func (p *floorPlan) F2(x, z *matrix.FloatMatrix) (f, Df, H *matrix.FloatMatrix, err error) {
f, Df, err = p.F1(x)
x17 := matrix.FloatVector(x.FloatArray()[17:])
tmp := matrix.Div(p.Amin, matrix.Pow(x17, 3.0))
tmp = matrix.Mul(z, tmp).Scale(2.0)
diag := matrix.FloatDiagonal(5, tmp.FloatArray()...)
H = matrix.FloatZeros(22, 22)
H.SetSubMatrix(17, 17, diag)
return
}
func (p *floorPlan) F1(x *matrix.FloatMatrix) (f, Df *matrix.FloatMatrix, err error) {
err = nil
mn := x.Min(-1, -2, -3, -4, -5)
if mn <= 0.0 {
f, Df = nil, nil
return
}
zeros := matrix.FloatZeros(5, 12)
dk1 := matrix.FloatDiagonal(5, -1.0)
dk2 := matrix.FloatZeros(5, 5)
x17 := matrix.FloatVector(x.FloatArray()[17:])
// -( Amin ./ (x17 .* x17) )
diag := matrix.Div(p.Amin, matrix.Mul(x17, x17)).Scale(-1.0)
dk2.SetIndexesFromArray(diag.FloatArray(), matrix.MakeDiagonalSet(5)...)
Df, _ = matrix.FloatMatrixStacked(matrix.StackRight, zeros, dk1, dk2)
x12 := matrix.FloatVector(x.FloatArray()[12:17])
// f = -x[12:17] + div(Amin, x[17:]) == div(Amin, x[17:]) - x[12:17]
f = matrix.Minus(matrix.Div(p.Amin, x17), x12)
return
}
// Computes analytic center of A*x <= b with A m by n of rank n.
// We assume that b > 0 and the feasible set is bounded.
func Acent(A, b *matrix.FloatMatrix, niters int) (*matrix.FloatMatrix, []float64) {
if niters <= 0 {
niters = MAXITERS
}
ntdecrs := make([]float64, 0, niters)
if A.Rows() != b.Rows() {
return nil, nil
}
m, n := A.Size()
x := matrix.FloatZeros(n, 1)
H := matrix.FloatZeros(n, n)
// Helper m*n matrix
Dmn := matrix.FloatZeros(m, n)
for i := 0; i < niters; i++ {
// Gradient is g = A^T * (1.0/(b - A*x)). d = 1.0/(b - A*x)
// d is m*1 matrix, g is n*1 matrix
d := matrix.Minus(b, matrix.Times(A, x)).Inv()
g := matrix.Times(A.Transpose(), d)
// Hessian is H = A^T * diag(1./(b-A*x))^2 * A.
// in the original python code expression d[:,n*[0]] creates
// a m*n matrix where each column is copy of column 0.
// We do it here manually.
for i := 0; i < n; i++ {
Dmn.SetColumn(i, d)
}
// Function mul creates element wise product of matrices.
Asc := matrix.Mul(Dmn, A)
blas.SyrkFloat(Asc, H, 1.0, 0.0, linalg.OptTrans)
// Newton step is v = H^-1 * g.
v := g.Copy().Scale(-1.0)
lapack.PosvFloat(H, v)
// Directional derivative and Newton decrement.
lam := blas.DotFloat(g, v)
ntdecrs = append(ntdecrs, math.Sqrt(-lam))
if ntdecrs[len(ntdecrs)-1] < TOL {
fmt.Printf("last Newton decrement < TOL(%v)\n", TOL)
return x, ntdecrs
}
// Backtracking line search.
// y = d .* A*v
y := d.Mul(A.Times(v))
step := 1.0
for 1-step*y.Max() < 0 {
step *= BETA
}
search:
for {
// t = -step*y
t := y.Copy().Scale(-step)
// t = (1 + t) [e.g. t = 1 - step*y]
t.Add(1.0)
// ts = sum(log(1-step*y))
ts := t.Log().Sum()
if -ts < ALPHA*step*lam {
break search
}
step *= BETA
}
v.Scale(step)
x = x.Plus(v)
}
// no solution !!
