Dear Liu,
as per documentation, the second argument of random_graph takes a function,
passing np.poisson(4) is like passing one single value form that
distribution.
Try changing your code to:
g = gt.random_graph(1000, deg_sampler=lambda:np.poisson(4), directed=False)
and see if it works as intended.
Best,
Giuseppe
2014-04-17 15:30 GMT+02:00 Liu <boxiliu.hust(a)gmail.com>: