NewBestSub/it.uniud.newbestsub.dataset/DatasetController

DatasetController

class DatasetController(targetToAchieve: String)

DatasetController

=================

Orchestrates:

DatasetModel lifecycle (load → optional expand → seal → solve)
Streaming consumption (printer coroutine reading ProgressEvents)
View writing for final snapshots and merged artifacts (CSV + Parquet)

Determinism

When Parameters.deterministic is true:

Runs execute sequentially (BEST → WORST → AVERAGE).
All randomness is routed through jMetal's singleton via RandomBridge.withSeed.
Expansion helpers (expandTopics, expandSystems) also use a SplittableRandom seeded via RandomBridge.childSeed for reproducible fake data.

Efficiency

A single printer coroutine consumes a back-pressured Channel of events.
Uses batched TopKReplaceBatch to avoid rewriting the TOP file multiple times per generation.

Constructors

DatasetController

constructor(targetToAchieve: String)

Properties

aggregatedDataResultPaths

var aggregatedDataResultPaths: MutableList<String>

functionValuesResultPaths

var functionValuesResultPaths: MutableList<String>

infoResultPaths

var infoResultPaths: MutableList<String>

models

var models: MutableList<DatasetModel>

Per-target models (when TARGET_ALL, indices: 0=BEST, 1=WORST, 2=AVERAGE; else a single entry).

topSolutionsResultPaths

var topSolutionsResultPaths: MutableList<String>

variableValuesResultPaths

var variableValuesResultPaths: MutableList<String>

Functions

clean

fun clean(dataList: MutableList<String>, logMessage: String)

Delete a list of files on disk and remove their paths from the given list.

copy

fun copy()

Copy the per-execution and merged results produced by the last solve/merge into the experiments destination tree (CSV + Parquet), preserving filenames.

expandSystems

fun expandSystems(expansionCoefficient: Int, trueNumberOfSystems: Int)

Expand the dataset by appending expansionCoefficient fake systems (or revert to a prefix), across all loaded models.

expandTopics

fun expandTopics(expansionCoefficient: Int)

Expand the dataset by appending expansionCoefficient fake topics to all loaded models.

load

fun load(datasetPath: String)

Load a dataset from a CSV path into one or three DatasetModels depending on targetToAchieve.

merge

fun merge(numberOfExecutions: Int)

Merge results from multiple executions (ex1..exN) into merged CSV/Parquet artifacts:

solve

fun solve(parameters: Parameters)

Run the experiment(s) according to parameters.targetToAchieve: