Boolean of a lists of arrays during modeling

Questions
1

This is a follow up question from How to slice an array of numbers during fitting

In that post I created fake array data and was able to execute code. However I am now using real data which have different lengths so I have to use a list.

This list of data with rows of different lengths. All elements in each row must be greater than a different row number.

a=[[22.2344534252345,23.2345345345,34.5687697],[23.56786789,2342.5678,234.56856,3456354.678]]
b=[[22.34544654],[222.567867]]

with pm.Model() as model_b:
#Priors
tau = pm.Normal(‘tau’, mu = 2, sd=2)
ϵ = pm.HalfCauchy(‘ϵ’, 5)

#Observed
active_ = a > tau * b

active should return [[False, True, True],[False,True,True,True]]

It returns a single value of False

# Likelihood
μ = pm.Deterministic('μ', pm.math.sum(a*active_))
kill_pred = pm.Normal('kill_pred', mu=μ, sd=ϵ, observed=kill)

trace_b = pm.sample()

pm.traceplot(trace_b);

Is there a way to get around this? Thank you!

2

took care of this by concat into single vector and referencing with an index