fmt.Printf("Iteration %d exhausted\n", niters)
return x, ntdecrs
}
func main() {
m := 6
Vdata := [][]float64{
[]float64{1.0, -1.0, -2.0, -2.0, 0.0, 1.5, 1.0},
[]float64{1.0, 2.0, 1.0, -1.0, -2.0, -1.0, 1.0}}
V := matrix.FloatMatrixFromTable(Vdata, matrix.RowOrder)
// V[1, :m] - V[1,1:]
a0 := matrix.Minus(V.GetSubMatrix(1, 0, 1, m), V.GetSubMatrix(1, 1, 1))
// V[0, :m] - V[0,1:]
a1 := matrix.Minus(V.GetSubMatrix(0, 0, 1, m), V.GetSubMatrix(0, 1, 1))
A0, _ := matrix.FloatMatrixStacked(matrix.StackDown, a0.Scale(-1.0), a1)
A0 = A0.Transpose()
b0 := matrix.Mul(A0, V.GetSubMatrix(0, 0, 2, m).Transpose())
b0 = matrix.Times(b0, matrix.FloatWithValue(2, 1, 1.0))
A := make([]*matrix.FloatMatrix, 0)
b := make([]*matrix.FloatMatrix, 0)
A = append(A, A0)
b = append(b, b0)
// List of symbols
C := make([]*matrix.FloatMatrix, 0)
C = append(C, matrix.FloatZeros(2, 1))
var row *matrix.FloatMatrix = nil
for k := 0; k < m; k++ {
row = A0.GetRow(k, row)
nrm := blas.Nrm2Float(row)
row.Scale(2.0 * b0.GetIndex(k) / (nrm * nrm))
C = append(C, row.Transpose())
}
// Voronoi set around C[1]
A1 := matrix.FloatZeros(3, 2)
A1.SetSubMatrix(0, 0, A0.GetSubMatrix(0, 0, 1).Scale(-1.0))
A1.SetSubMatrix(1, 0, matrix.Minus(C[m], C[1]).Transpose())
A1.SetSubMatrix(2, 0, matrix.Minus(C[2], C[1]).Transpose())
b1 := matrix.FloatZeros(3, 1)
b1.SetIndex(0, -b0.GetIndex(0))
v := matrix.Times(A1.GetRow(1, nil), matrix.Plus(C[m], C[1])).Float() * 0.5
b1.SetIndex(1, v)
v = matrix.Times(A1.GetRow(2, nil), matrix.Plus(C[2], C[1])).Float() * 0.5
b1.SetIndex(2, v)
A = append(A, A1)
b = append(b, b1)
// Voronoi set around C[2] ... C[5]
for k := 2; k < 6; k++ {
A1 = matrix.FloatZeros(3, 2)
A1.SetSubMatrix(0, 0, A0.GetSubMatrix(k-1, 0, 1).Scale(-1.0))
A1.SetSubMatrix(1, 0, matrix.Minus(C[k-1], C[k]).Transpose())
A1.SetSubMatrix(2, 0, matrix.Minus(C[k+1], C[k]).Transpose())
b1 = matrix.FloatZeros(3, 1)
b1.SetIndex(0, -b0.GetIndex(k-1))
v := matrix.Times(A1.GetRow(1, nil), matrix.Plus(C[k-1], C[k])).Float() * 0.5
b1.SetIndex(1, v)
v = matrix.Times(A1.GetRow(2, nil), matrix.Plus(C[k+1], C[k])).Float() * 0.5
b1.SetIndex(2, v)
A = append(A, A1)
b = append(b, b1)
}
// Voronoi set around C[6]
A1 = matrix.FloatZeros(3, 2)
A1.SetSubMatrix(0, 0, A0.GetSubMatrix(5, 0, 1).Scale(-1.0))
A1.SetSubMatrix(1, 0, matrix.Minus(C[1], C[6]).Transpose())
A1.SetSubMatrix(2, 0, matrix.Minus(C[5], C[6]).Transpose())
b1 = matrix.FloatZeros(3, 1)
b1.SetIndex(0, -b0.GetIndex(5))
v = matrix.Times(A1.GetRow(1, nil), matrix.Plus(C[1], C[6])).Float() * 0.5
b1.SetIndex(1, v)
v = matrix.Times(A1.GetRow(2, nil), matrix.Plus(C[5], C[6])).Float() * 0.5
b1.SetIndex(2, v)
A = append(A, A1)
b = append(b, b1)
P := matrix.FloatIdentity(2)
q := matrix.FloatZeros(2, 1)
solopts := &cvx.SolverOptions{ShowProgress: false, MaxIter: 30}
ovals := make([]float64, 0)
for k := 1; k < 7; k++ {
sol, err := cvx.Qp(P, q, A[k], b[k], nil, nil, solopts, nil)
_ = err
x := sol.Result.At("x")[0]
ovals = append(ovals, math.Pow(blas.Nrm2Float(x), 2.0))
}
optvals := matrix.FloatVector(ovals)
//fmt.Printf("optvals=\n%v\n", optvals)
rangeFunc := func(n int) []float64 {
r := make([]float64, 0)
for i := 0; i < n; i++ {
r = append(r, float64(i))
}
return r
}
nopts := 200
sigmas := matrix.FloatVector(rangeFunc(nopts))
sigmas.Scale((0.5 - 0.2) / float64(nopts)).Add(0.2)
bndsVal := func(sigma float64) float64 {
// 1.0 - sum(exp( -optvals/(2*sigma**2)))
return 1.0 - matrix.Exp(matrix.Scale(optvals, -1.0/(2*sigma*sigma))).Sum()
}
bnds := matrix.FloatZeros(sigmas.NumElements(), 1)
for j, v := range sigmas.FloatArray() {
bnds.SetIndex(j, bndsVal(v))
}
plotData("plot.png", sigmas.FloatArray(), bnds.FloatArray())
}
func main() {
flag.Parse()
m := len(udata)
nvars := 2 * m
u := matrix.FloatVector(udata[:m])
y := matrix.FloatVector(ydata[:m])
// minimize (1/2) * || yhat - y ||_2^2
// subject to yhat[j] >= yhat[i] + g[i]' * (u[j] - u[i]), j, i = 0,...,m-1
//
// Variables yhat (m), g (m).
