Thanks for your explain, so it seems different from matlab and numpy.
在 2014年1月15日星期三UTC+8下午7时06分50秒，Billou Bielour写道： > > Julia has N-vectors in addition to matrices, for performance reasons, as > far as I know. N-vectors have only one dimension while matrices have two > (even Nx1 and 1xN ones). > > In most linear algebra operation N-vectors behave like Nx1 matrices, this > means you can multiply a NxN matrix with N-vector. In addition when you > extract a column from a matrix ( A[:,1] ) it gives you a N-vector and not a > Nx1 matrix, while getting a row gives you a 1xN matrix. > > Transposition automatically convert your N-vector to a matrix, this leads > to the weird inequality a != a'' because a and a'' are not of the same type. > > All of this seems a bit weird, and I don't completely understand why it > works like that, but there's some reasons, either performances or design, > for why it's like that. It's not a big deal in practice though, once you > understand how it behave. > > >
