• What does the error message 'number of child results does not match tensor dimensions' mean?

I have the following code

 ns.unity << Meq.Constant(1.0)
 ns.elements << Meq.Sum(ns.unity)
 for p,q in IFRS:    
   ns.a11(p,q) << Meq.Sum(ns.predict_ok(p,q) * ns.predict_ok_conj(p,q))
   ns.a12(p,q) << Meq.Sum(ns.predict_ok_conj(p,q))
   ns.a21(p,q) << Meq.Sum(ns.predict_ok(p,q))
   ns.a22(p,q) << Meq.Composer(dims=(2,2), \
       children=(ns.elements,ns.elements,ns.elements,ns.elements))
   ns.components(p,q)<< Meq.Matrix22(ns.a11(p,q),ns.a12(p,q), \
       ns.a21(p,q),ns.a22(p,q))

But I am getting a failure with the creation of ns.components, all of which fail with errors of the form:

03:40:45: node 'components:12:24': number of child results does not match tensor dimensions (while getting result for request ev.0.0.0.0.1)

Essentially the ns.axx are each just scalar vells of dim (2,2).

What's happening?

I don't understand what you're trying to do in components(p,q)! You're telling it to build a 2x2 matrix out of 4 2x2 matrices, so naturally it complains...