The MAlign technique involves spreading the 36 mirror segments out by a known amount, taking one or more images of the resulting pattern, and analyzing that pattern for the proper movement of the individual segment images. This is very similar to a Shack-Hartmann test, in that variations of the pattern from the ideal pattern yield a measure of the errors in various terms of the telescope configuration. In particular, it is used to measure errors in piston (i.e. focus), secondary tilt (coma), and stack. It can also measure other aberrations, such as astigmatism and higher order terms, but other than changing the stack we have no way of using that information.
Most often called "malacq," this is sometimes called "malign" (e.g. NIRC). In most cases (again, NIRC is an exception) the appropriate tool is run by the OAs to take the sequence of images. In broad outline the procedure must:
What the OAs generally call "malign," but is called "align" from the command line. This tool reads in each focus mode image and displays it. The OA can then mark the segments, usually in an automatic mode where (s)he only need mark the center of the hexagonal segment pattern. The software then analyzes a set of image measurements, reporting qualitative blur values for each of piston, tilt, and stack errors. The OA can send any or all of the suggested corrections.