Getting mcsolve function from qutip to work

Question

I am using the qutip package under anaconda. I simply call the function mcsolve in my script. Now, when running under a Jupyter notebook, the script runs fine in seconds with results. But when running within Spyder, it gets stuck and never finishes. I Never had this problem with other functions within qutip.

Running this in Spyder prints 'aa' and then 'bb' but never gets to 'cc' line:

import qutip as qt
import numpy as np
import matplotlib.pyplot as plt

# Some constants
g = 100
delta = 10
Ka = g/2.5
Gamma = g/200
Gammap = g/50
Nu = 2*g
Delta_a = -4.12*g
Delta_c = Delta_a +delta*g

N = 4

sm = qt.tensor(qt.qeye(N),qt.sigmam())
sp = sm.dag()
a = qt.tensor(qt.destroy(N),qt.qeye(2))
adag = a.dag()

H = Delta_aspsm + Delta_cadaga + g*(adagsm+spa)+Nu*(sp+sm)

sz = qt.tensor(qt.qeye(N),qt.sigmaz())
C1 = np.sqrt(2*Ka)*a C2 = np.sqrt(Gamma)*sm
C3 = np.sqrt(Gammap/4.0)*(spsm-smsp)

rhoss = qt.steadystate(H,[C1,C2,C3])
print('aa')

times = np.linspace(0,10,1000)
IC = qt.tensor(qt.basis(N,0),qt.basis(2,1))
print('bb')

data = qt.mcsolve(H,IC,times,[C1,C2,C3],[adaga,spsm],ntraj = 1)
print('cc') 

Answer 1

I modified your code so I can use for python 3, I am using the Eclipse IDE but the problem is not IDE (Spider/Jupyter) as I have the same problem.

ERROR Image

It is some multiprocessing problem, maybe your Jupyter notebook using different python interpreter. Here is the modified version of your code:

import qutip as qt
import numpy as np
import matplotlib.pyplot as plt

# Some constants
g = 100
delta = 10
Ka = g/2.5
Gamma = g/200
Gammap = g/50
Nu = 2*g
Delta_a = -4.12*g
Delta_c = Delta_a +delta*g

N = 4

sm = qt.tensor(qt.qeye(N),qt.sigmam())
sp = sm.dag()
a = qt.tensor(qt.destroy(N),qt.qeye(2))
adag = a.dag()

H = Delta_a*sp*sm + Delta_c*adag*a + g*(adag*sm+sp*a)+Nu*(sp+sm)

sz = qt.tensor(qt.qeye(N),qt.sigmaz())
C1 = np.sqrt(2*Ka)*a 
C2 = np.sqrt(Gamma)*sm
C3 = np.sqrt(Gammap/4.0)*(sp*sm-sm*sp)

rhoss = qt.steadystate(H,c_op_list=[C1,C2,C3])
print('aa')

times = np.linspace(0,10,1000)
IC = qt.tensor(qt.basis(N,0),qt.basis(2,1))
print('bb')

data = qt.mcsolve(H,IC,times,[C1,C2,C3],[adag*a,sp*sm],ntraj = 10)
print('cc') 

It seems there is something with the parallel processing, look here: https://github.com/pytorch/pytorch/issues/5858 I got the same error when I was running it. It does not have any safe way of dealing with it unless you ask the conda's developers. However, by monkey patching I managed to run it this is my console image:

Execution without error

Caution the solution is not stable and may cause some inconsistency Go to

C:\Users\Your_username\AppData\Local\Continuum\anaconda3\Lib\multiprocess\spawn.py

Now add this at the very end of the script:

def _check_not_importing_main():
     pass