The feature about is_cached

[happyych]

Hello,

I have a question about using is_cached option.
When I read about the option in tutorial, I think it is super cool, because I wanted to reference two or more points in time for a variable simultaneously.

When I tried the option, I faced an error messege as below.
I guess setting is_cached option as False doesn't work in vectorized model.

Does the is_cached option also works in vectorized model?

If I have to set the option in other way, pleas let me know.

Thank you always!

[fumitoh]

Have you updated modelx to v0.27.0?

[happyych]

Yes I tried the code above based on modelx v0.27.0

[fumitoh]

Thank you. Judging from the traceback message, the error seems to be raised when trying to print the error message for another error.
So RENEWAL([1, 2]) is failing already. I think passing the tuple (1, 2) instead of the list shows the traceback of the original error.
I'll fix the error during outputting the error message.

[happyych]

Thank you for the fast reply. I tried tuple as your recomendation. Below is the error messege.

[fumitoh]

Yes, this is a formula error, so the ValueError message is indicating there's an error around a shape mismatch of numpy arrays.
I think the error message should show what line in RENEWAL the error occured although not captured in the screenshot.

[happyych]

Oh, sorry. I missed the line showing what line the error occur.
At line 6, the error occur.

(t <= 1 + DURATIONIF_TO_CVRG()) | ((t - DURATIONIF_TO_CVRG() -1) % (POL_TERM_U() * 12) != 0)

And the value of DURATIONIF_TO_CVRG() and POL_TERM_U() are below.

[fumitoh]

Yes, so as the errormessage says, you can't compare a tuple with 2 elements and an array with 5 elements.
I tested outside modelx:

import numpy as np
t = (1, 2)
DURATIONIF_TO_CVRG = np.array([0.,0.,0.,0.,0.])
POL_TERM_U = np.array([1.,1.,1.,1.,1.])
(t <= 1 + DURATIONIF_TO_CVRG | ((t - DURATIONIF_TO_CVRG -1) % (POL_TERM_U * 12) != 0))

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[52], line 5
      3 DURATIONIF_TO_CVRG = np.array([0.,0.,0.,0.,0.])
      4 POL_TERM_U = np.array([1.,1.,1.,1.,1.])
----> 5 (t <= 1 + DURATIONIF_TO_CVRG | ((t - DURATIONIF_TO_CVRG -1) % (POL_TERM_U * 12) != 0))

ValueError: operands could not be broadcast together with shapes (2,) (5,) 

[happyych]

Sorry for the late reply. It takes some time to figure out the error of my code. Now I understand the situation and feature of is_cached feature. Thank you for your kind explanation!