Quantization

nClearly if we encode a coefficient to less precision we need fewer bits to do so. nA suitable technique for this is to divide the coefficients by an integer and ignore the remainder on encoding and do the reverse on decoding.