Monday, November 13, 2017

Mapping the auto pricer to infinite rank

Our auto pricer uses operations on generators, a generator will generate the typical price sequence. Importantly, our system supplies an error term to make the generator compact dense?  .Compact means the generator covers the whole range of typical prices, within the error term. It should generate no duplicates and be closed.  But I am not the mathematicians here.

Now we compare two compact generators, subtract them even, node by node. We make a bold assumption here, mean equals variance down each path of the generator. We can treat the generators as a stable queued system.  The generator is algebraic, one can sum variances down the path.



By ratio, we can exchange value generating the match within. bounded error term, The finial sep is finite hyperbolic, the error term becoming a variance, which is supposed to be one unit.


Let the rank of your generator rise indefinitely and the system approach continuous hyperbolic.  The matched sequence defines a series of onion layers, each i Th element being the next layer.  The error term maintains the boundaries, the quantizations. This is a class of sphere packing, that is our assumption.  Price is indexible because we sphere pack.  We spread out and look for the center.

Let us interpret

What do we make of the longest path through the generator? It accumulates the highest queue size and variance?  That is really spare capacity from elements paths not that long, it accumulates until the long queue is full, then we get the rare price.  Like an unpacking function, whenever the case is unloaded, and unpacked, room is recovered in the delivery truck. In this interpretation, I am counting unused capacity in the distribution. I should be free ti do this, count negative flow of empty contained instead of positive flow of filled container.

No comments: