Hey Ruopeng,

I just wanted to confirm that you do in fact normalize the gradient matrix that is given for dti_recon.  For example, my gradient table looks something like this:

40, 40, 0

0, 40, 0 etc…

Where the 40's represent 40 G/cm gradient strength.  I am manually passing my maximum b-vale with the -b flag.  I am wondering if you normalize the matrix (i.e. is this the same as using a gradient matrix made of 1's and zeros?) 

thanks for your help,