Absolute Error = Original Number - Rounded/Truncated Number
Relative Error = Absolute Error / Original Number
The process of carrying out arithmetic shifts on a two's complement negative binary number moves one space left/right and a binary 0 is added in the respective spot (either MSB or LSB).
The process of carrying out arithmetic shifts on a two's complement negative binary number moves one space left/right and a binary 0 is added for left shift or 1 for right shift to the MSB.
A logical shift operation shifts each binary digit left or right and the vacated bits are replaced with 0's.
Arithmetic shifts depend on which way they are going. If it is a left arithmetic shift, then the sign will change. But for the right, it will keep its current sign.
Overflow is when a shift has occurred and the binary number does not 'fit' in the register as it is too great.
Underflow is when a shift has occurred and the binary number does not 'fit' in the register as it is too small.