P := matrix.FloatZeros(nvars, nvars)
// set m first diagonal indexes to 1.0
//P.SetIndexes(1.0, matrix.DiagonalIndexes(P)[:m]...)
P.Diag().SubMatrix(0, 0, 1, m).SetIndexes(1.0)
q := matrix.FloatZeros(nvars, 1)
q.SubMatrix(0, 0, y.NumElements(), 1).Plus(matrix.Scale(y, -1.0))
// m blocks (i = 0,...,m-1) of linear inequalities
//
// yhat[i] + g[i]' * (u[j] - u[i]) <= yhat[j], j = 0,...,m-1.
G := matrix.FloatZeros(m*m, nvars)
I := matrix.FloatDiagonal(m, 1.0)
for i := 0; i < m; i++ {
// coefficients of yhat[i] (column i)
//G.Set(1.0, matrix.ColumnIndexes(G, i)[i*m:(i+1)*m]...)
column(G, i).SetIndexes(1.0)
// coefficients of gi[i] (column i, rows i*m ... (i+1)*m)
//rows := matrix.Indexes(i*m, (i+1)*m)
//G.SetAtColumnArray(m+i, rows, matrix.Add(u, -u.GetIndex(i)).FloatArray())
// coefficients of gi[i] (column i, rows i*m ... (i+1)*m)
// from column m+i staring at row i*m select m rows and one column
G.SubMatrix(i*m, m+i, m, 1).Plus(matrix.Add(u, -u.GetIndex(i)))
// coeffients of yhat[i]) from rows i*m ... (i+1)*m, cols 0 ... m
//G.SetSubMatrix(i*m, 0, matrix.Minus(G.GetSubMatrix(i*m, 0, m, m), I))
G.SubMatrix(i*m, 0, m, m).Minus(I)
}
h := matrix.FloatZeros(m*m, 1)
var A, b *matrix.FloatMatrix = nil, nil
var solopts cvx.SolverOptions
solopts.ShowProgress = true
solopts.KKTSolverName = solver
sol, err := cvx.Qp(P, q, G, h, A, b, &solopts, nil)
if err != nil {
fmt.Printf("error: %v\n", err)
return
}
if sol != nil && sol.Status != cvx.Optimal {
fmt.Printf("status not optimal\n")
return
}
x := sol.Result.At("x")[0]
//yhat := matrix.FloatVector(x.FloatArray()[:m])
//g := matrix.FloatVector(x.FloatArray()[m:])
yhat := x.SubMatrix(0, 0, m, 1).Copy()
g := x.SubMatrix(m, 0).Copy()
rangeFunc := func(n int) []float64 {
r := make([]float64, 0)
for i := 0; i < n; i++ {
r = append(r, float64(i)*2.2/float64(n))
}
return r
}
ts := rangeFunc(1000)
fitFunc := func(points []float64) []float64 {
res := make([]float64, len(points))
for k, t := range points {
res[k] = matrix.Plus(yhat, matrix.Mul(g, matrix.Scale(u, -1.0).Add(t))).Max()
}
return res
}
fs := fitFunc(ts)
plotData("cvxfit.png", u.FloatArray(), y.FloatArray(), ts, fs)
}
func mcsdp(w *matrix.FloatMatrix) (*Solution, error) {
//
// Returns solution x, z to
//
// (primal) minimize sum(x)
// subject to w + diag(x) >= 0
//
// (dual) maximize -tr(w*z)
// subject to diag(z) = 1
// z >= 0.
//
n := w.Rows()
G := &matrixFs{n}
cngrnc := func(r, x *matrix.FloatMatrix, alpha float64) (err error) {
// Congruence transformation
//
// x := alpha * r'*x*r.
//
// r and x are square matrices.
