Rejection sampling on uniform distribution?

Try this:

numPoints = 1000  % Total number of points to save.
numLow = 1;
numHigh = 1;
% Start collecting elements using acceptance if in range,
% and rejection if out of range.
for k = 1 : 10000000  % enough to make sure we will collect at least numPoints elements.
  r = rand();
  % Save the number if it's in the low range.
  if r <= 0.2 && numLow <= 0.20 * numPoints
    % In range.  Store/save the number.
    lowKeepers(numLow) = r;
    numLow = numLow + 1;
  else
    % Not in range so reject/ignore it.
  end
  % Save the number if it's in the high range.
  if r >= 0.8 && numHigh <= 0.80 * numPoints
    % In range.  Store/save the number.
    highKeepers(numHigh) = r;
    numHigh = numHigh + 1;
  else
    % Not in range so reject/ignore it.
  end
end