In a directly related problem, I am able to compute the wavelet likelihood given the model_prediction and data in `np.array`

format, to later subtract as the residuals.

The function call is

`dwt_chisq(model_prediction, data, [gamma, sigma_r, sigma_w])`

To run this with `xo`

via `PyMC`

it places `model_prediction`

as a tensor (i.e. `theano.tensor.var.TensorVariable`

); also, `gamma`

, `sigma_r`

, and `sigma_w`

are each of type `theano.tensor.var.TensorVariable`

.

In order to run the `dwt_chisq`

as is, I would need to output a numpy array from each of those tensors. What is the appropriate method to do so?

I tried `.eval()`

; but that had very little useful outputs. Please let me know if I'm off base, such as "it's not possible to output a `np.array`

type" or "that's now how you use `.eval()`

"

Thank you!