func Test_mtdeb01(tst *testing.T) {
//verbose()
chk.PrintTitle("mtdeb01. Deb's mutation")
var ops OpsData
ops.SetDefault()
ops.Pm = 1.0
ops.Xrange = [][]float64{{-3, 3}, {-4, 4}}
ops.EnfRange = true
rnd.Init(0)
A := []float64{-1, 1}
io.Pforan("before: A = %v\n", A)
FltMutationDeb(A, 10, &ops)
io.Pforan("after: A = %v\n", A)
ha0 := rnd.Histogram{Stations: utl.LinSpace(-3, 3, 11)}
nsamples := 1000
aa := make([]float64, len(A))
a0s := make([]float64, nsamples)
for _, t := range []int{0, 50, 100} {
for i := 0; i < nsamples; i++ {
copy(aa, A)
FltMutationDeb(aa, t, &ops)
a0s[i] = aa[0]
}
ha0.Count(a0s, true)
io.Pf("\ntime = %d\n", t)
io.Pf("%s", rnd.TextHist(ha0.GenLabels("%.1f"), ha0.Counts, 60))
}
}
func Test_ind01(tst *testing.T) {
//verbose()
chk.PrintTitle("ind01. representation and copying")
rnd.Init(0)
nbases := 3
A := get_individual(0, nbases)
B := A.GetCopy()
chk.Scalar(tst, "ova0", 1e-17, B.Ovas[0], 123)
chk.Scalar(tst, "ova1", 1e-17, B.Ovas[1], 345)
chk.Scalar(tst, "oor0", 1e-17, B.Oors[0], 10)
chk.Scalar(tst, "oor1", 1e-17, B.Oors[1], 20)
chk.Scalar(tst, "oor2", 1e-17, B.Oors[2], 30)
fmts := map[string][]string{"int": {" %d"}, "flt": {" %.1f"}, "str": {" %q"}, "key": {" %x"}, "byt": {" %q"}, "fun": {" %q"}}
oA := A.Output(fmts, false)
oB := B.Output(fmts, false)
io.Pfyel("\n%v\n", oA)
io.Pfyel("%v\n\n", oB)
chk.String(tst, oA, " 1 20 300 4.4 5.5 666.0 \"abc\" \"b\" \"c\" 53 47 41 \"ABC\" \"DEF\" \"GHI\" \"f0\" \"f1\" \"f2\"")
chk.String(tst, oB, " 1 20 300 4.4 5.5 666.0 \"abc\" \"b\" \"c\" 53 47 41 \"ABC\" \"DEF\" \"GHI\" \"f0\" \"f1\" \"f2\"")
A.SetFloat(1, 33)
A.SetFloat(2, 88)
oA = A.Output(fmts, false)
io.Pfyel("\n%v\n", oA)
chk.String(tst, oA, " 1 20 300 4.4 33.0 88.0 \"abc\" \"b\" \"c\" 53 47 41 \"ABC\" \"DEF\" \"GHI\" \"f0\" \"f1\" \"f2\"")
}
func Test_binmut01(tst *testing.T) {
//verbose()
chk.PrintTitle("binmut01. mutation: binary")
var ops OpsData
ops.SetDefault()
ops.Pm = 1.0
ops.Tmax = 10
ops.Nchanges = 3
rnd.Init(0)
A := []int{0, 1, 1, 1, 0, 0, 1, 0, 1, 1}
a := make([]int, len(A))
copy(a, A)
io.Pforan("before: A = %v\n", A)
IntBinMutation(A, 0, &ops)
io.Pforan("after: A = %v\n", A)
ndiff := 0
for i := 0; i < len(A); i++ {
if A[i] != a[i] {
ndiff++
}
}
io.Pforan("number of changes = %v\n", ndiff)
if ndiff != ops.Nchanges {
tst.Errorf("binary mutation failed\n")
}
}
func Test_intordmut01(tst *testing.T) {
//verbose()
chk.PrintTitle("intordmut01")
var ops OpsData
ops.SetDefault()
ops.Pm = 1
rnd.Init(0)
a := []int{1, 2, 3, 4, 5, 6, 7, 8}
io.Pforan("before: a = %v\n", a)
ops.OrdSti = []int{2, 5, 4}
IntOrdMutation(a, 0, &ops)
io.Pfcyan("after: a = %v\n", a)
chk.Ints(tst, "a", a, []int{1, 2, 6, 7, 3, 4, 5, 8})
nums := utl.IntRange2(1, 9)
sort.Ints(a)
chk.Ints(tst, "asorted = 12345678", a, nums)
a = []int{1, 2, 3, 4, 5, 6, 7, 8}
io.Pforan("\nbefore: a = %v\n", a)
ops.OrdSti = nil
IntOrdMutation(a, 0, &ops)
io.Pfcyan("after: a = %v\n", a)
sort.Ints(a)
chk.Ints(tst, "asorted = 12345678", a, nums)
}
func Test_pop02(tst *testing.T) {
//verbose()
chk.PrintTitle("pop02")
rnd.Init(0)
genes := [][]float64{
{1, 5}, // 0
{1, 3}, // 1
{5, 7}, // 2
{1, 2}, // 3
{2, 4}, // 4
{3, 6}, // 5
{4, 8}, // 6
{4, 6}, // 7
{1, 3}, // 8
{0, 0}, // 9
}
// objective values, oor and demerits
// 0 1 2 3 4 5 6 7 8 9
ovs := []float64{11, 21, 10, 12, 13, 31, 41, 11.1, 31.5, 11.5}
dem := []float64{0.2, 0.7, 0.1, 0.5, 0.6, 0.8, 1.0, 0.3, 0.9, 0.4}
// init population
ninds := len(genes)
nova := 1
noor := 0
nbases := 2
var pop Population
pop = make([]*Individual, ninds)
for i := 0; i < ninds; i++ {
pop[i] = NewIndividual(nova, noor, nbases, genes[i])
pop[i].Ovas[0] = ovs[i]
