math-optim/algo/algo.go

// Copyright 2023 wanderer <a_mirre at utb dot cz>
// SPDX-License-Identifier: GPL-3.0-or-later

package algo

import (
	"fmt"
	"log"
	"sort"
	"sync"

	"git.dotya.ml/wanderer/math-optim/algo/cec2020"
	"git.dotya.ml/wanderer/math-optim/algo/de"
	"git.dotya.ml/wanderer/math-optim/bench"
	c20 "git.dotya.ml/wanderer/math-optim/bench/cec2020"
	"git.dotya.ml/wanderer/math-optim/report"
	"git.dotya.ml/wanderer/math-optim/stats"
)

// var Algos = []string{"Random Search", "Stochastic Hill Climbing"}

// mu protects access to meanStats.
var mu sync.Mutex

// mCoMPL protexts access to comparisonOfMeansPicList.
var mCoMPL sync.Mutex

var meanStats = &stats.MeanStats{}

var comparisonOfMeansPicList = &report.PicList{Algo: "Comparison of Means"}

// getComparisonOfMeansPics returns a sorted slice of pics field from the
// package global 'algoMeanPics'.
func getComparisonOfMeansPics() []report.Pic {
	// note: sorting by filename (dimens value being 0-padded at generation
	// time), relying on this as a hack so that we didn't have to implement our
	// own natural-order sorter.
	sort.Sort(comparisonOfMeansPicList.Pics)

	return comparisonOfMeansPicList.Pics
}

// saveAlgoMeans saves algo bench means safely.
func saveAlgoMeans(sabm stats.AlgoBenchMean) {
	mu.Lock()
	meanStats.AlgoMeans = append(meanStats.AlgoMeans, sabm)
	mu.Unlock()
}

// GetMeanStats returns a pointer of type stats.MeanStats to a sorted package
// global 'meanStats'.
func GetMeanStats() *stats.MeanStats {
	sort.Sort(meanStats)

	return meanStats
}

// PrepComparisonOfMeans returns a pointer to a slice of pics (of type
// report.PicList) and an integer - the count of unique benchmarking functions
// used.
func PrepComparisonOfMeans(wg *sync.WaitGroup) (*report.PicList, int) {
	pL := report.NewPicList()
	meanStats := GetMeanStats()
	algos := make([]string, 0)

	// learn how many algos were processed based on the data.
	for _, v := range meanStats.AlgoMeans {
		// if algos is empty just add the value directly, else determine if
		// it's already been added or not.
		if len(algos) > 0 {
			alreadyadded := false

			for _, algoName := range algos {
				if algoName == v.Algo {
					// early bail if already added.
					alreadyadded = true
					break
				}
			}

			if !alreadyadded {
				algos = append(algos, v.Algo)
			}
		} else {
			algos = append(algos, v.Algo)
		}
	}

	// construct title consisting of names of all involved algorithms.
	for _, v := range algos {
		switch pL.Algo {
		case "":
			pL.Algo = v
		default:
			pL.Algo += " vs " + v
		}
	}

	log.Println(`generating "Comparison of Means" plots`)

	algoCount := len(algos)
	dimLen := len(bench.Dimensions)
	// without Rastrigin in active duty for the moment.
	benchCount := len(bench.Functions) - 1

	// note: this is a wee bit ugly.
	for d := 0; d < dimLen; d++ {
		for i := 0; i < benchCount; i++ {
			dimXAlgoMeanVals := make([]stats.AlgoMeanVals, 0, algoCount)

			for j := 0; j < algoCount*benchCount; j += benchCount {
				neighbInfo := ""

