How do I create a scalar value that does not depend on the independent variables ?

[bradbell]

I would try using Jax , but the install does not work on my system and Jax has 1.3k open issues, so I think that my install questions would get lost in the noise; see
jax-ml/jax#18084

How do I create a scalar value that does not depend on the independent variables ?
The following function works (fails) if you set flag to False (True).

flag = False
import autograd

def my_fun(x) :  
   Scalar = type( x[0] )
   result = x[0] * x[1]
   if flag :
      result = result + Scalar(0.0) 
   return result

my_grad = autograd.grad(my_fun)
my_x    = autograd.numpy.array( [ 2.0, 3.0 ] )
my_g    = my_grad( my_x )
ok      = my_g[0] == my_x[1] and my_g[1] == my_x[0]
print( f'ok = {ok}' )