To understand the theory, consider a simplification. We have a loan queue and a deposit queue. The pit boss wants to run down both queues and match S to L as best it van, making up the difference with created or destroyed money, the bit error. In fact, this would work if all parties operated optimally in tyher sandbox. Less than optimum matchers use less cycles and thus traders have more looks at the stack.
The optimum is to make two moving window Huffman encoders for each queue, but add in pit boss loan and deposit events to make both queues binary balanced twos complement trees. Then you have a node by node match, the interest charges should be the path length times the aggregate ratio.
When the bit error exceeds bounds, the pit boss lengthens or shortens the Huffman window as need be. All participants can traverse the tree, in turn or drop deposit and loan asynchronously.
How is sub optimum matching and red/gree connected?
The secure element uses the red/green indicator to warn when the pricing surface has a convcavity, and he should price around it .
This means, warn if you but two dozen eggs and forget the bacon. If that happens, the coherence the two supply chains had altered, and you jumped, bought two dozen eggs thinking they were cheap compared to bacon. Should have just bought the same, and watched one more time, then you price around the bend. Then find out later you only can eat half of much bacon as before.
But it is highly quantized, choices become locally limited so a sub optimum price in the pit still avoids jumping the shark. Then you add in the buyers club effect where buyers coordinate to reduce supply chain volatility.
A lot of options for robotic pit bosses. Keep it simple as it can then be validated.
No comments:
Post a Comment