//
err = nil
// tx = matrix(x, (n,n)) is copying and reshaping
// scale diagonal of x by 1/2, (x is (n,n))
tx := x.Copy()
matrix.Reshape(tx, n, n)
tx.Diag().Scale(0.5)
// a := tril(x)*r
// (python: a = +r is really making a copy of r)
a := r.Copy()
err = blas.TrmmFloat(tx, a, 1.0, linalg.OptLeft)
// x := alpha*(a*r' + r*a')
err = blas.Syr2kFloat(r, a, tx, alpha, 0.0, linalg.OptTrans)
// x[:] = tx[:]
tx.CopyTo(x)
return
}
Fkkt := func(W *sets.FloatMatrixSet) (KKTFunc, error) {
// Solve
// -diag(z) = bx
// -diag(x) - inv(rti*rti') * z * inv(rti*rti') = bs
//
// On entry, x and z contain bx and bs.
// On exit, they contain the solution, with z scaled
// (inv(rti)'*z*inv(rti) is returned instead of z).
//
// We first solve
//
// ((rti*rti') .* (rti*rti')) * x = bx - diag(t*bs*t)
//
// and take z = -rti' * (diag(x) + bs) * rti.
var err error = nil
rti := W.At("rti")[0]
// t = rti*rti' as a nonsymmetric matrix.
t := matrix.FloatZeros(n, n)
err = blas.GemmFloat(rti, rti, t, 1.0, 0.0, linalg.OptTransB)
if err != nil {
return nil, err
}
// Cholesky factorization of tsq = t.*t.
tsq := matrix.Mul(t, t)
err = lapack.Potrf(tsq)
if err != nil {
return nil, err
}
f := func(x, y, z *matrix.FloatMatrix) (err error) {
// tbst := t * zs * t = t * bs * t
tbst := z.Copy()
matrix.Reshape(tbst, n, n)
cngrnc(t, tbst, 1.0)
// x := x - diag(tbst) = bx - diag(rti*rti' * bs * rti*rti')
diag := tbst.Diag().Transpose()
x.Minus(diag)
// x := (t.*t)^{-1} * x = (t.*t)^{-1} * (bx - diag(t*bs*t))
err = lapack.Potrs(tsq, x)
// z := z + diag(x) = bs + diag(x)
// z, x are really column vectors here
z.AddIndexes(matrix.MakeIndexSet(0, n*n, n+1), x.FloatArray())
// z := -rti' * z * rti = -rti' * (diag(x) + bs) * rti
cngrnc(rti, z, -1.0)
return nil
}
return f, nil
}
c := matrix.FloatWithValue(n, 1, 1.0)
// initial feasible x: x = 1.0 - min(lmbda(w))
lmbda := matrix.FloatZeros(n, 1)
wp := w.Copy()
lapack.Syevx(wp, lmbda, nil, 0.0, nil, []int{1, 1}, linalg.OptRangeInt)
x0 := matrix.FloatZeros(n, 1).Add(-lmbda.GetAt(0, 0) + 1.0)
s0 := w.Copy()
s0.Diag().Plus(x0.Transpose())
matrix.Reshape(s0, n*n, 1)
// initial feasible z is identity
z0 := matrix.FloatIdentity(n)
matrix.Reshape(z0, n*n, 1)
dims := sets.DSetNew("l", "q", "s")
dims.Set("s", []int{n})
primalstart := sets.FloatSetNew("x", "s")
dualstart := sets.FloatSetNew("z")
primalstart.Set("x", x0)
primalstart.Set("s", s0)
dualstart.Set("z", z0)
var solopts SolverOptions
solopts.MaxIter = 30
solopts.ShowProgress = false
h := w.Copy()
matrix.Reshape(h, h.NumElements(), 1)
return ConeLpCustomMatrix(c, G, h, nil, nil, dims, Fkkt, &solopts, primalstart, dualstart)
}
func qcl1(A, b *matrix.FloatMatrix) (*cvx.Solution, error) {
// Returns the solution u, z of
//
// (primal) minimize || u ||_1
// subject to || A * u - b ||_2 <= 1
//
// (dual) maximize b^T z - ||z||_2
// subject to || A'*z ||_inf <= 1.
//
// Exploits structure, assuming A is m by n with m >= n.