pop[i].Demerit = dem[i]
}
pop.Sort()
genes_sorted := [][]float64{
{5, 7}, // 2
{1, 5}, // 0
{4, 6}, // 7
{0, 0}, // 9
{1, 2}, // 3
{2, 4}, // 4
{1, 3}, // 1
{3, 6}, // 5
{1, 3}, // 8
{4, 8}, // 6
}
for i, ind := range pop {
for j := 0; j < ind.Nfltgenes; j++ {
chk.Scalar(tst, io.Sf("i%dg%d", i, j), 1e-14, ind.GetFloat(j), genes_sorted[i][j])
}
}
}
func main() {
// GA parameters
C := goga.ReadConfParams("tsp-simple.json")
rnd.Init(C.Seed)
// location / coordinates of stations
locations := [][]float64{
{60, 200}, {180, 200}, {80, 180}, {140, 180}, {20, 160}, {100, 160}, {200, 160},
{140, 140}, {40, 120}, {100, 120}, {180, 100}, {60, 80}, {120, 80}, {180, 60},
{20, 40}, {100, 40}, {200, 40}, {20, 20}, {60, 20}, {160, 20},
}
nstations := len(locations)
C.SetIntOrd(nstations)
C.CalcDerived()
// objective value function
C.OvaOor = func(ind *goga.Individual, idIsland, time int, report *bytes.Buffer) {
L := locations
ids := ind.Ints
dist := 0.0
for i := 1; i < nstations; i++ {
a, b := ids[i-1], ids[i]
dist += math.Sqrt(math.Pow(L[b][0]-L[a][0], 2.0) + math.Pow(L[b][1]-L[a][1], 2.0))
}
a, b := ids[nstations-1], ids[0]
dist += math.Sqrt(math.Pow(L[b][0]-L[a][0], 2.0) + math.Pow(L[b][1]-L[a][1], 2.0))
ind.Ovas[0] = dist
return
}
// evolver
nova, noor := 1, 0
evo := goga.NewEvolver(nova, noor, C)
evo.Run()
// results
io.Pfgreen("best = %v\n", evo.Best.Ints)
io.Pfgreen("best OVA = %v (871.117353844847)\n\n", evo.Best.Ovas[0])
// plot travelling salesman path
if C.DoPlot {
plt.SetForEps(1, 300)
X, Y := make([]float64, nstations), make([]float64, nstations)
for k, id := range evo.Best.Ints {
X[k], Y[k] = locations[id][0], locations[id][1]
plt.PlotOne(X[k], Y[k], "'r.', ms=5, clip_on=0, zorder=20")
plt.Text(X[k], Y[k], io.Sf("%d", id), "fontsize=7, clip_on=0, zorder=30")
}
plt.Plot(X, Y, "'b-', clip_on=0, zorder=10")
plt.Plot([]float64{X[0], X[nstations-1]}, []float64{Y[0], Y[nstations-1]}, "'b-', clip_on=0, zorder=10")
plt.Equal()
plt.AxisRange(10, 210, 10, 210)
plt.Gll("$x$", "$y$", "")
plt.SaveD("/tmp/goga", "test_evo04.eps")
}
}
func Test_ind02(tst *testing.T) {
//verbose()
chk.PrintTitle("ind02. copy into")
rnd.Init(0)
nbases := 1
A := get_individual(0, nbases)
B := get_individual(1, nbases)
fmts := map[string][]string{
"int": {"%2d", "%4d", "%5d"}, // ints
"flt": {"%6g", "%6g", "%5g"}, // floats
"str": {"%4s", "%2s", "%2s"}, // strings
"key": {"%3x", "%3x", "%3x"}, // keys
"byt": {"%4s", "%4s", "%4s"}, // bytes
"fun": {"%3s", "%3s", "%3s"}, // funcs
}
io.Pfpink("A = %v\n", A.Output(fmts, false))
io.Pfcyan("B = %v\n", B.Output(fmts, false))
var ops OpsData
ops.SetDefault()
ops.Pc = 1.0
ops.Cuts = []int{1, 2}
ops.Xrange = [][]float64{{0, 1}, {-20, 20}, {-300, 300}}
a := A.GetCopy()
b := A.GetCopy()
IndCrossover(a, b, A, B, 0, &ops)
io.Pforan("a = %v\n", a.Output(fmts, false))
io.Pfblue2("b = %v\n", b.Output(fmts, false))
chk.Ints(tst, "a.Ints ", a.Ints, []int{1, -20, 300})
chk.Ints(tst, "b.Ints ", b.Ints, []int{-1, 20, -300})
chk.Strings(tst, "a.Strings", a.Strings, []string{"abc", "Y", "c"})
chk.Strings(tst, "b.Strings", b.Strings, []string{"X", "b", "Z"})
// TODO: add other tests here
io.Pf("\n")
x := get_individual(0, nbases)
x.Ovas = []float64{0, 0}
x.Oors = []float64{0, 0, 0}
io.Pfblue2("x = %v\n", x.Output(fmts, false))
B.CopyInto(x)
chk.Scalar(tst, "ova0", 1e-17, x.Ovas[0], 200)
chk.Scalar(tst, "ova1", 1e-17, x.Ovas[1], 100)
chk.Scalar(tst, "oor0", 1e-17, x.Oors[0], 15)
chk.Scalar(tst, "oor1", 1e-17, x.Oors[1], 25)
chk.Scalar(tst, "oor2", 1e-17, x.Oors[2], 35)
io.Pforan("x = %v\n", x.Output(fmts, false))
chk.String(tst, x.Output(fmts, false), B.Output(fmts, false))
