// CompareJacMpi compares Jacobian matrix (e.g. for testing)
func CompareJacMpi(tst *testing.T, ffcn Cb_f, Jfcn Cb_J, x []float64, tol float64, distr bool) {
// numerical
n := len(x)
fx := make([]float64, n)
w := make([]float64, n) // workspace
ffcn(fx, x)
var Jnum la.Triplet
Jnum.Init(n, n, n*n)
JacobianMpi(&Jnum, ffcn, x, fx, w, distr)
jn := Jnum.ToMatrix(nil)
// analytical
var Jana la.Triplet
Jana.Init(n, n, n*n)
Jfcn(&Jana, x)
ja := Jana.ToMatrix(nil)
// compare
max_diff := la.MatMaxDiff(jn.ToDense(), ja.ToDense())
if max_diff > tol {
tst.Errorf("[1;31mmax_diff = %g[0m\n", max_diff)
} else {
io.Pf("[1;32mmax_diff = %g[0m\n", max_diff)
}
}
func CompareJac(tst *testing.T, ffcn Cb_f, Jfcn Cb_J, x []float64, tol float64, distr bool) {
n := len(x)
// numerical
fx := make([]float64, n)
w := make([]float64, n) // workspace
ffcn(fx, x)
var Jnum la.Triplet
Jnum.Init(n, n, n*n)
Jacobian(&Jnum, ffcn, x, fx, w, distr)
jn := Jnum.ToMatrix(nil)
// analytical
var Jana la.Triplet
Jana.Init(n, n, n*n)
Jfcn(&Jana, x)
ja := Jana.ToMatrix(nil)
// compare
//la.PrintMat(fmt.Sprintf("Jana(%d)",mpi.Rank()), ja.ToDense(), "%13.6f", false)
//la.PrintMat(fmt.Sprintf("Jnum(%d)",mpi.Rank()), jn.ToDense(), "%13.6f", false)
max_diff := la.MatMaxDiff(jn.ToDense(), ja.ToDense())
if max_diff > tol {
tst.Errorf("[1;31mmax_diff = %g[0m\n", max_diff)
} else {
io.Pf("[1;32mmax_diff = %g[0m\n", max_diff)
}
}
// CheckJ check Jacobian matrix
// Ouptut: cnd -- condition number (with Frobenius norm)
func (o *NlSolver) CheckJ(x []float64, tol float64, chkJnum, silent bool) (cnd float64, err error) {
// Jacobian matrix
var Jmat [][]float64
if o.useDn {
Jmat = la.MatAlloc(o.neq, o.neq)
err = o.JfcnDn(Jmat, x)
if err != nil {
return 0, chk.Err(_nls_err5, "dense", err.Error())
}
} else {
if o.numJ {
err = Jacobian(&o.Jtri, o.Ffcn, x, o.fx, o.w, false)
if err != nil {
return 0, chk.Err(_nls_err5, "sparse", err.Error())
}
} else {
err = o.JfcnSp(&o.Jtri, x)
if err != nil {
return 0, chk.Err(_nls_err5, "sparse(num)", err.Error())
}
}
Jmat = o.Jtri.ToMatrix(nil).ToDense()
}
//la.PrintMat("J", Jmat, "%23g", false)
// condition number
cnd, err = la.MatCondG(Jmat, "F", 1e-10)
if err != nil {
return cnd, chk.Err(_nls_err6, err.Error())
}
if math.IsInf(cnd, 0) || math.IsNaN(cnd) {
return cnd, chk.Err(_nls_err7, cnd)
}
// numerical Jacobian
if !chkJnum {
return
}
var Jtmp la.Triplet
ws := make([]float64, o.neq)
err = o.Ffcn(o.fx, x)
if err != nil {
return
}
Jtmp.Init(o.neq, o.neq, o.neq*o.neq)
Jacobian(&Jtmp, o.Ffcn, x, o.fx, ws, false)
Jnum := Jtmp.ToMatrix(nil).ToDense()
for i := 0; i < o.neq; i++ {
for j := 0; j < o.neq; j++ {
chk.PrintAnaNum(io.Sf("J[%d][%d]", i, j), tol, Jmat[i][j], Jnum[i][j], !silent)
}
}
maxdiff := la.MatMaxDiff(Jmat, Jnum)
if maxdiff > tol {
err = chk.Err(_nls_err8, maxdiff)
}
return
}
// AddToKb adds element K to global Jacobian matrix Kb
func (o *ElastRod) AddToKb(Kb *la.Triplet, sol *Solution, firstIt bool) (err error) {
for i, I := range o.Umap {
for j, J := range o.Umap {
Kb.Put(I, J, o.K[i][j])
}
}
return
}
// AddToKb adds element K to global Jacobian matrix Kb
func (o *Rod) AddToKb(Kb *la.Triplet, sol *Solution, firstIt bool) (err error) {
// zero K matrix
la.MatFill(o.K, 0)
la.MatFill(o.M, 0) // TODO: implement mass matrix