m, n := A.Size()
Fkkt := func(W *sets.FloatMatrixSet) (f cvx.KKTFunc, err error) {
minor := 0
if !checkpnt.MinorEmpty() {
minor = checkpnt.MinorTop()
}
err = nil
f = nil
beta := W.At("beta")[0].GetIndex(0)
v := W.At("v")[0]
// As = 2 * v *(v[1:].T * A)
//v_1 := matrix.FloatNew(1, v.NumElements()-1, v.FloatArray()[1:])
v_1 := v.SubMatrix(1, 0).Transpose()
As := matrix.Times(v, matrix.Times(v_1, A)).Scale(2.0)
//As_1 := As.GetSubMatrix(1, 0, m, n)
//As_1.Scale(-1.0)
//As.SetSubMatrix(1, 0, matrix.Minus(As_1, A))
As_1 := As.SubMatrix(1, 0, m, n)
As_1.Scale(-1.0)
As_1.Minus(A)
As.Scale(1.0 / beta)
S := matrix.Times(As.Transpose(), As)
checkpnt.AddMatrixVar("S", S)
d1 := W.At("d")[0].SubMatrix(0, 0, n, 1).Copy()
d2 := W.At("d")[0].SubMatrix(n, 0).Copy()
// D = 4.0 * (d1**2 + d2**2)**-1
d := matrix.Plus(matrix.Mul(d1, d1), matrix.Mul(d2, d2)).Inv().Scale(4.0)
// S[::n+1] += d
S.Diag().Plus(d.Transpose())
err = lapack.Potrf(S)
checkpnt.Check("00-Fkkt", minor)
if err != nil {
return
}
f = func(x, y, z *matrix.FloatMatrix) (err error) {
minor := 0
if !checkpnt.MinorEmpty() {
minor = checkpnt.MinorTop()
} else {
loopf += 1
minor = loopf
}
checkpnt.Check("00-f", minor)
// -- z := - W**-T * z
// z[:n] = -div( z[:n], d1 )
z_val := z.SubMatrix(0, 0, n, 1)
z_res := matrix.Div(z_val, d1).Scale(-1.0)
z.SubMatrix(0, 0, n, 1).Set(z_res)
// z[n:2*n] = -div( z[n:2*n], d2 )
z_val = z.SubMatrix(n, 0, n, 1)
z_res = matrix.Div(z_val, d2).Scale(-1.0)
z.SubMatrix(n, 0, n, 1).Set(z_res)
// z[2*n:] -= 2.0*v*( v[0]*z[2*n] - blas.dot(v[1:], z[2*n+1:]) )
v0_z2n := v.GetIndex(0) * z.GetIndex(2*n)
v1_z2n := blas.DotFloat(v, z, &linalg.IOpt{"offsetx", 1}, &linalg.IOpt{"offsety", 2*n + 1})
z_res = matrix.Scale(v, -2.0*(v0_z2n-v1_z2n))
z.SubMatrix(2*n, 0, z_res.NumElements(), 1).Plus(z_res)
// z[2*n+1:] *= -1.0
z.SubMatrix(2*n+1, 0).Scale(-1.0)
// z[2*n:] /= beta
z.SubMatrix(2*n, 0).Scale(1.0 / beta)
// -- x := x - G' * W**-1 * z
// z_n = z[:n], z_2n = z[n:2*n], z_m = z[-(m+1):],
z_n := z.SubMatrix(0, 0, n, 1)
z_2n := z.SubMatrix(n, 0, n, 1)
z_m := z.SubMatrix(z.NumElements()-(m+1), 0)
// x[:n] -= div(z[:n], d1) - div(z[n:2*n], d2) + As.T * z[-(m+1):]
z_res = matrix.Minus(matrix.Div(z_n, d1), matrix.Div(z_2n, d2))
a_res := matrix.Times(As.Transpose(), z_m)
z_res.Plus(a_res).Scale(-1.0)
x.SubMatrix(0, 0, n, 1).Plus(z_res)
// x[n:] += div(z[:n], d1) + div(z[n:2*n], d2)
z_res = matrix.Plus(matrix.Div(z_n, d1), matrix.Div(z_2n, d2))
x.SubMatrix(n, 0, z_res.NumElements(), 1).Plus(z_res)
checkpnt.Check("15-f", minor)
// Solve for x[:n]:
//
// S*x[:n] = x[:n] - (W1**2 - W2**2)(W1**2 + W2**2)^-1 * x[n:]
// w1 = (d1**2 - d2**2), w2 = (d1**2 + d2**2)
w1 := matrix.Minus(matrix.Mul(d1, d1), matrix.Mul(d2, d2))
w2 := matrix.Plus(matrix.Mul(d1, d1), matrix.Mul(d2, d2))
// x[:n] += -mul( div(w1, w2), x[n:])
x_n := x.SubMatrix(n, 0)
x_val := matrix.Mul(matrix.Div(w1, w2), x_n).Scale(-1.0)
x.SubMatrix(0, 0, n, 1).Plus(x_val)
checkpnt.Check("25-f", minor)
// Solve for x[n:]:
//
// (d1**-2 + d2**-2) * x[n:] = x[n:] + (d1**-2 - d2**-2)*x[:n]
err = lapack.Potrs(S, x)
if err != nil {
fmt.Printf("Potrs error: %s\n", err)
}
checkpnt.Check("30-f", minor)
// Solve for x[n:]:
//
