使用 Numpy 生成范围内的随机数

Question

I have ranges , where each tuple represents the range of a random number. eg:

ranges = [(-1,100), (0,1), (50, 5000)]

Now, I want to create a numpy array where each element of array is randomly generated from the corresponding tuple of ranges .

Pseudo code:

rand_array = numpy.array[randomly generate element1 from (-1,100), randomly generate element2 from (0,1), randomly generate element3 from (50,500)]

Of course, naive way to do this is:

rand_array = [random.uniform(coord[0], coord[1]) for coord in ranges]

But I want to do it in numpy way, as my whole code accepts numpy array data-types. Another solution can be to convert naively generated rand_array to numpy array. But I think it is not efficient. Is there a numpy way of doing this?

Answer 1

Most random functions accept array parameters. So you can do

np.random.uniform(*np.transpose(ranges))

The usual broadcasting rules apply, for example, to get 10 triplets

np.random.uniform(*np.transpose(ranges),(10,3))