// for each integration point
var E float64
nverts := o.Cell.Shp.Nverts
for idx, ip := range o.IpsElem {
// interpolation functions, gradients and variables @ ip
err = o.ipvars(idx, sol)
if err != nil {
return
}
// auxiliary
coef := ip[3]
Jvec := o.Cell.Shp.Jvec3d
G := o.Cell.Shp.Gvec
J := o.Cell.Shp.J
// add contribution to consistent tangent matrix
for m := 0; m < nverts; m++ {
for n := 0; n < nverts; n++ {
for i := 0; i < o.Ndim; i++ {
for j := 0; j < o.Ndim; j++ {
r := i + m*o.Ndim
c := j + n*o.Ndim
E, err = o.Model.CalcD(o.States[idx], firstIt)
if err != nil {
return
}
o.K[r][c] += coef * o.A * E * G[m] * G[n] * Jvec[i] * Jvec[j] / J
}
}
}
}
}
// add K to sparse matrix Kb
for i, I := range o.Umap {
for j, J := range o.Umap {
Kb.Put(I, J, o.K[i][j])
}
}
return
}
// adds element K to global Jacobian matrix Kb
func (o Rod) AddToKb(Kb *la.Triplet, sol *Solution, firstIt bool) (ok bool) {
// zero K matrix
la.MatFill(o.K, 0)
la.MatFill(o.M, 0)
// for each integration point
nverts := o.Shp.Nverts
ndim := Global.Ndim
for idx, ip := range o.IpsElem {
// interpolation functions, gradients and variables @ ip
if !o.ipvars(idx, sol) {
return
}
// auxiliary
coef := ip.W
Jvec := o.Shp.Jvec3d
G := o.Shp.Gvec
J := o.Shp.J
// add contribution to consistent tangent matrix
for m := 0; m < nverts; m++ {
for n := 0; n < nverts; n++ {
for i := 0; i < ndim; i++ {
for j := 0; j < ndim; j++ {
r := i + m*ndim
c := j + n*ndim
E, err := o.Model.CalcD(o.States[idx], firstIt)
if LogErr(err, "AddToKb") {
return
}
o.K[r][c] += coef * o.A * E * G[m] * G[n] * Jvec[i] * Jvec[j] / J
}
}
}
}
}
// add K to sparse matrix Kb
for i, I := range o.Umap {
for j, J := range o.Umap {
Kb.Put(I, J, o.K[i][j])
}
}
return true
}
// adds element K to global Jacobian matrix Kb
func (o Beam) AddToKb(Kb *la.Triplet, sol *Solution, firstIt bool) (ok bool) {
if Global.Sim.Data.Steady {
for i, I := range o.Umap {
for j, J := range o.Umap {
Kb.Put(I, J, o.K[i][j])
}
}
} else {
dc := Global.DynCoefs
for i, I := range o.Umap {
for j, J := range o.Umap {
Kb.Put(I, J, o.M[i][j]*dc.α1+o.K[i][j])
}
}
}
return true
}
// adds element K to global Jacobian matrix Kb
func (o *Beam) AddToKb(Kb *la.Triplet, sol *Solution, firstIt bool) (err error) {
if sol.Steady {
for i, I := range o.Umap {
for j, J := range o.Umap {
Kb.Put(I, J, o.K[i][j])
}
}
return
}
α1 := sol.DynCfs.α1
for i, I := range o.Umap {
for j, J := range o.Umap {
Kb.Put(I, J, o.M[i][j]*α1+o.K[i][j])
}
}
return
}
func main() {
mpi.Start(false)
defer func() {
mpi.Stop(false)
}()
if mpi.Rank() == 0 {
chk.PrintTitle("Test SumToRoot 01")
}
M := [][]float64{
{1000, 1000, 1000, 1011, 1021, 1000},
{1000, 1000, 1000, 1012, 1022, 1000},
{1000, 1000, 1000, 1013, 1023, 1000},
{1011, 1012, 1013, 1000, 1000, 1000},
{1021, 1022, 1023, 1000, 1000, 1000},
{1000, 1000, 1000, 1000, 1000, 1000},
}
id, sz, m := mpi.Rank(), mpi.Size(), len(M)
start, endp1 := (id*m)/sz, ((id+1)*m)/sz
if sz > 6 {
chk.Panic("this test works with at most 6 processors")
}
var J la.Triplet
J.Init(m, m, m*m)
for i := start; i < endp1; i++ {
for j := 0; j < m; j++ {
J.Put(i, j, M[i][j])
}
}
la.PrintMat(fmt.Sprintf("J @ proc # %d", id), J.ToMatrix(nil).ToDense(), "%10.1f", false)
la.SpTriSumToRoot(&J)
var tst testing.T
if mpi.Rank() == 0 {