// (d1**-2 + d2**-2) * x[n:] = x[n:] + (d1**-2 - d2**-2)*x[:n]
// w1 = (d1**-2 - d2**-2), w2 = (d1**-2 + d2**-2)
w1 = matrix.Minus(matrix.Mul(d1, d1).Inv(), matrix.Mul(d2, d2).Inv())
w2 = matrix.Plus(matrix.Mul(d1, d1).Inv(), matrix.Mul(d2, d2).Inv())
x_n = x.SubMatrix(0, 0, n, 1)
// x[n:] += mul( d1**-2 - d2**-2, x[:n])
x_val = matrix.Mul(w1, x_n)
x.SubMatrix(n, 0, x_val.NumElements(), 1).Plus(x_val)
checkpnt.Check("35-f", minor)
// x[n:] = div( x[n:], d1**-2 + d2**-2)
x_n = x.SubMatrix(n, 0)
x_val = matrix.Div(x_n, w2)
x.SubMatrix(n, 0, x_val.NumElements(), 1).Set(x_val)
checkpnt.Check("40-f", minor)
// x_n = x[:n], x-2n = x[n:2*n]
x_n = x.SubMatrix(0, 0, n, 1)
x_2n := x.SubMatrix(n, 0, n, 1)
// z := z + W^-T * G*x
// z[:n] += div( x[:n] - x[n:2*n], d1)
x_val = matrix.Div(matrix.Minus(x_n, x_2n), d1)
z.SubMatrix(0, 0, n, 1).Plus(x_val)
checkpnt.Check("44-f", minor)
// z[n:2*n] += div( -x[:n] - x[n:2*n], d2)
x_val = matrix.Div(matrix.Plus(x_n, x_2n).Scale(-1.0), d2)
z.SubMatrix(n, 0, n, 1).Plus(x_val)
checkpnt.Check("48-f", minor)
// z[2*n:] += As*x[:n]
x_val = matrix.Times(As, x_n)
z.SubMatrix(2*n, 0, x_val.NumElements(), 1).Plus(x_val)
checkpnt.Check("50-f", minor)
return nil
}
return
}
// matrix(n*[0.0] + n*[1.0])
c := matrix.FloatZeros(2*n, 1)
c.SubMatrix(n, 0).SetIndexes(1.0)
h := matrix.FloatZeros(2*n+m+1, 1)
h.SetIndexes(1.0, 2*n)
// h[2*n+1:] = -b
h.SubMatrix(2*n+1, 0).Plus(b).Scale(-1.0)
G := &matrixFs{A}
dims := sets.DSetNew("l", "q", "s")
dims.Set("l", []int{2 * n})
dims.Set("q", []int{m + 1})
var solopts cvx.SolverOptions
solopts.ShowProgress = true
if maxIter > 0 {
solopts.MaxIter = maxIter
}
if len(solver) > 0 {
solopts.KKTSolverName = solver
}
return cvx.ConeLpCustomMatrix(c, G, h, nil, nil, dims, Fkkt, &solopts, nil, nil)
}
/*
Returns the Nesterov-Todd scaling W at points s and z, and stores the
scaled variable in lmbda.
W * z = W^{-T} * s = lmbda.
W is a MatrixSet with entries:
- W['dnl']: positive vector
- W['dnli']: componentwise inverse of W['dnl']
- W['d']: positive vector
- W['di']: componentwise inverse of W['d']
- W['v']: lists of 2nd order cone vectors with unit hyperbolic norms
- W['beta']: list of positive numbers
- W['r']: list of square matrices
- W['rti']: list of square matrices. rti[k] is the inverse transpose
of r[k].
*/
func computeScaling(s, z, lmbda *matrix.FloatMatrix, dims *sets.DimensionSet, mnl int) (W *sets.FloatMatrixSet, err error) {
/*DEBUGGED*/
err = nil
W = sets.NewFloatSet("dnl", "dnli", "d", "di", "v", "beta", "r", "rti")
// For the nonlinear block:
//
// W['dnl'] = sqrt( s[:mnl] ./ z[:mnl] )
// W['dnli'] = sqrt( z[:mnl] ./ s[:mnl] )
// lambda[:mnl] = sqrt( s[:mnl] .* z[:mnl] )
var stmp, ztmp, lmd *matrix.FloatMatrix
if mnl > 0 {
stmp = matrix.FloatVector(s.FloatArray()[:mnl])
ztmp = matrix.FloatVector(z.FloatArray()[:mnl])
//dnl := stmp.Div(ztmp)
//dnl.Apply(dnl, math.Sqrt)
dnl := matrix.Sqrt(matrix.Div(stmp, ztmp))
//dnli := dnl.Copy()
//dnli.Apply(dnli, func(a float64)float64 { return 1.0/a })
dnli := matrix.Inv(dnl)
W.Set("dnl", dnl)
W.Set("dnli", dnli)
//lmd = stmp.Mul(ztmp)
//lmd.Apply(lmd, math.Sqrt)
lmd = matrix.Sqrt(matrix.Mul(stmp, ztmp))
lmbda.SetIndexesFromArray(lmd.FloatArray(), matrix.MakeIndexSet(0, mnl, 1)...)