}
func Test_pop03(tst *testing.T) {
//verbose()
chk.PrintTitle("pop03")
rnd.Init(0)
C := NewConfParams()
C.NumInts = 7
pop := PopBinGen(0, C)
io.Pforan("%v\n", pop.Output(C))
}
func Test_cxdeb01(tst *testing.T) {
//verbose()
chk.PrintTitle("cxdeb01. Deb's crossover")
var ops OpsData
ops.SetDefault()
ops.Pc = 1.0
ops.Xrange = [][]float64{{-3, 3}, {-4, 4}}
ops.EnfRange = true
rnd.Init(0)
A := []float64{-1, 1}
B := []float64{1, 2}
a := make([]float64, len(A))
b := make([]float64, len(A))
FltCrossoverDeb(a, b, A, B, 0, &ops)
io.Pforan("A = %v\n", A)
io.Pforan("B = %v\n", B)
io.Pfcyan("a = %.6f\n", a)
io.Pfcyan("b = %.6f\n", b)
nsamples := 1000
a0s, a1s := make([]float64, nsamples), make([]float64, nsamples)
b0s, b1s := make([]float64, nsamples), make([]float64, nsamples)
for i := 0; i < nsamples; i++ {
FltCrossoverDeb(a, b, B, A, 0, &ops)
a0s[i], a1s[i] = a[0], a[1]
b0s[i], b1s[i] = b[0], b[1]
}
ha0 := rnd.Histogram{Stations: []float64{-4, -3.5, -3, -2.5, -2, -1.5, -1, -0.5, 0, 0.5, 1}}
hb0 := rnd.Histogram{Stations: []float64{0, 0.5, 1, 1.5, 2, 2.5, 3, 3.5, 5, 5.5, 6}}
ha1 := rnd.Histogram{Stations: utl.LinSpace(-4, 4, 11)}
hb1 := rnd.Histogram{Stations: utl.LinSpace(-4, 4, 11)}
ha0.Count(a0s, true)
hb0.Count(b0s, true)
ha1.Count(a1s, true)
hb1.Count(b1s, true)
io.Pforan("\na0s\n")
io.Pf("%s", rnd.TextHist(ha0.GenLabels("%.1f"), ha0.Counts, 60))
io.Pforan("b0s\n")
io.Pf("%s", rnd.TextHist(hb0.GenLabels("%.1f"), hb0.Counts, 60))
io.Pforan("\na1s\n")
io.Pf("%s", rnd.TextHist(ha1.GenLabels("%.1f"), ha1.Counts, 60))
io.Pforan("b1s\n")
io.Pf("%s", rnd.TextHist(hb1.GenLabels("%.1f"), hb1.Counts, 60))
}
func Test_simplechromo01(tst *testing.T) {
//verbose()
chk.PrintTitle("simplechromo01")
rnd.Init(0)
nbases := 2
for i := 0; i < 10; i++ {
chromo := SimpleChromo([]float64{1, 10, 100}, nbases)
io.Pforan("chromo = %v\n", chromo)
chk.IntAssert(len(chromo), 3*nbases)
chk.Scalar(tst, "gene0", 1e-14, chromo[0]+chromo[1], 1)
chk.Scalar(tst, "gene1", 1e-14, chromo[2]+chromo[3], 10)
chk.Scalar(tst, "gene2", 1e-13, chromo[4]+chromo[5], 100)
}
}
func Test_ends02(tst *testing.T) {
//verbose()
chk.PrintTitle("ends02")
rnd.Init(0)
size := 20
ncuts := 10
nsamples := 100
hist := rnd.IntHistogram{Stations: utl.IntRange(size + 3)}
for i := 0; i < nsamples; i++ {
ends := GenerateCxEnds(size, ncuts, nil)
hist.Count(ends, false)
}
io.Pf("%s\n", rnd.TextHist(hist.GenLabels("%d"), hist.Counts, 60))
}
func Test_mwicz01(tst *testing.T) {
//verbose()
chk.PrintTitle("mwicz01. Michalewicz mutation")
var ops OpsData
ops.SetDefault()
ops.Pm = 1.0
ops.Tmax = 10
rnd.Init(0)
ops.Xrange = [][]float64{{0, 2}, {1, 3}, {2, 4}, {3, 5}, {4, 6}}
T := utl.IntRange(int(ops.Tmax))
for _, t := range T {
io.Pf("t=%v Δ=%v\n", t, ops.MwiczDelta(float64(t), 1))
}
for _, t := range T {
A := []float64{0, 1, 2, 3, 4}
FltMutationMwicz(A, t, &ops)
io.Pforan("A = %.8f\n", A)
}
if chk.Verbose {
b := 2.0
f := func(r, tb float64) float64 {
return math.Pow(r, math.Pow(1.0-tb, b))
}
np := 21
r, tb := utl.MeshGrid2D(0, 1, 0, 1, np, np) // tb = t/tmax
z := la.MatAlloc(np, np)
for i := 0; i < np; i++ {
for j := 0; j < np; j++ {
z[i][j] = f(r[i][j], tb[i][j])
}
}
plt.Surface(tb, r, z, "linewidth=0.8")
plt.Gll("tb", "r", "")
plt.SaveD("/tmp/goga", "test_mwicz01.eps")
//plt.Show()
}
}
func Test_flt01(tst *testing.T) {
//verbose()
chk.PrintTitle("flt01. quadratic with inequalities")
// parameters
C := NewConfParams()
C.Pll = false
C.Nisl = 1
C.Ninds = 12
C.GAtype = "crowd"
C.CrowdSize = 3
C.DiffEvol = true
C.RangeFlt = [][]float64{
{-2, 2}, // gene # 0: min and max
{-2, 2}, // gene # 1: min and max
