T_HOMATE

1. Introduction

What?

The THOMATE [1] method is a method achieving the same results as the backtracing [2] but with much less statistics and a control on the errors involved. The goal is to extract from data a distribution of unknown parameters, not only a mean value as with a typical χ².  THOMATE has been applied successfully to analyze the experiment of my Ph.D. disseration [3].

The anecdote is that I developed this method while I was working on the nuclear physics experiment for my PhD thesis, in Dubna (Russia) without an internet connection. I tried to re-develop the backtracing from scratch to try to understand why it was so limited by statistics. I ended up with something completely new, simpler and more efficient (a proof that a bad memory can actually be beneficial).

What for?

THOMATE can have important applications in many fields beyond nuclear physics or even general physics. For example, let's imagine that a study has been made in Tenessee recording what people buy at the supermarket, as a function of their household income. A supermarket manager in Texas could extract thanks to THOMATE the distribution of incomes of his customers, based on the detail of his store's sales. He could then target better his products and services: if, for example, it appears that 20% of his customer are earning $100,000 and above, and studies show that people with this kind of salary buy a fair amount of electronic devices, it is definitely a good idea to dedicate a section of his store to that kind of products.

How?

Here is a little article explaning in more details (and a few equations) what THOMATE really is. The basic idea is actually extremely simple and I wonder why nobody thought of it before (except for the much more complicated and less efficient backtracing).

Table of contents

1. Introduction