Scipy minimize using tensorflow matmul got the input error

Question

I am using the tensorflow2 and using scipy minimize, my code as following

x_init=float(1/500)
xopti=tf.fill([p,1],x_init)
d