}
C.PopFltGen = PopFltGen
if chk.Verbose {
C.DoPlot = chk.Verbose
}
C.CalcDerived()
rnd.Init(C.Seed)
// functions
fcn := func(f, g, h []float64, x []float64) {
f[0] = x[0]*x[0]/2.0 + x[1]*x[1] - x[0]*x[1] - 2.0*x[0] - 6.0*x[1]
g[0] = 2.0 - x[0] - x[1] // ≥ 0
g[1] = 2.0 + x[0] - 2.0*x[1] // ≥ 0
g[2] = 3.0 - 2.0*x[0] - x[1] // ≥ 0
g[3] = x[0] // ≥ 0
g[4] = x[1] // ≥ 0
}
// simple problem
sim := NewSimpleFltProb(fcn, 1, 5, 0, C)
sim.Run(chk.Verbose)
// plot
sim.Plot("test_flt01")
}
func Test_blx01(tst *testing.T) {
//verbose()
chk.PrintTitle("blx01. blended crossover")
var ops OpsData
ops.SetDefault()
ops.Pc = 1.0
ops.Xrange = [][]float64{{-1, 2}, {0, 3}, {1, 4}, {3, 6}, {4, 7}}
rnd.Init(0)
A := []float64{0, 1, 2, 4, 5}
B := []float64{1, 2, 3, 5, 6}
a := make([]float64, len(A))
b := make([]float64, len(A))
FltCrossoverBlx(a, b, A, B, 0, &ops)
io.Pforan("A = %v\n", A)
io.Pforan("B = %v\n", B)
io.Pfcyan("a = %v\n", a)
io.Pfcyan("b = %v\n", b)
}
func Test_flt03(tst *testing.T) {
//verbose()
chk.PrintTitle("flt03. sin⁶(5 π x) multimodal")
// configuration
C := NewConfParams()
C.Nova = 1
C.Noor = 2
C.Nisl = 4
C.Ninds = 24
C.GAtype = "crowd"
C.DiffEvol = true
C.CrowdSize = 3
C.ParetoPhi = 0.01
C.CompProb = true
C.Tf = 100
C.Dtmig = 60
C.RangeFlt = [][]float64{{0, 0.9999999999999}}
C.PopFltGen = PopFltGen
C.CalcDerived()
rnd.Init(C.Seed)
// post-processing function
values := utl.Deep3alloc(C.Tf/10, C.Nisl, C.Ninds)
C.PostProc = func(idIsland, time int, pop Population) {
if time%10 == 0 {
k := time / 10
for i, ind := range pop {
values[k][idIsland][i] = ind.GetFloat(0)
}
}
}
// functions
yfcn := func(x float64) float64 { return math.Pow(math.Sin(5.0*math.Pi*x), 6.0) }
fcn := func(f, g, h []float64, x []float64) {
f[0] = -yfcn(x[0])
}
// simple problem
sim := NewSimpleFltProb(fcn, 1, 0, 0, C)
sim.Run(chk.Verbose)
// write histograms and plot
if chk.Verbose {
// write histograms
var buf bytes.Buffer
hist := rnd.Histogram{Stations: utl.LinSpace(0, 1, 13)}
for k := 0; k < C.Tf/10; k++ {
for i := 0; i < C.Nisl; i++ {
clear := false
if i == 0 {
clear = true
}
hist.Count(values[k][i], clear)
}
io.Ff(&buf, "\ntime=%d\n%v", k*10, rnd.TextHist(hist.GenLabels("%4.2f"), hist.Counts, 60))
}
io.WriteFileVD("/tmp/goga", "test_flt03_hist.txt", &buf)
// plot
plt.SetForEps(0.8, 300)
xmin := sim.Evo.Islands[0].Pop[0].GetFloat(0)
xmax := xmin
for k := 0; k < C.Nisl; k++ {
for _, ind := range sim.Evo.Islands[k].Pop {
x := ind.GetFloat(0)
y := yfcn(x)
xmin = utl.Min(xmin, x)
xmax = utl.Max(xmax, x)
plt.PlotOne(x, y, "'r.',clip_on=0,zorder=20")
}
}
np := 401
X := utl.LinSpace(0, 1, np)
Y := make([]float64, np)
for i := 0; i < np; i++ {
Y[i] = yfcn(X[i])
}
plt.Plot(X, Y, "'b-',clip_on=0,zorder=10")
plt.Gll("$x$", "$y$", "")
plt.SaveD("/tmp/goga", "test_flt03_func.eps")
}
}
func Test_intordcx01(tst *testing.T) {
//verbose()
chk.PrintTitle("intordcx01")
var ops OpsData
ops.SetDefault()
ops.Pc = 1
rnd.Init(0)
A := []int{1, 2, 3, 4, 5, 6, 7, 8}
B := []int{2, 4, 6, 8, 7, 5, 3, 1}
a := make([]int, len(A))
b := make([]int, len(A))
ops.Cuts = []int{2, 5}
IntOrdCrossover(a, b, A, B, 0, &ops)
io.Pforan("A = %v\n", A)
io.Pfblue2("B = %v\n", B)
io.Pfgreen("a = %v\n", a)
io.Pfyel("b = %v\n", b)
chk.Ints(tst, "A", A, []int{1, 2, 3, 4, 5, 6, 7, 8})
chk.Ints(tst, "B", B, []int{2, 4, 6, 8, 7, 5, 3, 1})
chk.Ints(tst, "a", a, []int{4, 5, 6, 8, 7, 1, 2, 3})
chk.Ints(tst, "b", b, []int{8, 7, 3, 4, 5, 1, 2, 6})
sort.Ints(a)
sort.Ints(b)
nums := utl.IntRange2(1, 9)
chk.Ints(tst, "asorted = 12345678", a, nums)
chk.Ints(tst, "bsorted = 12345678", b, nums)
A = []int{1, 3, 5, 7, 6, 2, 4, 8}