chk.Matrix(&tst, "J @ proc 0", 1.0e-17, J.ToMatrix(nil).ToDense(), [][]float64{
{1000, 1000, 1000, 1011, 1021, 1000},
{1000, 1000, 1000, 1012, 1022, 1000},
{1000, 1000, 1000, 1013, 1023, 1000},
{1011, 1012, 1013, 1000, 1000, 1000},
{1021, 1022, 1023, 1000, 1000, 1000},
{1000, 1000, 1000, 1000, 1000, 1000},
})
}
}
func main() {
mpi.Start(false)
defer func() {
mpi.Stop(false)
}()
myrank := mpi.Rank()
if myrank == 0 {
chk.PrintTitle("Test MUMPS Sol 02")
}
ndim := 10
id, sz := mpi.Rank(), mpi.Size()
start, endp1 := (id*ndim)/sz, ((id+1)*ndim)/sz
if mpi.Size() > ndim {
chk.Panic("the number of processors must be smaller than or equal to %d", ndim)
}
b := make([]float64, ndim)
var t la.Triplet
t.Init(ndim, ndim, ndim*ndim)
for i := start; i < endp1; i++ {
j := i
if i > 0 {
j = i - 1
}
for ; j < ndim; j++ {
val := 10.0 - float64(j)
if i > j {
val -= 1.0
}
t.Put(i, j, val)
}
b[i] = float64(i + 1)
}
x_correct := []float64{-1, 8, -65, 454, -2725, 13624, -54497, 163490, -326981, 326991}
sum_b_to_root := true
la.RunMumpsTestR(&t, 1e-4, b, x_correct, sum_b_to_root)
}
// AddToKb adds element K to global Jacobian matrix Kb
func (o *ElemPhi) AddToKb(Kb *la.Triplet, sol *Solution, firstIt bool) (err error) {
// auxiliary
nverts := o.Cell.Shp.Nverts
// zero K matrix
la.MatFill(o.K, 0)
dt := sol.Dt * 2
// for each integration point
for iip, ip := range o.IpsElem {
// interpolation functions and gradients
err = o.Cell.Shp.CalcAtIp(o.X, ip, true)
if err != nil {
return
}
// auxiliary variables
coef := o.Cell.Shp.J * ip[3]
S := o.Cell.Shp.S
G := o.Cell.Shp.G
// add to right hand side vector
for m := 0; m < nverts; m++ {
for n := 0; n < nverts; n++ {
o.K[m][n] -= coef * (S[m]*S[n] + dt*dt/6.0*(o.v_0[iip][0]*G[m][0]+o.v_0[iip][1]*G[m][1])*(o.v_0[iip][0]*G[n][0]+o.v_0[iip][1]*G[n][1])) / dt
}
}
}
// add K to sparse matrix Kb
for i, I := range o.Umap {
for j, J := range o.Umap {
Kb.Put(I, J, o.K[i][j])
}
}
return
}
// AddToKb adds element K to global Jacobian matrix Kb
func (o *ElemPhi) AddToKb(Kb *la.Triplet, sol *Solution, firstIt bool) (ok bool) {
// auxiliary
β1 := Global.DynCoefs.β1
nverts := o.Shp.Nverts
// zero K matrix
la.MatFill(o.K, 0)
// for each integration point
for _, ip := range o.IpsElem {
// interpolation functions and gradients
if LogErr(o.Shp.CalcAtIp(o.X, ip, true), "InterpStarVars") {
return
}
// auxiliary variables
coef := o.Shp.J * ip.W
S := o.Shp.S
// add to right hand side vector
for m := 0; m < nverts; m++ {
for n := 0; n < nverts; n++ {
o.K[m][n] += coef * S[m] * S[n] * β1
}
}
}
// add K to sparse matrix Kb
for i, I := range o.Umap {
for j, J := range o.Umap {
Kb.Put(I, J, o.K[i][j])
}
}
return true
}
func Assemble(K11, K12 *la.Triplet, F1 []float64, src Cb_src, g *Grid2D, e *Equations) {
K11.Start()
K12.Start()
la.VecFill(F1, 0.0)
kx, ky := 1.0, 1.0
alp, bet, gam := 2.0*(kx/g.Dxx+ky/g.Dyy), -kx/g.Dxx, -ky/g.Dyy
mol := []float64{alp, bet, bet, gam, gam}
for i, I := range e.RF1 {
col, row := I%g.Nx, I/g.Nx
nodes := []int{I, I - 1, I + 1, I - g.Nx, I + g.Nx} // I, left, right, bottom, top
if col == 0 {
nodes[1] = nodes[2]
}
if col == g.Nx-1 {
nodes[2] = nodes[1]
}
if row == 0 {
nodes[3] = nodes[4]
}
if row == g.Ny-1 {
nodes[4] = nodes[3]
}
for k, J := range nodes {
j1, j2 := e.FR1[J], e.FR2[J] // 1 or 2?
