Monday, July 27, 2015

You don't scare me Peter Keevash

Abstract: We give a new proof of the Frankl-Rodl theorem on forbidden intersections, via the probabilistic method of dependent random choice. Our method extends to codes with forbidden distances, where over large alphabets our bound is significantly better than that obtained by Frankl and Rodl. We also apply our bound to a question of Ellis on sets of permutations with forbidden distances, and to establish a weak form of a conjecture of Alon, Shpilka and Umans on sun flowers.
This is Peter's work, along with Long, co author.  We learn about sets of school girls who, each in set pairs, exclude one unique set. Peter gets a better bound on the number of sets in this family of sets.  This is going to be fun.  Its like asking how much net do I need to catch roller coaster passengers being flung.

Schoolgirls, they are everywhere these days. Now, as we read this and have fun, Banker Coin goes up in value (Peter owns 1/12 of the coin). Peter makes us much more comfortable about banker bot.

Peter's Plot:
The bot is weaving a web of safety nets along our journey's path. So, our smart card light goes yellow when we make a lousy bet on the number of school girls.  Thus our SmartCards are more fun, and trustable. We are going to have fun.

How would I do this?
A queuing problem on a finite graph, without loops.  I know that local node properties have to obey this l-avoiding problem. So I have a restriction on two colliding Poisson, at stability. I apply the information metric to my Skellam distribution, after hiring a professional mathematician. My information metric gives me a relationship between information efficiency and variance on that Skellam. My solution minimizes the second derivative on that compression ratio between set sizes on my schoolgirls. No I know there is a relationship between equipartition and kinetic energy, so I got self adapted statistics, and I have to find a bounded path for variance on a finite graph.

So, our bot just helps the owner avoid long waits in line. Literally, we will find this to be true. We get that red light, long wait, warning. And, that bot computes with the largest, most accurate and secure collection of information available.  Peter's students are going to be gazillionares.

The banker interface:
The bank programmer sees his client as a large set of spreadsheets, just like he does now. Except the banker can write a protocol graph which instructs the smart card to perform spread rebalances in real time as purchases are made.  Bankers will literally be in the business of writing central banker rules, specific to their client groups.


No comments: