simply the ones-complement plus one. Thus the ones-complement of 42 becomes -43, therefore the twos-complement of 42 is -43+1 which is -42. Thus -42 is represented by the binary value 11010110 in twos-complement notation.

With twos-complement, there is only one representation for the value 0. This is because the ones-complement of 00000000 is 11111111 and if we add 00000001 we get 00000000. Note that we don't get 100000000 because the result cannot have any more bits than were in the original value. When an "overflow" occurs, we cycle back to zero. As a result, incrementing and decrementing signed values has exactly the same logic as incrementing or decrementing unsigned values and flipping the sign of any value is only slightly more complicated by the extra addition operation. However, flipping the sign of a value is a much rarer operation than counting so the cost is trivial compared to the cost of counting operations using ones-complement (because there are two values for zero).

Note that ones-complement notation allows an 8-bit value to store signed values in the range -127 to +127, whereas twos-complement allows a range of -128 to +127 (through the elimination of the extra zero). But in unsigned notation, both allow the same range: 0 to 255.

Although we rarely encounter ones-complement notation, it is important to keep in mind that not all systems use twos-complement notation, particularly when working with low-level but portable programming languages. This is the reason why both the C and the C++ standards state that the range of an 8-bit signed value is only guaranteed to store values in the range -127 to +127.