if j1 > -1 { // 11
K11.Put(i, j1, mol[k])
} else { // 12
K12.Put(i, j2, mol[k])
}
}
if src != nil {
x := float64(col) * g.Dx
y := float64(row) * g.Dy
F1[i] += src(x, y)
}
}
}
// Jacobian computes Jacobian (sparse) matrix
// Calculates (with N=n-1):
// df0dx0, df0dx1, df0dx2, ... df0dxN
// df1dx0, df1dx1, df1dx2, ... df1dxN
// . . . . . . . . . . . . .
// dfNdx0, dfNdx1, dfNdx2, ... dfNdxN
// INPUT:
// ffcn : f(x) function
// x : station where dfdx has to be calculated
// fx : f @ x
// w : workspace with size == n == len(x)
// RETURNS:
// J : dfdx @ x [must be pre-allocated]
func Jacobian(J *la.Triplet, ffcn Cb_f, x, fx, w []float64) (err error) {
ndim := len(x)
start, endp1 := 0, ndim
if J.Max() == 0 {
J.Init(ndim, ndim, ndim*ndim)
}
J.Start()
var df float64
for col := 0; col < ndim; col++ {
xsafe := x[col]
delta := math.Sqrt(EPS * max(CTE1, math.Abs(xsafe)))
x[col] = xsafe + delta
err = ffcn(w, x) // w := f(x+δx[col])
if err != nil {
return
}
for row := start; row < endp1; row++ {
df = w[row] - fx[row]
J.Put(row, col, df/delta)
}
x[col] = xsafe
}
return
}
func Test_assemb01(tst *testing.T) {
chk.PrintTitle("assemb01")
// grid
var g Grid2D
g.Init(1.0, 1.0, 3, 3)
// equations numbering
var e Equations
e.Init(g.N, []int{0, 3, 6})
// K11 and K12
var K11, K12 la.Triplet
InitK11andK12(&K11, &K12, &e)
// assembly
F1 := make([]float64, e.N1)
Assemble(&K11, &K12, F1, nil, &g, &e)
// check
K11d := K11.ToMatrix(nil).ToDense()
K12d := K12.ToMatrix(nil).ToDense()
K11c := [][]float64{
{16.0, -4.0, -8.0, 0.0, 0.0, 0.0},
{-8.0, 16.0, 0.0, -8.0, 0.0, 0.0},
{-4.0, 0.0, 16.0, -4.0, -4.0, 0.0},
{0.0, -4.0, -8.0, 16.0, 0.0, -4.0},
{0.0, 0.0, -8.0, 0.0, 16.0, -4.0},
{0.0, 0.0, 0.0, -8.0, -8.0, 16.0},
}
K12c := [][]float64{
{-4.0, 0.0, 0.0},
{0.0, 0.0, 0.0},
{0.0, -4.0, 0.0},
{0.0, 0.0, 0.0},
{0.0, 0.0, -4.0},
{0.0, 0.0, 0.0},
}
chk.Matrix(tst, "K11: ", 1e-16, K11d, K11c)
chk.Matrix(tst, "K12: ", 1e-16, K12d, K12c)
}
func Test_linipm01(tst *testing.T) {
//verbose()
chk.PrintTitle("linipm01")
// linear programming problem
// min -4*x0 - 5*x1
// s.t. 2*x0 + x1 ≤ 3
// x0 + 2*x1 ≤ 3
// x0,x1 ≥ 0
// standard:
// 2*x0 + x1 + x2 = 3
// x0 + 2*x1 + x3 = 3
// x0,x1,x2,x3 ≥ 0
var T la.Triplet
T.Init(2, 4, 6)
T.Put(0, 0, 2.0)
T.Put(0, 1, 1.0)
T.Put(0, 2, 1.0)
T.Put(1, 0, 1.0)
T.Put(1, 1, 2.0)
T.Put(1, 3, 1.0)
Am := T.ToMatrix(nil)
A := Am.ToDense()
c := []float64{-4, -5, 0, 0}
b := []float64{3, 3}
// print LP
la.PrintMat("A", A, "%6g", false)
la.PrintVec("b", b, "%6g", false)
la.PrintVec("c", c, "%6g", false)
io.Pf("\n")
// solve LP
var ipm LinIpm
defer ipm.Clean()
ipm.Init(Am, b, c, nil)
err := ipm.Solve(chk.Verbose)
if err != nil {
tst.Errorf("ipm failed:\n%v", err)
return
}
// check
io.Pf("\n")
io.Pforan("x = %v\n", ipm.X)
io.Pfcyan("λ = %v\n", ipm.L)
io.Pforan("s = %v\n", ipm.S)
x := ipm.X[:2]
bres := make([]float64, 2)
la.MatVecMul(bres, 1, A, x)
io.Pforan("bres = %v\n", bres)
