I put the description up on one of the pages to the left. It is the standard sandbox S/L system, I just added the member in and out to and from the master BTC block chain. Open to free use by anyone who wants to make a bitcoin savings and loan.
Let us define a redundancy in our S/L in terms of the Walmart checkout manager. The checkout manger sets items per basket as needed to keep the various checkout lines stable. If two different people with five items in their basket are in two different lines than that is a redundancy. It is redundant because all baskets with five items should be interchangeable within the same line, their redundancy has been extinguished by a commutative property.
In a bank S/L the equivalent is two $300 deposits earn a different interest payment. This is exactly the problem a Huffman encoder solves. So we have two queues, deposits and loans. If both are minimally redundant then their interest swaps should create a unique, and minimal, fit to the final S/L balance, and some linear scaling between the two results in minimal market risk.
So, if I have a 'compressed' version of a sequence then I have a distributive property going down the encoding tree. I can balance the tree (eliminate skew)by adding entries as needed at each node. Two such trees will be co-balanced, and the ratio of commutative entries on each node will be the same ratio as the opposite tree, I can scale one relative to the other. My scaling is such as to maximally cancel out the balancers I have put, and the resulting differences in the balancers will be minimized. That is minimizing market risk for the pit boss. Th end result should be unique, and most of these proofs are about proving uniqueness on a graph configuration.
No comments:
Post a Comment