				// only add info about neighbours if it was changed (from
				// the default of -1).
				if n := meanStats.AlgoMeans[d+j].BenchMeans[i].Neighbours; n != -1 {
					neighbInfo = fmt.Sprintf(" (N: %d)", n)
				}

				ms := &stats.AlgoMeanVals{
					Title:    meanStats.AlgoMeans[d+j].Algo + neighbInfo,
					MeanVals: meanStats.AlgoMeans[d+j].BenchMeans[i].MeanVals,
				}
				dimXAlgoMeanVals = append(dimXAlgoMeanVals, *ms)
			}

			dimens := meanStats.AlgoMeans[d].BenchMeans[i].Dimens
			iterations := meanStats.AlgoMeans[d].BenchMeans[i].Iterations
			bench := meanStats.AlgoMeans[d].BenchMeans[i].Bench

			wg.Add(1)

			// construct plots concurrently.
			go PlotMeanValsMulti(
				wg, dimens, iterations, bench, "plot-", ".pdf",
				dimXAlgoMeanVals...,
			)
		}
	}

	// wait for all the plotting to conclude.
	wg.Wait()

	pL.Pics = getComparisonOfMeansPics()

	return pL, benchCount
}

// DoRandomSearch executes a search using the 'Random search' method.
func DoRandomSearch(wg *sync.WaitGroup, m *sync.Mutex) {
	defer wg.Done()

	printRandomSearch("starting...")

	// funcCount is the number of bench functions available.
	// without Rastrigin in active duty for the moment.
	funcCount := len(bench.Functions) - 1
	// stats for the current algo (RandomSearch).
	algoStats := make([][]stats.Stats, funcCount)
	// ch serves as a way to get the actual computed output.
	ch := make(chan []stats.Stats, funcCount)

	defer close(ch)

	for i := range algoStats {
		// ng y'all.
		go RandomSearchNG(10000, 30, bench.Dimensions, bench.FuncNames[i], ch)
	}

	// get results.
	for i := range algoStats {
		s := <-ch

		algoStats[i] = s
	}

	// save stats to json.
	// stats.SaveStats(schw, "schwefel")
	// stats.SaveStats(djg1, "djg1")
	// stats.SaveStats(djg2, "djg2")

	pCh := make(chan report.PicList, funcCount*len(bench.Dimensions))
	pMeanCh := make(chan report.PicList, funcCount*len(bench.Dimensions))

	defer close(pCh)
	defer close(pMeanCh)

	for i := range algoStats {
		go plotAllDims(algoStats[i], "plot", ".pdf", pCh, pMeanCh)
	}

	pLs := []report.PicList{}
	pLsMean := []report.PicList{}

	for range algoStats {
		pL := <-pCh
		pLMean := <-pMeanCh

		pLs = append(pLs, pL)
		pLsMean = append(pLsMean, pLMean)
	}

	algoName := "Random Search"

	// protect access to shared data.
	m.Lock()
	report.SavePicsToFile(pLs, pLsMean, algoName)

	stats.SaveTable(algoName, algoStats)
	m.Unlock()
}

// TODO(me): split this package to multiple - package per algo, common code here.
// TODO(me): implement Simulated Annaeling.
// TODO(me): implement a variant of Stochastic Hill Climber that tweaks its
//				Neighbourhood size or MaxNeighbourVariancePercent based on the
//				latest 5 values, if they don't change, params get tweaked to
//				broaden the search space to make sure it's not stuck in a local
//				extreme.

// DoStochasticHillClimbing performs a search using the 'Stochastic Hill
// Climbing' method.
func DoStochasticHillClimbing(wg *sync.WaitGroup, m *sync.Mutex) {
	defer wg.Done()

	printSHC("starting...")

