Consider a matrix with m < 10000 rows and n < 1000 columns containing ones and zeros. I need to select k <= n columns so that the number of rows containing more ones than zeros is maximal (k is fixed).