B = []int{5, 6, 3, 8, 2, 1, 4, 7}
ops.Cuts = []int{3, 6}
IntOrdCrossover(a, b, A, B, 0, &ops)
io.Pforan("\nA = %v\n", A)
io.Pfblue2("B = %v\n", B)
io.Pfgreen("a = %v\n", a)
io.Pfyel("b = %v\n", b)
chk.Ints(tst, "A", A, []int{1, 3, 5, 7, 6, 2, 4, 8})
chk.Ints(tst, "B", B, []int{5, 6, 3, 8, 2, 1, 4, 7})
chk.Ints(tst, "a", a, []int{5, 7, 6, 8, 2, 1, 4, 3})
chk.Ints(tst, "b", b, []int{3, 8, 1, 7, 6, 2, 4, 5})
sort.Ints(a)
sort.Ints(b)
chk.Ints(tst, "asorted = 12345678", a, nums)
chk.Ints(tst, "bsorted = 12345678", b, nums)
A = []int{1, 2, 3, 4, 5, 6, 7, 8}
B = []int{2, 4, 6, 8, 7, 5, 3, 1}
ops.Cuts = []int{}
IntOrdCrossover(a, b, A, B, 0, &ops)
io.Pforan("\nA = %v\n", A)
io.Pfblue2("B = %v\n", B)
io.Pfgreen("a = %v\n", a)
io.Pfyel("b = %v\n", b)
sort.Ints(a)
sort.Ints(b)
chk.Ints(tst, "asorted = 12345678", a, nums)
chk.Ints(tst, "bsorted = 12345678", b, nums)
C := []int{1, 2, 3}
D := []int{3, 1, 2}
c := make([]int, len(C))
d := make([]int, len(D))
IntOrdCrossover(c, d, C, D, 0, &ops)
io.Pforan("\nC = %v\n", C)
io.Pfblue2("D = %v\n", D)
io.Pfgreen("c = %v\n", c)
io.Pfyel("d = %v\n", d)
chk.Ints(tst, "c", c, []int{2, 1, 3})
chk.Ints(tst, "d", d, []int{1, 2, 3})
sort.Ints(c)
sort.Ints(d)
chk.Ints(tst, "csorted = 123", c, []int{1, 2, 3})
chk.Ints(tst, "dsorted = 123", d, []int{1, 2, 3})
}
// CalcDerived computes derived variables and checks consistency
func (o *Parameters) CalcDerived() {
// check
if o.Nova < 1 {
chk.Panic("number of objective values (nova) must be greater than 0")
}
if o.Nsol < 6 {
chk.Panic("number of solutions must greater than 6. Nsol = %d is invalid", o.Nsol)
}
if o.Ncpu < 2 {
o.Ncpu = 1
o.Pll = false
o.DtExc = 1
}
if o.Ncpu > o.Nsol/2 {
chk.Panic("number of CPU must be smaller than or equal to half the number of solutions. Ncpu=%d > Nsol/2=%d", o.Ncpu, o.Nsol/2)
}
if o.Tf < 1 {
o.Tf = 1
}
if o.DtExc < 1 {
o.DtExc = o.Tf / 10
}
if o.DtOut < 1 {
o.DtOut = o.Tf / 5
}
// derived
o.Nflt = len(o.FltMin)
o.Nint = len(o.IntMin)
if o.BinInt > 0 {
o.Nint = o.BinInt
}
if o.Nflt == 0 && o.Nint == 0 {
chk.Panic("either floats and ints must be set (via FltMin/Max or IntMin/Max)")
}
// floats
if o.Nflt > 0 {
chk.IntAssert(len(o.FltMax), o.Nflt)
o.DelFlt = make([]float64, o.Nflt)
for i := 0; i < o.Nflt; i++ {
o.DelFlt[i] = o.FltMax[i] - o.FltMin[i]
}
}
// mesh
if o.Nflt < 2 {
o.UseMesh = false
}
if o.UseMesh {
if o.Nbry < 2 {
o.Nbry = 2
}
o.NumXiXjPairs = (o.Nflt*o.Nflt - o.Nflt) / 2
o.NumXiXjBryPts = (o.Nbry-2)*4 + 4
o.NumExtraSols = o.NumXiXjPairs * o.NumXiXjBryPts
io.PfYel("NumXiXjPairs=%d NumXiXjBryPts=%d NumExtraSols=%d\n", o.NumXiXjPairs, o.NumXiXjBryPts, o.NumExtraSols)
o.Nsol += o.NumExtraSols
}
// generic ints
if o.BinInt == 0 && o.Nint > 0 {
chk.IntAssert(len(o.IntMax), o.Nint)
o.DelInt = make([]int, o.Nint)
for i := 0; i < o.Nint; i++ {
o.DelInt[i] = o.IntMax[i] - o.IntMin[i]
}
}
if o.Nint != o.Nflt {
o.ClearFlt = false
}
// number of cuts and changes in ints
if o.Nint > 0 {
if o.IntNcuts > o.Nint {
o.IntNcuts = o.Nint
}
if o.IntNchanges > o.Nint {
o.IntNchanges = o.Nint
}
}
// initialise random numbers generator
rnd.Init(o.Seed)
}
func Test_pop01(tst *testing.T) {
//verbose()
chk.PrintTitle("pop01")
rnd.Init(0)
// genes
genes := [][]float64{
{1, 5, -200},
{1, 3, -300},
{5, 7, -400},
{1, 2, -500},
{2, 4, -300},
}
// objective values and fitness values
ovs := []float64{11, 21, 10, 12, 13}
// init population
ninds := len(genes)
nova := 1
noor := 0
nbases := 2
var pop Population
pop = make([]*Individual, ninds)
for i := 0; i < ninds; i++ {
pop[i] = NewIndividual(nova, noor, nbases, genes[i])
pop[i].Ovas[0] = ovs[i]
}
// check floats and subfloats
