Arellano_solutions migration to matplotlib 1.4.3 in python 3.4

[finmod]

The following error appears in arellano_solutions in the plotting the last three time series.

---

TypeError Traceback (most recent call last)
C:\Users\Denis\Anaconda3\lib\site-packages\quantecon\markov\core.py in simulate(self, ts_length, init, num_reps, random_state)
351 try:
--> 352 k = len(init) # init is an array
353 dim = 2

TypeError: object of type 'numpy.int64' has no len()

During handling of the above exception, another exception occurred:

ValueError Traceback (most recent call last)
in ()
1 T = 250
----> 2 y_vec, B_vec, q_vec, default_vec = ae.simulate(T)
3
4 # Pick up default start and end dates
5 start_end_pairs = []

C:\Users\Denis\Anaconda3\lib\site-packages\quantecon\models\arellano_vfi.py in simulate(self, T, y_init, B_init)
164 mc = qe.markov.MarkovChain(self.Py)
165
--> 166 y_sim_indices = mc.simulate(T, init=y_init)
167 B_sim_indices = np.empty(T, dtype=np.int64)
168 B_sim_indices[0] = B_init

C:\Users\Denis\Anaconda3\lib\site-packages\quantecon\markov\core.py in simulate(self, ts_length, init, num_reps, random_state)
367 else:
368 raise ValueError(
--> 369 'init must be int, array_like of ints, or None'
370 )
371

ValueError: init must be int, array_like of ints, or None

[oyamad]

@finmod Thank you for the report. This seems to be a bug in MarkovChain.

This line should have been elif isinstance(init, (numbers.Integral, np.integer)): as in here, since "numpy.int64 is not instance of int" (numpy/numpy#2951) (I didn't know this).

To double check, I have to set up a Python 3 environment, and I will then fix this tomorrow.

[oyamad]

Maybe elif isinstance(init, numbers.Integral): is enough (numpy/numpy#4547).

[finmod]

None of the two patches.

TypeError Traceback (most recent call last)

C:\Users\Denis\Anaconda3\lib\site-packages\quantecon\markov\core.py in simulate(self, ts_length, init, num_reps, random_state)

351         try:

--> 352 k = len(init) # init is an array

353             dim = 2

TypeError: object of type 'numpy.int64' has no len()

During handling of the above exception, another exception occurred:

NameError Traceback (most recent call last)

in ()

  1 T = 250

----> 2 y_vec, B_vec, q_vec, default_vec = ae.simulate(T)

  3 

  4 # Pick up default start and end dates

  5 start_end_pairs = []

C:\Users\Denis\Anaconda3\lib\site-packages\quantecon\models\arellano_vfi.py in simulate(self, T, y_init, B_init)

164         mc = qe.markov.MarkovChain(self.Py)

165 

--> 166 y_sim_indices = mc.simulate(T, init=y_init)

167         B_sim_indices = np.empty(T, dtype=np.int64)

168         B_sim_indices[0] = B_init

C:\Users\Denis\Anaconda3\lib\site-packages\quantecon\markov\core.py in simulate(self, ts_length, init, num_reps, random_state)

363             if init is None:

364                 init_states = random_state.randint(self.n, size=k)

--> 365 elif isinstance(init, numbers.Integral):

366                 init_states = np.ones(k, dtype=int) * init

367             else:

NameError: name 'numbers' is not defined

From: Daisuke Oyama [mailto:[email protected]]
Sent: Thursday, September 17, 2015 2:35 PM
To: QuantEcon/QuantEcon.py [email protected]
Cc: finmod [email protected]
Subject: Re: [QuantEcon.py] Arellano_solutions migration to matplotlib 1.4.3 in python 3.4 (#190)

Maybe elif isinstance(init, numbers.Integral): is enough (numpy/numpy#4547 numpy/numpy#4547 ).

—
Reply to this email directly or view it on GitHub #190 (comment) . https://github.com/notifications/beacon/AMHyIojNBexz_kRQYFUGeei_ZvCQh7_bks5oyqsOgaJpZM4F_BcE.gif

[oyamad]

Thanks @finmod, the numbers module has to be imported by import numbers.

[finmod]

Ok. Spot on by inserting import numbers in core.py and also in the arellano notebook. Thanks

Denis Richard
*

---

• Via Panama, 88
• Rome 00198, Italy
• Tel: +39 06 83 90 66 29
• Cel: +39 32 79 76 12 96
• Email: [email protected]
  *

  ---

From: Daisuke Oyama [mailto:[email protected]]
Sent: Thursday, September 17, 2015 4:39 PM
To: QuantEcon/QuantEcon.py [email protected]
Cc: finmod [email protected]
Subject: Re: [QuantEcon.py] Arellano_solutions migration to matplotlib 1.4.3 in python 3.4 (#190)

Thanks @finmod https://github.com/finmod , the numbers module has to be imported by import numbers.

—
Reply to this email directly or view it on GitHub #190 (comment) . https://github.com/notifications/beacon/AMHyIhY6zmHJooQo21MxlvWZSt8zCrlWks5oysgegaJpZM4F_BcE.gif