Each column of A corresponds to a potential shift for one employee. For example, find(A(:,1)) shows that column 1 has entries in rows 7, 8, 9, 10, 11, 12, and 25. This means that Employee #1 (because there is an entry in row 25, the first row of the bottom half of A) might work hours 7, 8, 9, 10, 11, and 12. Likewise, column 1000 (entries in rows 7, 8, 9, 40) tells us that Employee #16 might work hours 7, 8, and 9.
The blog post describes a decision vector x, with 1295 elements corresponding to the 1295 columns of A. If an employee is chosen to work a shift, the corresponding element of x will be 1. For example, when I run the example, x(24) = 1. This means that Employee #1 will work hours 13-20 (This is a different solution than in the blog post, but has the same total cost).
A shift with dead time will never be selected because there is no column of A that corresponds to a shift with dead time.