for i, ind := range pop {
for j := 0; j < ind.Nfltgenes; j++ {
chk.Scalar(tst, io.Sf("before: i%dg%d", i, j), 1e-12, ind.GetFloat(j), genes[i][j])
}
}
// print bases
io.Pf("\nbases (before)\n")
io.Pf("%s\n", pop.OutFloatBases("%10.4f"))
// change subfloats
bases := [][]float64{
{10, 1, 14, 5, -10, -1},
{10, 1, 12, 1, -20, -1},
{10, 5, 12, 1, -30, -1},
{10, 1, 12, 2, -40, -1},
{10, 2, 11, 1, -20, -1},
}
for i, b := range bases {
copy(pop[i].Floats, b)
}
// print bases
io.Pfyel("bases (after)\n")
io.Pfyel("%s\n", pop.OutFloatBases("%4g"))
// checkf floats
chk.Scalar(tst, "after: i0g0", 1e-16, pop[0].GetFloat(0), 11)
chk.Scalar(tst, "after: i0g1", 1e-16, pop[0].GetFloat(1), 19)
chk.Scalar(tst, "after: i0g2", 1e-16, pop[0].GetFloat(2), -11)
chk.Scalar(tst, "after: i1g0", 1e-16, pop[1].GetFloat(0), 11)
chk.Scalar(tst, "after: i1g1", 1e-16, pop[1].GetFloat(1), 13)
chk.Scalar(tst, "after: i1g2", 1e-16, pop[1].GetFloat(2), -21)
chk.Scalar(tst, "after: i2g0", 1e-16, pop[2].GetFloat(0), 15)
chk.Scalar(tst, "after: i2g1", 1e-16, pop[2].GetFloat(1), 13)
chk.Scalar(tst, "after: i2g2", 1e-16, pop[2].GetFloat(2), -31)
chk.Scalar(tst, "after: i3g0", 1e-16, pop[3].GetFloat(0), 11)
chk.Scalar(tst, "after: i3g1", 1e-16, pop[3].GetFloat(1), 14)
chk.Scalar(tst, "after: i3g2", 1e-16, pop[3].GetFloat(2), -41)
chk.Scalar(tst, "after: i4g0", 1e-16, pop[4].GetFloat(0), 12)
chk.Scalar(tst, "after: i4g1", 1e-16, pop[4].GetFloat(1), 12)
chk.Scalar(tst, "after: i4g2", 1e-16, pop[4].GetFloat(2), -21)
}
func Test_int01(tst *testing.T) {
//verbose()
chk.PrintTitle("int01. organise sequence of ints")
io.Pf("\n")
// initialise random numbers generator
rnd.Init(0) // 0 => use current time as seed
// parameters
C := NewConfParams()
C.Nova = 1
C.Noor = 0
C.Nisl = 1
C.Ninds = 20
C.RegTol = 0
C.NumInts = 20
//C.GAtype = "crowd"
C.CrowdSize = 2
C.Tf = 50
C.Verbose = chk.Verbose
C.CalcDerived()
// mutation function
C.Ops.MtInt = func(A []int, time int, ops *OpsData) {
size := len(A)
if !rnd.FlipCoin(ops.Pm) || size < 1 {
return
}
pos := rnd.IntGetUniqueN(0, size, ops.Nchanges)
for _, i := range pos {
if A[i] == 1 {
A[i] = 0
}
if A[i] == 0 {
A[i] = 1
}
}
}
// generation function
C.PopIntGen = func(id int, cc *ConfParams) Population {
o := make([]*Individual, cc.Ninds)
genes := make([]int, cc.NumInts)
for i := 0; i < cc.Ninds; i++ {
for j := 0; j < cc.NumInts; j++ {
genes[j] = rand.Intn(2)
}
o[i] = NewIndividual(cc.Nova, cc.Noor, cc.Nbases, genes)
}
return o
}
// objective function
C.OvaOor = func(ind *Individual, idIsland, time int, report *bytes.Buffer) {
score := 0.0
count := 0
for _, val := range ind.Ints {
if val == 0 && count%2 == 0 {
score += 1.0
}
if val == 1 && count%2 != 0 {
score += 1.0
}
count++
}
ind.Ovas[0] = 1.0 / (1.0 + score)
return
}
// run optimisation
evo := NewEvolver(C)
evo.Run()
// results
ideal := 1.0 / (1.0 + float64(C.NumInts))
io.PfGreen("\nBest = %v\nBestOV = %v (ideal=%v)\n", evo.Best.Ints, evo.Best.Ovas[0], ideal)
}
func Test_int02(tst *testing.T) {
//verbose()
chk.PrintTitle("int02. TSP")
// location / coordinates of stations
locations := [][]float64{
{60, 200}, {180, 200}, {80, 180}, {140, 180}, {20, 160}, {100, 160}, {200, 160},
{140, 140}, {40, 120}, {100, 120}, {180, 100}, {60, 80}, {120, 80}, {180, 60},
{20, 40}, {100, 40}, {200, 40}, {20, 20}, {60, 20}, {160, 20},
}
nstations := len(locations)
// parameters
C := NewConfParams()
C.Nova = 1
C.Noor = 0
C.Nisl = 4
C.Ninds = 24
C.RegTol = 0.3
C.RegPct = 0.2
//C.Dtmig = 30
C.GAtype = "crowd"
C.ParetoPhi = 0.1
C.Elite = false
C.DoPlot = false //chk.Verbose
//C.Rws = true
C.SetIntOrd(nstations)
C.CalcDerived()
