Generator-Utils

Overview

A bunch of functions so you don't have to write them. The main purpose of the repo is to unify the most used tools and make them deterministic to the random seed.

Installation

git clone https://github.com/establishment/generator-utils
cd generator-utils
make install

Design goals

• All functions which take a range as parameter generate values in [left, right). Adding a x in front of these functions makes the generated values [left, right]. For example, Range(2) returns a vector with [0, 1] while xRange(2) returns [0, 1, 2].
• Nothing throws exceptions, just assert(0) for now

Short documentation overview

Utils

vector<Type> Range(left, right, step)

• Returns a vector with [left, left + step, .. right) Other versions Range(left, right), Range(right) and all of these with x

Sort RandomShuffle Reverse Unique

• Applies basic stl function directly on containers(vectors)

Actual signature Container<Type>& Sort(Container<Type>&) Container<Type>&& Sort(Container<Type>&&) - for function call return values

vector<int> x = {1, 3, 2, 2, 3, 1, 1, 5};
Reverse(Unique(x)); // Unique sorts as well, because you want that
// x will be {5, 3, 2, 1}
auto v = Sort(FunctionThatReturnsAVector());

Random

UniqueNumbers(left, right, num_elements)

• Returns a vector of num_elements numbers in range [left, right) Other versions UniqueNumbers(right, num_elements) - all version work with x as well

Honorable mentions UniqueNumbers(left, right, num_elements, const std::function<bool(Type)>& valid_number) - returns random numbers, but only numbers for which the function returns true

Partition::Uniform(num_elements, num_buckets, min_val)

• Returns a vector of num_buckets elements. The sum of the elements is num_elements and each of them is >= min_val. Assuming that min_val = 0, the returned partition is a random, each possible valid partition having the same chance to be chosen.

Rand RandLog RandLogScaled

• Returns a random number in [left, right). The weight of the number depends on the function
• All Rand functions have a x version, for which the retured value will be in [left, right].

Rand(left, right)

• Basic rand in interval [left, right) Other versions Rand() Rand(right)

RandLog(left, right)

• Give a random value in [left, right), value x having the weight equal to 1/x

RandLogScaled(left, right)

• same as RandLog but element i has weight equal to 1 / (i - left + 1)