chk.Vector(tst, "x", 1e-9, x, []float64{1, 1})
chk.Vector(tst, "A*x=b", 1e-8, bres, b)
// plot
if true && chk.Verbose {
f := func(x []float64) float64 {
return c[0]*x[0] + c[1]*x[1]
}
g := func(x []float64, i int) float64 {
return A[i][0]*x[0] + A[i][1]*x[1] - b[i]
}
np := 41
vmin, vmax := []float64{-2.0, -2.0}, []float64{2.0, 2.0}
PlotTwoVarsContour("/tmp/gosl", "test_linipm01", x, np, nil, true, vmin, vmax, f,
func(x []float64) float64 { return g(x, 0) },
func(x []float64) float64 { return g(x, 1) },
)
}
}
func InitK11andK12(K11, K12 *la.Triplet, e *Equations) {
K11.Init(e.N1, e.N1, e.N1*5)
K12.Init(e.N1, e.N2, e.N1*5)
}
// AddToKb adds element K to global Jacobian matrix Kb
func (o *ElemU) AddToKb(Kb *la.Triplet, sol *Solution, firstIt bool) (ok bool) {
// zero K matrix
la.MatFill(o.K, 0)
// for each integration point
dc := Global.DynCoefs
ndim := Global.Ndim
nverts := o.Shp.Nverts
for idx, ip := range o.IpsElem {
// interpolation functions, gradients and variables @ ip
if !o.ipvars(idx, sol) {
return
}
// check Jacobian
if o.Shp.J < 0 {
LogErrCond(true, "ElemU: eid=%d: Jacobian is negative = %g\n", o.Id(), o.Shp.J)
return
}
// auxiliary
coef := o.Shp.J * ip.W * o.Thickness
S := o.Shp.S
G := o.Shp.G
// consistent tangent model matrix
if LogErr(o.MdlSmall.CalcD(o.D, o.States[idx], firstIt), "AddToKb") {
return
}
// add contribution to consistent tangent matrix
if o.UseB {
radius := 1.0
if Global.Sim.Data.Axisym {
radius = o.Shp.AxisymGetRadius(o.X)
coef *= radius
}
IpBmatrix(o.B, ndim, nverts, G, radius, S)
la.MatTrMulAdd3(o.K, coef, o.B, o.D, o.B) // K += coef * tr(B) * D * B
} else {
IpAddToKt(o.K, nverts, ndim, coef, G, o.D)
}
// dynamic term
if !Global.Sim.Data.Steady {
for m := 0; m < nverts; m++ {
for i := 0; i < ndim; i++ {
r := i + m*ndim
for n := 0; n < nverts; n++ {
c := i + n*ndim
o.K[r][c] += coef * S[m] * S[n] * (o.Rho*dc.α1 + o.Cdam*dc.α4)
}
}
}
}
}
// add K to sparse matrix Kb
for i, I := range o.Umap {
for j, J := range o.Umap {
Kb.Put(I, J, o.K[i][j])
}
}
return true
}
func TestDiffusion1D(tst *testing.T) {
//verbose()
chk.PrintTitle("Test Diffusion 1D (cooling)")
// solution parameters
silent := false
fixstp := true
//fixstp := false
//method := "FwEuler"
method := "BwEuler"
//method := "Dopri5"
//method := "Radau5"
//numjac := true
numjac := false
rtol := 1e-4
atol := rtol
// timestep
t0, tf, dt := 0.0, 0.2, 0.03
// problem data
kx := 1.0 // conductivity
N := 6 // number of nodes
//Nb := N + 2 // augmented system dimension
xmax := 1.0 // length
dx := xmax / float64(N-1) // spatial step size
dxx := dx * dx
mol := []float64{kx / dxx, -2.0 * kx / dxx, kx / dxx}
// function
fcn := func(f []float64, t float64, y []float64, args ...interface{}) error {
for i := 0; i < N; i++ {
f[i] = 0
if i == 0 || i == N-1 {
continue // skip presc node
}
for p, j := range []int{i - 1, i, i + 1} {
if j < 0 {
j = i + 1
} // left boundary
