par = [1, 0, 0, 0, 0]

values = np.array([ complex(0, 0),
complex(0, 1), complex(0, -1),
complex(1, 0), complex(-1, 0)])

mat = np.full((n, n), complex(0,0))


for i in range(1, n):
mat[i, i-1] = random.choice(values) if par[0] else np.random.beta(0.1, 0.1)*2 - 1
mat[i-1, i] = random.choice(values) if par[1] else np.random.beta(0.1, 0.1)*2 - 1


mat[0, 0], mat[n-1, n-1] = 1, 1