} else {
// set for empty matrices
//W.Set("dnl", matrix.FloatZeros(0, 1))
//W.Set("dnli", matrix.FloatZeros(0, 1))
mnl = 0
}
// For the 'l' block:
//
// W['d'] = sqrt( sk ./ zk )
// W['di'] = sqrt( zk ./ sk )
// lambdak = sqrt( sk .* zk )
//
// where sk and zk are the first dims['l'] entries of s and z.
// lambda_k is stored in the first dims['l'] positions of lmbda.
m := dims.At("l")[0]
//td := s.FloatArray()
stmp = matrix.FloatVector(s.FloatArray()[mnl : mnl+m])
//zd := z.FloatArray()
ztmp = matrix.FloatVector(z.FloatArray()[mnl : mnl+m])
//fmt.Printf(".Sqrt()=\n%v\n", matrix.Div(stmp, ztmp).Sqrt().ToString("%.17f"))
//d := stmp.Div(ztmp)
//d.Apply(d, math.Sqrt)
d := matrix.Div(stmp, ztmp).Sqrt()
//di := d.Copy()
//di.Apply(di, func(a float64)float64 { return 1.0/a })
di := matrix.Inv(d)
//fmt.Printf("d:\n%v\n", d)
//fmt.Printf("di:\n%v\n", di)
W.Set("d", d)
W.Set("di", di)
//lmd = stmp.Mul(ztmp)
//lmd.Apply(lmd, math.Sqrt)
lmd = matrix.Mul(stmp, ztmp).Sqrt()
// lmd has indexes mnl:mnl+m and length of m
lmbda.SetIndexesFromArray(lmd.FloatArray(), matrix.MakeIndexSet(mnl, mnl+m, 1)...)
//fmt.Printf("after l:\n%v\n", lmbda)
/*
For the 'q' blocks, compute lists 'v', 'beta'.
The vector v[k] has unit hyperbolic norm:
(sqrt( v[k]' * J * v[k] ) = 1 with J = [1, 0; 0, -I]).
beta[k] is a positive scalar.
The hyperbolic Householder matrix H = 2*v[k]*v[k]' - J
defined by v[k] satisfies
(beta[k] * H) * zk = (beta[k] * H) \ sk = lambda_k
where sk = s[indq[k]:indq[k+1]], zk = z[indq[k]:indq[k+1]].
lambda_k is stored in lmbda[indq[k]:indq[k+1]].
*/
ind := mnl + dims.At("l")[0]
var beta *matrix.FloatMatrix
for _, k := range dims.At("q") {
W.Append("v", matrix.FloatZeros(k, 1))
}
beta = matrix.FloatZeros(len(dims.At("q")), 1)
W.Set("beta", beta)
vset := W.At("v")
for k, m := range dims.At("q") {
v := vset[k]
// a = sqrt( sk' * J * sk ) where J = [1, 0; 0, -I]
aa := jnrm2(s, m, ind)
// b = sqrt( zk' * J * zk )
bb := jnrm2(z, m, ind)
// beta[k] = ( a / b )**1/2
beta.SetIndex(k, math.Sqrt(aa/bb))
// c = sqrt( (sk/a)' * (zk/b) + 1 ) / sqrt(2)
c0 := blas.DotFloat(s, z, &la_.IOpt{"n", m},
&la_.IOpt{"offsetx", ind}, &la_.IOpt{"offsety", ind})
cc := math.Sqrt((c0/aa/bb + 1.0) / 2.0)
// vk = 1/(2*c) * ( (sk/a) + J * (zk/b) )
blas.CopyFloat(z, v, &la_.IOpt{"offsetx", ind}, &la_.IOpt{"n", m})
blas.ScalFloat(v, -1.0/bb)
v.SetIndex(0, -1.0*v.GetIndex(0))
blas.AxpyFloat(s, v, 1.0/aa, &la_.IOpt{"offsetx", ind}, &la_.IOpt{"n", m})
blas.ScalFloat(v, 1.0/2.0/cc)
// v[k] = 1/sqrt(2*(vk0 + 1)) * ( vk + e ), e = [1; 0]
v.SetIndex(0, v.GetIndex(0)+1.0)
blas.ScalFloat(v, (1.0 / math.Sqrt(2.0*v.GetIndex(0))))
/*
To get the scaled variable lambda_k
d = sk0/a + zk0/b + 2*c
lambda_k = [ c;
(c + zk0/b)/d * sk1/a + (c + sk0/a)/d * zk1/b ]
lambda_k *= sqrt(a * b)
*/
lmbda.SetIndex(ind, cc)
dd := 2*cc + s.GetIndex(ind)/aa + z.GetIndex(ind)/bb
blas.CopyFloat(s, lmbda, &la_.IOpt{"offsetx", ind + 1}, &la_.IOpt{"offsety", ind + 1},
&la_.IOpt{"n", m - 1})
zz := (cc + z.GetIndex(ind)/bb) / dd / aa
ss := (cc + s.GetIndex(ind)/aa) / dd / bb
blas.ScalFloat(lmbda, zz, &la_.IOpt{"offset", ind + 1}, &la_.IOpt{"n", m - 1})
blas.AxpyFloat(z, lmbda, ss, &la_.IOpt{"offsetx", ind + 1},
&la_.IOpt{"offsety", ind + 1}, &la_.IOpt{"n", m - 1})
blas.ScalFloat(lmbda, math.Sqrt(aa*bb), &la_.IOpt{"offset", ind}, &la_.IOpt{"n", m})
ind += m
//fmt.Printf("after q[%d]:\n%v\n", k, lmbda)
}
/*
For the 's' blocks: compute two lists 'r' and 'rti'.