if j == N {
j = i - 1
} // right boundary
f[i] += mol[p] * y[j]
}
}
//io.Pfgrey("y = %v\n", y)
//io.Pfcyan("f = %v\n", f)
return nil
}
// Jacobian
jac := func(dfdy *la.Triplet, t float64, y []float64, args ...interface{}) error {
//chk.Panic("jac is not available")
if dfdy.Max() == 0 {
//dfdy.Init(Nb, Nb, 3*N)
dfdy.Init(N, N, 3*N)
}
dfdy.Start()
for i := 0; i < N; i++ {
if i == 0 || i == N-1 {
dfdy.Put(i, i, 0.0)
continue
}
for p, j := range []int{i - 1, i, i + 1} {
if j < 0 {
j = i + 1
} // left boundary
if j == N {
j = i - 1
} // right boundary
dfdy.Put(i, j, mol[p])
}
}
return nil
}
// initial values
x := utl.LinSpace(0.0, xmax, N)
y := make([]float64, N)
//y := make([]float64, Nb)
for i := 0; i < N; i++ {
y[i] = 4.0*x[i] - 4.0*x[i]*x[i]
}
// debug
f0 := make([]float64, N)
//f0 := make([]float64, Nb)
fcn(f0, 0, y)
if false {
io.Pforan("y0 = %v\n", y)
io.Pforan("f0 = %v\n", f0)
var J la.Triplet
jac(&J, 0, y)
la.PrintMat("J", J.ToMatrix(nil).ToDense(), "%8.3f", false)
}
//chk.Panic("stop")
/*
// constraints
var A la.Triplet
A.Init(2, N, 2)
A.Put(0, 0, 1.0)
A.Put(1, N-1, 1.0)
io.Pfcyan("A = %+v\n", A)
Am := A.ToMatrix(nil)
c := make([]float64, 2)
la.SpMatVecMul(c, 1, Am, y) // c := Am*y
la.PrintMat("A", Am.ToDense(), "%3g", false)
io.Pfcyan("c = %v ([0, 0] => consistent)\n", c)
*/
/*
// mass matrix
var M la.Triplet
M.Init(Nb, Nb, N + 4)
for i := 0; i < N; i++ {
M.Put(i, i, 1.0)
}
M.PutMatAndMatT(&A)
Mm := M.ToMatrix(nil)
la.PrintMat("M", Mm.ToDense(), "%3g", false)
*/
// output
var b0, b1, b2 bytes.Buffer
fmt.Fprintf(&b0, "from gosl import *\n")
fmt.Fprintf(&b1, "T = array([")
fmt.Fprintf(&b2, "U = array([")
out := func(first bool, dt, t float64, y []float64, args ...interface{}) error {
fmt.Fprintf(&b1, "%23.15E,", t)
fmt.Fprintf(&b2, "[")
for i := 0; i < N; i++ {
fmt.Fprintf(&b2, "%23.15E,", y[i])
}
fmt.Fprintf(&b2, "],")
return nil
}
defer func() {
fmt.Fprintf(&b1, "])\n")
fmt.Fprintf(&b2, "])\n")
fmt.Fprintf(&b2, "X = linspace(0.0, %g, %d)\n", xmax, N)
fmt.Fprintf(&b2, "tt, xx = meshgrid(T, X)\n")
fmt.Fprintf(&b2, "ax = PlotSurf(tt, xx, vstack(transpose(U)), 't', 'x', 'u', 0.0, 1.0)\n")
fmt.Fprintf(&b2, "ax.view_init(20.0, 30.0)\n")
fmt.Fprintf(&b2, "show()\n")
io.WriteFileD("/tmp/gosl", "plot_diffusion_1d.py", &b0, &b1, &b2)
}()
// ode solver
var Jfcn Cb_jac
var osol ODE
if !numjac {
Jfcn = jac
}
osol.Init(method, N, fcn, Jfcn, nil, out, silent)
//osol.Init(method, Nb, fcn, Jfcn, &M, out, silent)
osol.SetTol(atol, rtol)
// constant Jacobian
if method == "BwEuler" {
osol.CteTg = true
osol.Verbose = true
}
// run
wallt0 := time.Now()
if !fixstp {
dt = tf - t0
}
osol.Solve(y, t0, tf, dt, fixstp)
io.Pfmag("elapsed time = %v\n", time.Now().Sub(wallt0))
}
// adds element K to global Jacobian matrix Kb
func (o ElemUP) AddToKb(Kb *la.Triplet, sol *Solution, firstIt bool) (ok bool) {
// clear matrices
ndim := Global.Ndim
u_nverts := o.U.Shp.Nverts
