NumPy's random integer methods can be used for sampling from a set of evenly-spaced real numbers, though it requires a bit of extra work: to pick a number from n evenly-spaced real numbers between a and b (inclusive), use:

In [x]: a + (b - a) * (np.random.random_integers(n) - 1) / (n - 1)

For example to sample from $[\frac{1}{2}, \frac{3}{2}, \frac{5}{2}, \frac{7}{2}]$,

In [x]: a, b, n = 0.5, 3.5, 4
In [x]: a + (b - a) * (np.random.random_integers(n, size=10) - 1) / (n - 1)
array([ 1.5,  0.5,  1.5,  1.5,  3.5,  2.5,  3.5,  3.5,  3.5,  3.5])

Note: the numpy.random.random_integers() function is now deprecated and may be removed in future versions of Python. The current recommended call is to np.random.randint(1, n+1) instead:

    In [x]: a + (b - a) * (np.random.randint(1, n+1) - 1) / (n - 1)