Complex-valued wave function about neural-network-quantum-state HOT 1 OPEN

tf.log(tf.float32) --> return tf.float32
And therefore, if x < 0 and with type tf.float32,
tf.log(x) would output nan

See example below

(Pdb) tf.log(-1)
*** TypeError: Value passed to parameter 'x' has DataType int32 not in list of allowed values: float16, bfloat16, float32, float64, complex64, complex128
(Pdb) tf.log(-1+1j)
<tf.Tensor 'Log_2:0' shape=() dtype=complex128>
(Pdb) tf.log(-1.)
<tf.Tensor 'Log_3:0' shape=() dtype=float32>
(Pdb) tf.log(-1.+0)
<tf.Tensor 'Log_4:0' shape=() dtype=float32>
(Pdb) tf.log(-1.+0j)
<tf.Tensor 'Log_5:0' shape=() dtype=complex128>
(Pdb) tf.log(-2.)
<tf.Tensor 'Log_6:0' shape=() dtype=float32>
(Pdb) tf.log(tf.cast(-1.+0j), tf.complex64)
*** TypeError: cast() missing 1 required positional argument: 'dtype'
(Pdb) tf.log(tf.cast(-1.+0j), dtype=tf.complex64)
*** TypeError: cast() missing 1 required positional argument: 'dtype'
(Pdb) tf.log(tf.cast(-1.+0j, dtype=tf.complex64))
<tf.Tensor 'Log_7:0' shape=() dtype=complex64>
(Pdb) sess=tf.InteractiveSession()
(Pdb) sess.run(tf.log(-1))
*** TypeError: Value passed to parameter 'x' has DataType int32 not in list of allowed values: float16, bfloat16, float32, float64, complex64, complex128
(Pdb) sess.run(tf.log(-1.))
nan

from neural-network-quantum-state.