The matrix library only truncates the printout of matrix entries. The output is, of course, full 32bit or 64bit floating point number.