Generating Randomness
Elara is a platform for simulation, optimization and machine learning. Randomness is very useful in each of these fields. Modelling of real-world situations will often involve a degree uncertainty or randomness. East supports the generation of random numbers and other values.
Reproducible randomness
East expressions are always evaluated within an environment. Typically this is a task on the Elara platform. Each task is assigned a fresh random number generator. When the same task is re-executed the random number generated will be seeded with the same value. (Distinct tasks may be seeded differently, however).
When randomness is employed in scenario simulation (in ProcessBuilder), the simulation result may vary from trajectory to trajectory.
Each trajectory will be initialized with a distinct random seed.
Random expressions
East has a series of built-in expressions for generated random values.
| East function | Description | Example usage | Result |
|---|---|---|---|
RandomUniform | Produce a random float between zero and one | RandomUniform() | 0.1290526065430484 |
RandomUniform | Produce a random float between a min and max value | RandomUniform(0, 10) | 6.484048884596946 |
RandomNormal | Produce a random float from a normal distribution (mean 0, standard deviation 1) | RandomNormal() | -0.4858409631668993 |
RandomRange | Produce a random integer between a min and max value | RandomRange(0n, 10n) | 8 |
RandomValue | Produce a random value from an array | RandomValue(["a", "b", "c"]) | "c" |
RandomKey | Produce a random key from a set | RandomKey(new Set(["a", "b", "c"])) | "b" |
RandomWeightedKey | Produce a random key from a dictionary with probability proportional to the value | RandomWeightedKey(new Map([["a", 0.99], ["b", 0.01]])) | "a" |
Furthermore, there are some more advanced distributions provided by the standard library.
| East function | Description | Example usage | Result |
|---|---|---|---|
RandomExponential | Produce a random float from the exponential distribution (mean 1) | RandomUniform() | 0.31412055716414494 |
RandomWeibull | Produce a random float from the Weibull distribution with shape parameter 3 | RandomWeibull(3) | 6.484048884596946 |
RandomKeys | Produce a given number of random keys from a set | RandomKeys(new Set(["a", "b", "c"]), 2n) | {"b", "c"} |
RandomValues | Produce a given number of random values from an array | RandomValues([1, 2, 3], 2n) | [1, 3] |
Next steps
Continue to the next tutorial to understand how to evaluate machine learning functions using East and Elara.