Computational methods

So we have a norm, minimum inventory redundancy, an identified Fibonacci polynomials of finite length as  eigenfunctions.  Then we add the dual Huffman code method in which we find the optimum Huffman tree that matches a channel of data variation; away we go.  We expect to see the generated Huffman tree will have spirals now and then, a sure sign of rank change upcoming.

All data is in the form y(x), a polynomial of short variation.  I say short variation here because I model the agent has having bounded functions of short variation.  Variation measured by zero crossings.

Condition the series for seasonal effects, yearly effects and so on. Then blur the data a bit, using a window and expand bins in X and Y.  Then run the result through the run time (windowed) Huffman encoder, and watch the rank and spiral of the generated tree. I think you can just interleave multiple streams, y1(x1), y2(x2),... as long as their variations retain order.