// initialise random numbers generator
rnd.Init(0)
// objective value function
C.OvaOor = func(ind *Individual, idIsland, t int, report *bytes.Buffer) {
L := locations
ids := ind.Ints
//io.Pforan("ids = %v\n", ids)
dist := 0.0
for i := 1; i < nstations; i++ {
a, b := ids[i-1], ids[i]
dist += math.Sqrt(math.Pow(L[b][0]-L[a][0], 2.0) + math.Pow(L[b][1]-L[a][1], 2.0))
}
a, b := ids[nstations-1], ids[0]
dist += math.Sqrt(math.Pow(L[b][0]-L[a][0], 2.0) + math.Pow(L[b][1]-L[a][1], 2.0))
ind.Ovas[0] = dist
return
}
// evolver
evo := NewEvolver(C)
// print initial population
pop := evo.Islands[0].Pop
//io.Pf("\n%v\n", pop.Output(nil, false))
// 0,4,8,11,14,17,18,15,12,19,13,16,10,6,1,3,7,9,5,2 894.363
if false {
for i, x := range []int{0, 4, 8, 11, 14, 17, 18, 15, 12, 19, 13, 16, 10, 6, 1, 3, 7, 9, 5, 2} {
pop[0].Ints[i] = x
}
evo.Islands[0].CalcOvs(pop, 0)
evo.Islands[0].CalcDemeritsAndSort(pop)
}
// check initial population
ints := make([]int, nstations)
if false {
for i := 0; i < C.Ninds; i++ {
for j := 0; j < nstations; j++ {
ints[j] = pop[i].Ints[j]
}
sort.Ints(ints)
chk.Ints(tst, "ints", ints, utl.IntRange(nstations))
}
}
// run
evo.Run()
//io.Pf("%v\n", pop.Output(nil, false))
io.Pfgreen("best = %v\n", evo.Best.Ints)
io.Pfgreen("best OVA = %v (871.117353844847)\n\n", evo.Best.Ovas[0])
// best = [18 17 14 11 8 4 0 2 5 9 12 7 6 1 3 10 16 13 19 15]
// best OVA = 953.4643474956656
// best = [8 11 14 17 18 15 12 19 16 13 10 6 1 3 7 9 5 2 0 4]
// best OVA = 871.117353844847
// best = [5 2 0 4 8 11 14 17 18 15 12 19 16 13 10 6 1 3 7 9]
// best OVA = 871.1173538448469
// best = [6 10 13 16 19 15 18 17 14 11 8 4 0 2 5 9 12 7 3 1]
// best OVA = 880.7760751923065
// check final population
if false {
for i := 0; i < C.Ninds; i++ {
for j := 0; j < nstations; j++ {
ints[j] = pop[i].Ints[j]
}
sort.Ints(ints)
chk.Ints(tst, "ints", ints, utl.IntRange(nstations))
}
}
// plot travelling salesman path
if C.DoPlot {
plt.SetForEps(1, 300)
X, Y := make([]float64, nstations), make([]float64, nstations)
for k, id := range evo.Best.Ints {
X[k], Y[k] = locations[id][0], locations[id][1]
plt.PlotOne(X[k], Y[k], "'r.', ms=5, clip_on=0, zorder=20")
plt.Text(X[k], Y[k], io.Sf("%d", id), "fontsize=7, clip_on=0, zorder=30")
}
plt.Plot(X, Y, "'b-', clip_on=0, zorder=10")
plt.Plot([]float64{X[0], X[nstations-1]}, []float64{Y[0], Y[nstations-1]}, "'b-', clip_on=0, zorder=10")
plt.Equal()
plt.AxisRange(10, 210, 10, 210)
plt.Gll("$x$", "$y$", "")
plt.SaveD("/tmp/goga", "test_evo04.eps")
}
}
func Test_flt02(tst *testing.T) {
//verbose()
chk.PrintTitle("flt02. circle with equality constraint")
// parameters
C := NewConfParams()
C.Eps1 = 1e-3
C.Pll = false
C.Nisl = 4
C.Ninds = 12
C.Ntrials = 1
if chk.Verbose {
C.Ntrials = 40
}
C.Verbose = false
C.Dtmig = 50
C.CrowdSize = 3
C.CompProb = false
C.GAtype = "crowd"
C.DiffEvol = true
C.RangeFlt = [][]float64{
{-1, 3}, // gene # 0: min and max
{-1, 3}, // gene # 1: min and max
}
C.Latin = true
C.PopFltGen = PopFltGen
if chk.Verbose {
C.FnKey = ""
if C.Ntrials == 1 {
C.DoPlot = true
}
}
C.Ops.EnfRange = true
C.NumFmts = map[string][]string{"flt": {"%8.4f", "%8.4f"}}
C.ShowDem = true
C.RegTol = 0.01
C.CalcDerived()
rnd.Init(C.Seed)
// geometry
xe := 1.0 // centre of circle
le := -0.4 // selected level of f(x)
ys := xe - (1.0+le)/math.Sqrt2 // coordinates of minimum point with level=le
y0 := 2.0*ys + xe // vertical axis intersect of straight line defined by c(x)
xc := []float64{xe, xe} // centre
nx := len(xc)
// functions
fcn := func(f, g, h []float64, x []float64) {
res := 0.0
for i := 0; i < nx; i++ {
res += (x[i] - xc[i]) * (x[i] - xc[i])
}
