def compute_mat_X(nbscans, lhrf, tr, tmax, onsets, durations=None):
x = numpy.zeros((nbscans,lhrf),dtype=float)
x_tmp = numpy.zeros(nbscans,dtype=float)
x_tmp[xrange(0,tmax)] = restarize_events(onsets,numpy.zeros_like(onsets),tr,tmax)
for i in xrange(0,lhrf):
for j in xrange(0,len(x)):
x[j,i] = x_tmp[j-i]
return x
def compute_mat_X_2(nbscans, tr, lhrf, dt, onsets, durations=None):
if durations is None: #assume spiked stimuli
durations = numpy.zeros_like(onsets)
osf = tr/dt # over-sampling factor
assert( int(osf) == osf) #construction will only work if dt is a multiple of tr
x = numpy.zeros((nbscans,lhrf),dtype=float)
tmax = nbscans * tr #total session duration
lgt = (nbscans + 2) * osf # nb of scans if tr=dt
paradigm_bins = restarize_events(onsets, numpy.zeros_like(onsets), dt, tmax)
firstcol = numpy.concatenate( (paradigm_bins, numpy.zeros(lgt-len(paradigm_bins))) )
firstrow = numpy.concatenate( ([paradigm_bins[0]], numpy.zeros(lhrf-1, dtype=int)) )
x_tmp = numpy.array(toeplitz( firstcol, firstrow), dtype=int)
os_indexes = [(numpy.arange(nbscans)*osf).astype(int)]
x = x_tmp[os_indexes]
return x
