When an input signal contains frequency components above half the analog input sampling rate, aliasing errors occur. These higher frequencies appear as erroneous signals that cannot be distinguished from valid sampled data. The best approach to eliminating this false data is to use a hardware low-pass filter, which limits the input signal bandwidth to below half the sampling rate. If even lower frequency filters are sometimes needed for certain applications, they may be achieved mathematically in software using the valid data.