p_nverts := o.P.Shp.Nverts
la.MatFill(o.P.Kpp, 0)
for i := 0; i < o.U.Nu; i++ {
for j := 0; j < o.P.Np; j++ {
o.Kup[i][j] = 0
o.Kpu[j][i] = 0
}
for j := 0; j < o.U.Nu; j++ {
o.U.K[i][j] = 0
}
}
if o.P.DoExtrap {
for i := 0; i < p_nverts; i++ {
o.P.ρl_ex[i] = 0
for j := 0; j < p_nverts; j++ {
o.P.dρldpl_ex[i][j] = 0
}
for j := 0; j < o.U.Nu; j++ {
o.dρldus_ex[i][j] = 0
}
}
}
// for each integration point
dc := Global.DynCoefs
var coef, plt, klr, ρL, Cl, divvs float64
var ρl, ρ, Cpl, Cvs, dρdpl, dpdpl, dCpldpl, dCvsdpl, dklrdpl, dCpldusM, dρldusM, dρdusM float64
var r, c int
for idx, ip := range o.U.IpsElem {
// interpolation functions, gradients and variables @ ip
if !o.ipvars(idx, sol) {
return
}
coef = o.U.Shp.J * ip.W
S := o.U.Shp.S
G := o.U.Shp.G
Sb := o.P.Shp.S
Gb := o.P.Shp.G
// axisymmetric case
radius := 1.0
if Global.Sim.Data.Axisym {
radius = o.U.Shp.AxisymGetRadius(o.U.X)
coef *= radius
}
// auxiliary
divvs = dc.α4*o.divus - o.U.divχs[idx] // divergence of Eq (35a) [1]
// tpm variables
plt = dc.β1*o.P.pl - o.P.ψl[idx] // Eq (35c) [1]
klr = o.P.Mdl.Cnd.Klr(o.P.States[idx].A_sl)
ρL = o.P.States[idx].A_ρL
Cl = o.P.Mdl.Cl
if LogErr(o.P.Mdl.CalcLs(o.P.res, o.P.States[idx], o.P.pl, o.divus, true), "AddToKb") {
return
}
ρl = o.P.res.A_ρl
ρ = o.P.res.A_ρ
Cpl = o.P.res.Cpl
Cvs = o.P.res.Cvs
dρdpl = o.P.res.Dρdpl
dpdpl = o.P.res.Dpdpl
dCpldpl = o.P.res.DCpldpl
dCvsdpl = o.P.res.DCvsdpl
dklrdpl = o.P.res.Dklrdpl
dCpldusM = o.P.res.DCpldusM
dρldusM = o.P.res.DρldusM
dρdusM = o.P.res.DρdusM
// Kpu, Kup and Kpp
for n := 0; n < p_nverts; n++ {
for j := 0; j < ndim; j++ {
// Kpu := ∂Rl^n/∂us^m and Kup := ∂Rus^m/∂pl^n; see Eq (47) of [1]
for m := 0; m < u_nverts; m++ {
c = j + m*ndim
// add ∂rlb/∂us^m: Eqs (A.3) and (A.6) of [1]
o.Kpu[n][c] += coef * Sb[n] * (dCpldusM*plt + dc.α4*Cvs) * G[m][j]
// add ∂(ρl.wl)/∂us^m: Eq (A.8) of [1]
for i := 0; i < ndim; i++ {
o.Kpu[n][c] += coef * Gb[n][i] * S[m] * dc.α1 * ρL * klr * o.P.Mdl.Klsat[i][j]
}
// add ∂rl/∂pl^n and ∂p/∂pl^n: Eqs (A.9) and (A.11) of [1]
o.Kup[c][n] += coef * (S[m]*Sb[n]*dρdpl*o.bs[j] - G[m][j]*Sb[n]*dpdpl)
// for seepage face
if o.P.DoExtrap {
o.dρldus_ex[n][c] += o.P.Emat[n][idx] * dρldusM * G[m][j]
}
}
// term in brackets in Eq (A.7) of [1]
o.P.tmp[j] = Sb[n]*dklrdpl*o.hl[j] - klr*(Sb[n]*Cl*o.bs[j]+Gb[n][j])
}
// Kpp := ∂Rl^m/∂pl^n; see Eq (47) of [1]
for m := 0; m < p_nverts; m++ {
// add ∂rlb/dpl^n: Eq (A.5) of [1]
o.P.Kpp[m][n] += coef * Sb[m] * Sb[n] * (dCpldpl*plt + dCvsdpl*divvs + dc.β1*Cpl)
// add ∂(ρl.wl)/∂us^m: Eq (A.7) of [1]
for i := 0; i < ndim; i++ {
for j := 0; j < ndim; j++ {
o.P.Kpp[m][n] -= coef * Gb[m][i] * o.P.Mdl.Klsat[i][j] * o.P.tmp[j]
}
}
// inner summation term in Eq (22) of [2]
if o.P.DoExtrap {
o.P.dρldpl_ex[m][n] += o.P.Emat[m][idx] * Cpl * Sb[n]
}
}
// Eq. (19) of [2]
if o.P.DoExtrap {
o.P.ρl_ex[n] += o.P.Emat[n][idx] * ρl
}
}
// Kuu: add ∂rub^m/∂us^n; see Eqs (47) and (A.10) of [1]