r[k]' * sk^{-1} * r[k] = diag(lambda_k)^{-1}
r[k]' * zk * r[k] = diag(lambda_k)
where sk and zk are the entries inds[k] : inds[k+1] of
s and z, reshaped into symmetric matrices.
rti[k] is the inverse of r[k]', so
rti[k]' * sk * rti[k] = diag(lambda_k)^{-1}
rti[k]' * zk^{-1} * rti[k] = diag(lambda_k).
The vectors lambda_k are stored in
lmbda[ dims['l'] + sum(dims['q']) : -1 ]
*/
for _, k := range dims.At("s") {
W.Append("r", matrix.FloatZeros(k, k))
W.Append("rti", matrix.FloatZeros(k, k))
}
maxs := maxdim(dims.At("s"))
work := matrix.FloatZeros(maxs*maxs, 1)
Ls := matrix.FloatZeros(maxs*maxs, 1)
Lz := matrix.FloatZeros(maxs*maxs, 1)
ind2 := ind
for k, m := range dims.At("s") {
r := W.At("r")[k]
rti := W.At("rti")[k]
// Factor sk = Ls*Ls'; store Ls in ds[inds[k]:inds[k+1]].
blas.CopyFloat(s, Ls, &la_.IOpt{"offsetx", ind2}, &la_.IOpt{"n", m * m})
lapack.PotrfFloat(Ls, &la_.IOpt{"n", m}, &la_.IOpt{"lda", m})
// Factor zs[k] = Lz*Lz'; store Lz in dz[inds[k]:inds[k+1]].
blas.CopyFloat(z, Lz, &la_.IOpt{"offsetx", ind2}, &la_.IOpt{"n", m * m})
lapack.PotrfFloat(Lz, &la_.IOpt{"n", m}, &la_.IOpt{"lda", m})
// SVD Lz'*Ls = U*diag(lambda_k)*V'. Keep U in work.
for i := 0; i < m; i++ {
blas.ScalFloat(Ls, 0.0, &la_.IOpt{"offset", i * m}, &la_.IOpt{"n", i})
}
blas.CopyFloat(Ls, work, &la_.IOpt{"n", m * m})
blas.TrmmFloat(Lz, work, 1.0, la_.OptTransA, &la_.IOpt{"lda", m}, &la_.IOpt{"ldb", m},
&la_.IOpt{"n", m}, &la_.IOpt{"m", m})
lapack.GesvdFloat(work, lmbda, nil, nil,
la_.OptJobuO, &la_.IOpt{"lda", m}, &la_.IOpt{"offsetS", ind},
&la_.IOpt{"n", m}, &la_.IOpt{"m", m})
// r = Lz^{-T} * U
blas.CopyFloat(work, r, &la_.IOpt{"n", m * m})
blas.TrsmFloat(Lz, r, 1.0, la_.OptTransA,
&la_.IOpt{"lda", m}, &la_.IOpt{"n", m}, &la_.IOpt{"m", m})
// rti = Lz * U
blas.CopyFloat(work, rti, &la_.IOpt{"n", m * m})
blas.TrmmFloat(Lz, rti, 1.0,
&la_.IOpt{"lda", m}, &la_.IOpt{"n", m}, &la_.IOpt{"m", m})
// r := r * diag(sqrt(lambda_k))
// rti := rti * diag(1 ./ sqrt(lambda_k))
for i := 0; i < m; i++ {
a := math.Sqrt(lmbda.GetIndex(ind + i))
blas.ScalFloat(r, a, &la_.IOpt{"offset", m * i}, &la_.IOpt{"n", m})
blas.ScalFloat(rti, 1.0/a, &la_.IOpt{"offset", m * i}, &la_.IOpt{"n", m})
}
ind += m
ind2 += m * m
}
return
}