	// funcCount is the number of bench functions available.
	// without Rastrigin in active duty for the moment.
	funcCount := len(bench.Functions) - 1
	// stats for the current algo (StochasticHillClimber).
	algoStats := make([][]stats.Stats, funcCount)
	// ch serves as a way to get the actual computed output.
	ch := make(chan []stats.Stats, funcCount)

	defer close(ch)

	for i := range algoStats {
		// params:
		//   maxFES, benchMinIters, neighbours int,
		//   theD []int,
		//   benchFunc string,
		//   ch chan []stats.Stats
		go HillClimb(10000, 30, 10, bench.Dimensions, bench.FuncNames[i], ch)
	}

	// get results.
	for i := range algoStats {
		s := <-ch

		algoStats[i] = s
	}

	pCh := make(chan report.PicList, funcCount*len(bench.Dimensions))
	pMeanCh := make(chan report.PicList, funcCount*len(bench.Dimensions))

	defer close(pCh)
	defer close(pMeanCh)

	for _, algoStat := range algoStats {
		go plotAllDims(algoStat, "plot", ".pdf", pCh, pMeanCh)
	}

	pLs := []report.PicList{}
	pLsMean := []report.PicList{}

	for range algoStats {
		pL := <-pCh
		pLMean := <-pMeanCh

		pLs = append(pLs, pL)
		pLsMean = append(pLsMean, pLMean)
	}

	algoName := "Stochastic Hill Climbing"

	// protect access to shared data.
	m.Lock()
	report.SavePicsToFile(pLs, pLsMean, algoName)

	stats.SaveTable(algoName, algoStats)
	m.Unlock()
}

func DoStochasticHillClimbing100Neigh(wg *sync.WaitGroup, m *sync.Mutex) {
	defer wg.Done()

	printSHC("starting...")

	// funcCount is the number of bench functions available.
	// without Rastrigin in active duty for the moment.
	funcCount := len(bench.Functions) - 1
	// stats for the current algo (StochasticHillClimber).
	algoStats := make([][]stats.Stats, funcCount)
	// ch serves as a way to get the actual computed output.
	ch := make(chan []stats.Stats, funcCount)

	for i := range algoStats {
		go HillClimb(10000, 30, 100, bench.Dimensions, bench.FuncNames[i], ch)
	}

	// get results.
	for i := range algoStats {
		s := <-ch

		algoStats[i] = s
	}

	pCh := make(chan report.PicList, funcCount*len(bench.Dimensions))
	pMeanCh := make(chan report.PicList, funcCount*len(bench.Dimensions))

	for _, algoStat := range algoStats {
		go plotAllDims(algoStat, "plot", ".pdf", pCh, pMeanCh)
	}

	pLs := []report.PicList{}
	pLsMean := []report.PicList{}

	for range algoStats {
		pL := <-pCh
		pLMean := <-pMeanCh

		pLs = append(pLs, pL)
		pLsMean = append(pLsMean, pLMean)
	}

	algoName := "Stochastic Hill Climbing 100 Neighbours"

	// protect access to shared data.
	m.Lock()
	report.SavePicsToFile(pLs, pLsMean, algoName)
	// report.SavePicsToFile(pLsMean, pLs, algoName)

	// stats.PrintStatisticTable(algoStats)
	stats.SaveTable(algoName, algoStats)
	m.Unlock()
}

func DojDE(wg *sync.WaitGroup, m *sync.Mutex) {
	defer wg.Done()

	de.LogPrintln("starting")

	// funcCount is the number of bench functions available and tested.
	funcCount := len(bench.Functions)
	// stats for the current algo.
	algoStats := make([][]stats.Stats, funcCount)
	// ch serves as a way to get the actual computed output.
	ch := make(chan []stats.Stats, funcCount)
	// chAlgoMeans := make(chan *stats.AlgoBenchMean, 1)
	chAlgoMeans := make(chan *stats.AlgoBenchMean, funcCount)

	defer close(ch)
	defer close(chAlgoMeans)

	// jDE params.
	np := 50
	f := 0.5
	cr := 0.9

	for i := range algoStats {
		jDE := de.NewjDE()

