Erroneous sqrt in demo notebook
andrewwarrington opened this issue · 1 comments
andrewwarrington commented
Hi all,
In the lgssm_learning.ipynb
demo here, I think there shouldn't be the sqrt
around the variance bounds in plot_predictions
:
plt.fill_between(
jnp.arange(num_timesteps),
spc * i + smoothed_emissions[:, i] - 2 * jnp.sqrt(smoothed_emissions_std[I]),
spc * i + smoothed_emissions[:, i] + 2 * jnp.sqrt(smoothed_emissions_std[I]),
color=ln.get_color(),
alpha=0.25,
)
given that the variable name is ..._std
and the return type in the model looks to be a standard deviation:
dynamax/dynamax/linear_gaussian_ssm/models.py
Line 262 in d88050c
I think the error is also persistent in the HMC version here.
A