utility-function-extractor/packages/utility-tools
2022-06-19 21:22:07 -04:00
..
input feat: Add utility tools 2022-06-15 23:04:46 -04:00
output feat: Add utility tools 2022-06-15 23:04:46 -04:00
src fix: aggregation bug 2022-06-19 21:22:07 -04:00
LICENSE.txt tweak: General code cleanup. 2022-06-19 20:50:53 -04:00
package.json v1.0.3 2022-06-19 21:19:39 -04:00
README.md tweak: General code cleanup. 2022-06-19 20:50:53 -04:00
yarn.lock fix: interface bug 2022-06-19 17:49:15 -04:00

Utility Tools

This package contains a series of utilities to work with the utility functions produced by this utility function extractor.

Built with

Getting started

Installation

yarn add utility-tools

then in your file:

import { mergeSort, findPaths, aggregatePaths } from "utility-tools";

Usage

You can find an example how to use and concatenate these functions in /src/example.js, as well as an example of the input format needed in the input folder.

Interface

Merge sort (mergeSort)

Given a list of elements and a list of utilitity comparisons, sort the list. If there are not enough comparisons to implement the merge sort algorithm, return one of the missing comparisons.

Gotcha: The list of elements has to be the same list, and in the same order, as that produced when initially doing the comparisons. This is because the merge-sort algorithm depends on the initial order of the list.

Find Paths (findPaths)

Given an (ordered) list of elements and a list of utility comparisons, find all possible monotonous paths from each element to each other. A monotonous path is a path that is either all increasing or all decreasing, relative to the ordering given.

Note: Some elements will have many more paths than others.

Note: The findPaths.js file has a few un-used functions which should make it easier to understand the code.

Aggregate paths (aggregatePaths)

Given a list of path, aggregate them to finally produce an estimate of the relative utility of each element.

There are two ways of doing this:

    1. Aggregate the means (expected values) for each path.
    • This method is fast
    • But has the disadvantage the expected value aggregation is tricky, particularly if one of the elements is positive and the other one negative (because then one can't)
    1. Aggregate the distributions given for each path.

Roadmap

I don't have any additions planned for this repository.

Contact

Feel free to shoot me any questions at nuno.semperelh@protonmail.com