f[0] = math.Sqrt(res) - 1
h[0] = x[0] + x[1] + xe - y0
}
// simple problem
sim := NewSimpleFltProb(fcn, 1, 0, 1, C)
sim.Run(chk.Verbose)
// stat
io.Pf("\n")
sim.Stat(0, 60, -0.4)
// plot
sim.PltExtra = func() {
plt.PlotOne(ys, ys, "'o', markeredgecolor='yellow', markerfacecolor='none', markersize=10")
}
sim.Plot("test_flt02")
}
func Test_flt04(tst *testing.T) {
//verbose()
chk.PrintTitle("flt04. two-bar truss. Pareto-optimal")
// configuration
C := NewConfParams()
C.Nisl = 4
C.Ninds = 24
C.GAtype = "crowd"
C.DiffEvol = true
C.CrowdSize = 3
C.ParetoPhi = 0.05
C.Tf = 100
C.Dtmig = 25
C.RangeFlt = [][]float64{{0.1, 2.25}, {0.5, 2.5}}
C.PopFltGen = PopFltGen
C.CalcDerived()
rnd.Init(C.Seed)
// data
// from Coelho (2007) page 19
ρ := 0.283 // lb/in³
H := 100.0 // in
P := 1e4 // lb
E := 3e7 // lb/in²
σ0 := 2e4 // lb/in²
// functions
TSQ2 := 2.0 * math.Sqrt2
fcn := func(f, g, h []float64, x []float64) {
f[0] = 2.0 * ρ * H * x[1] * math.Sqrt(1.0+x[0]*x[0])
f[1] = P * H * math.Pow(1.0+x[0]*x[0], 1.5) * math.Sqrt(1.0+math.Pow(x[0], 4.0)) / (TSQ2 * E * x[0] * x[0] * x[1])
g[0] = σ0 - P*(1.0+x[0])*math.Sqrt(1.0+x[0]*x[0])/(TSQ2*x[0]*x[1])
g[1] = σ0 - P*(1.0-x[0])*math.Sqrt(1.0+x[0]*x[0])/(TSQ2*x[0]*x[1])
}
// objective value function
C.OvaOor = func(ind *Individual, idIsland, t int, report *bytes.Buffer) {
x := ind.GetFloats()
f := make([]float64, 2)
g := make([]float64, 2)
fcn(f, g, nil, x)
ind.Ovas[0] = f[0]
ind.Ovas[1] = f[1]
ind.Oors[0] = utl.GtePenalty(g[0], 0, 1)
ind.Oors[1] = utl.GtePenalty(g[1], 0, 1)
}
// simple problem
sim := NewSimpleFltProb(fcn, 2, 2, 0, C)
sim.Run(chk.Verbose)
// results
if chk.Verbose {
// reference data
_, dat, _ := io.ReadTable("data/coelho-fig1.6.dat")
// Pareto-front
feasible := sim.Evo.GetFeasible()
ovas, _ := sim.Evo.GetResults(feasible)
ovafront, _ := sim.Evo.GetParetoFront(feasible, ovas, nil)
xova, yova := sim.Evo.GetFrontOvas(0, 1, ovafront)
// plot
plt.SetForEps(0.75, 355)
plt.Plot(dat["f1"], dat["f2"], "'k+',ms=3")
x := utl.DblsGetColumn(0, ovas)
y := utl.DblsGetColumn(1, ovas)
plt.Plot(x, y, "'r.'")
plt.Plot(xova, yova, "'ko',markerfacecolor='none',ms=6")
plt.Gll("$f_1$", "$f_2$", "")
plt.SaveD("/tmp/goga", "test_flt04.eps")
}
}
func Test_mut01(tst *testing.T) {
//verbose()
chk.PrintTitle("mut01")
var ops OpsData
ops.SetDefault()
ops.Pm = 1
rnd.Init(0)
A := []int{1, 2, 3, 4, 5, 6, 7, 8, 9}
ops.Nchanges = 3
ops.Mmax = 10
io.Pforan("before: A = %v\n", A)
IntMutation(A, 0, &ops)
io.Pforan("after: A = %v\n", A)
io.Pf("\n")
B := []float64{1.1, 2.2, 3.3, 4.4, 5.5, 6.6, 7.7, 8.8, 9.9}
io.Pforan("before: B = %v\n", B)
FltMutation(B, 0, &ops)
io.Pforan("after: B = %v\n", B)
io.Pf("\n")
C := []string{"a", "b", "c", "d", "e", "f"}
ops.Nchanges = 2
io.Pforan("before: C = %v\n", C)
StrMutation(C, 0, &ops)
io.Pforan("after: C = %v\n", C)
io.Pf("\n")
D := []byte("abcdefghijklm")
ops.Nchanges = 3
io.Pforan("before: D = %s\n", D)
KeyMutation(D, 0, &ops)
io.Pforan("after: D = %s\n", D)
io.Pf("\n")
E := [][]byte{[]byte("abc"), []byte("def"), []byte("ghi"), []byte("jkl")}
io.Pforan("before: E = %s\n", E)
BytMutation(E, 0, &ops)
io.Pforan("after: E = %s\n", E)
io.Pf("\n")
F := []Func_t{
func(o *Individual) string { return "f0" },
func(o *Individual) string { return "f1" },
func(o *Individual) string { return "f2" },
func(o *Individual) string { return "g0" },
func(o *Individual) string { return "g1" },
func(o *Individual) string { return "g2" },
}
io.Pforan("before: F =")
for _, f := range F {
io.Pforan(" %q", f(nil))
}
FunMutation(F, 0, &ops)
io.Pforan("\nafter: F =")
for _, f := range F {
io.Pforan(" %q", f(nil))
}
io.Pf("\n")
io.Pf("\n")
}