		// params:
		//  Generations, minimum bench iterations, mutation strategy, parameter
		//  self-adaptation scheme, initial population size, differential
		//  weight, mutation constant, dimensions, bench name and a
		//  synchronisation channel.
		//
		//  -1 to disable generation limits,
		//  n > 0 for minimum bench iterations
		//  0..17 to choose a mutation strategy,
		//  0..1 to select a parameter self-adaptation scheme,
		//  np >= 4 as initial population size.
		jDE.Init(-1, 30, 0, 0, np, f, cr, bench.DimensionsGA, bench.FuncNames[i], ch, chAlgoMeans)

		go jDE.Run()
	}

	// get results.
	for i := range algoStats {
		s := <-ch
		aM := <-chAlgoMeans

		algoStats[i] = s

		saveAlgoMeans(*aM)
	}

	pCh := make(chan report.PicList, funcCount*len(bench.DimensionsGA))
	pMeanCh := make(chan report.PicList, funcCount*len(bench.DimensionsGA))

	for _, algoStat := range algoStats {
		go plotAllDims(algoStat, "plot", ".pdf", pCh, pMeanCh)
	}

	pLs := []report.PicList{}
	pLsMean := []report.PicList{}

	for range algoStats {
		pL := <-pCh
		pLMean := <-pMeanCh

		pLs = append(pLs, pL)
		pLsMean = append(pLsMean, pLMean)
	}

	algoName := "Differential Evolution"

	// protect access to shared data.
	m.Lock()
	report.SavePicsToFile(pLs, pLsMean, algoName)

	stats.SaveTable(algoName, algoStats)
	m.Unlock()
}

func DoCEC2020jDE(wg *sync.WaitGroup, m *sync.Mutex) {
	defer wg.Done()

	cec2020.LogPrintln("starting")

	// funcCount is the number of bench functions available and tested.
	funcCount := len(c20.Functions)
	// stats for the current algo.
	algoStats := make([][]stats.Stats, funcCount)
	// ch serves as a way to get the actual computed output.
	ch := make(chan []stats.Stats, funcCount)
	// chAlgoMeans := make(chan *stats.AlgoBenchMean, 1)
	chAlgoMeans := make(chan *stats.AlgoBenchMean, funcCount)

	defer close(ch)
	defer close(chAlgoMeans)

	// jDE params.
	np := 50
	f := 0.5
	cr := 0.9

	for i := range algoStats {
		jDE := cec2020.NewjDE()

		// params:
		//  Generations, minimum bench iterations, mutation strategy, parameter
		//  self-adaptation scheme, initial population size, differential
		//  weight, mutation constant, dimensions, bench name and a
		//  synchronisation channel.
		//
		//  -1 to disable generation limits,
		//  n > 0 for minimum bench iterations
		//  0..17 to choose a mutation strategy,
		//  0..1 to select a parameter self-adaptation scheme,
		//  np >= 4 as initial population size.
		jDE.Init(-1, 30, 0, 0, np, f, cr, c20.Dimensions, c20.FuncNames[i], ch, chAlgoMeans)

		go jDE.Run()
	}

	// get results.
	for i := range algoStats {
		s := <-ch
		aM := <-chAlgoMeans

		algoStats[i] = s

		saveAlgoMeans(*aM)
	}

	pCh := make(chan report.PicList, funcCount*len(c20.Dimensions))
	pMeanCh := make(chan report.PicList, funcCount*len(c20.Dimensions))

	for _, algoStat := range algoStats {
		go plotAllDims(algoStat, "plot", ".pdf", pCh, pMeanCh)
	}

	pLs := []report.PicList{}
	pLsMean := []report.PicList{}

	for range algoStats {
		pL := <-pCh
		pLMean := <-pMeanCh

		pLs = append(pLs, pL)
		pLsMean = append(pLsMean, pLMean)
	}

	algoName := "Self-adapting Differential Evolution"

	// protect access to shared data.
	m.Lock()
	report.SavePicsToFile(pLs, pLsMean, algoName)

	stats.SaveTable(algoName, algoStats)
	m.Unlock()
}