for m := 0; m < u_nverts; m++ {
for i := 0; i < ndim; i++ {
r = i + m*ndim
for n := 0; n < u_nverts; n++ {
for j := 0; j < ndim; j++ {
c = j + n*ndim
o.U.K[r][c] += coef * S[m] * (S[n]*dc.α1*ρ*tsr.It[i][j] + dρdusM*o.bs[i]*G[n][j])
}
}
}
}
// consistent tangent model matrix
if LogErr(o.U.MdlSmall.CalcD(o.U.D, o.U.States[idx], firstIt), "AddToKb") {
return
}
// Kuu: add stiffness term ∂(σe・G^m)/∂us^n
if o.U.UseB {
IpBmatrix(o.U.B, ndim, u_nverts, G, radius, S)
la.MatTrMulAdd3(o.U.K, coef, o.U.B, o.U.D, o.U.B) // K += coef * tr(B) * D * B
} else {
IpAddToKt(o.U.K, u_nverts, ndim, coef, G, o.U.D)
}
}
// contribution from natural boundary conditions
if o.P.HasSeep {
if !o.P.add_natbcs_to_jac(sol) {
return
}
if !o.add_natbcs_to_jac(sol) {
return
}
}
// add K to sparse matrix Kb
// _ _
// | Kuu Kup 0 |
// | Kpu Kpp Kpf |
// |_ Kfu Kfp Kff _|
//
for i, I := range o.P.Pmap {
for j, J := range o.P.Pmap {
Kb.Put(I, J, o.P.Kpp[i][j])
}
for j, J := range o.P.Fmap {
Kb.Put(I, J, o.P.Kpf[i][j])
Kb.Put(J, I, o.P.Kfp[j][i])
}
for j, J := range o.U.Umap {
Kb.Put(I, J, o.Kpu[i][j])
Kb.Put(J, I, o.Kup[j][i])
}
}
for i, I := range o.P.Fmap {
for j, J := range o.P.Fmap {
Kb.Put(I, J, o.P.Kff[i][j])
}
}
for i, I := range o.U.Umap {
for j, J := range o.U.Umap {
Kb.Put(I, J, o.U.K[i][j])
}
}
return true
}
func IpBmatrix_sparse(B *la.Triplet, ndim, nne int, G [][]float64, radius float64, S []float64, axisym bool) {
B.Start()
if ndim == 3 {
for i := 0; i < nne; i++ {
B.Put(0, 0+i*3, G[i][0])
B.Put(1, 1+i*3, G[i][1])
B.Put(2, 2+i*3, G[i][2])
B.Put(3, 0+i*3, G[i][1]/SQ2)
B.Put(4, 1+i*3, G[i][2]/SQ2)
B.Put(5, 2+i*3, G[i][0]/SQ2)
B.Put(3, 1+i*3, G[i][0]/SQ2)
B.Put(4, 2+i*3, G[i][1]/SQ2)
B.Put(5, 0+i*3, G[i][2]/SQ2)
}
return
}
if axisym {
for i := 0; i < nne; i++ {
B.Put(0, 0+i*2, G[i][0])
B.Put(1, 1+i*2, G[i][1])
B.Put(2, 0+i*2, S[i]/radius)
B.Put(3, 0+i*2, G[i][1]/SQ2)
B.Put(3, 1+i*2, G[i][0]/SQ2)
}
return
}
for i := 0; i < nne; i++ {
B.Put(0, 0+i*2, G[i][0])
B.Put(1, 1+i*2, G[i][1])
B.Put(3, 0+i*2, G[i][1]/SQ2)
B.Put(3, 1+i*2, G[i][0]/SQ2)
}
}
func main() {
// input matrix in Triplet format
// including repeated positions. e.g. (0,0)
var A la.Triplet
A.Init(5, 5, 13)
A.Put(0, 0, 1.0) // << repeated
A.Put(0, 0, 1.0) // << repeated
A.Put(1, 0, 3.0)
A.Put(0, 1, 3.0)
A.Put(2, 1, -1.0)
A.Put(4, 1, 4.0)
A.Put(1, 2, 4.0)
A.Put(2, 2, -3.0)
A.Put(3, 2, 1.0)
A.Put(4, 2, 2.0)
A.Put(2, 3, 2.0)
A.Put(1, 4, 6.0)
A.Put(4, 4, 1.0)
// right-hand-side
b := []float64{8.0, 45.0, -3.0, 3.0, 19.0}
// solve
x, err := la.SolveRealLinSys(&A, b)
if err != nil {
io.Pfred("solver failed:\n%v", err)
return
}
// output
la.PrintMat("a", A.ToMatrix(nil).ToDense(), "%5g", false)
la.PrintVec("b", b, "%v ", false)
la.PrintVec("x", x, "%v ", false)
}
func Test_nls04(tst *testing.T) {
//verbose()
chk.PrintTitle("nls04. finite differences problem")
// grid
var g fdm.Grid2D
g.Init(1.0, 1.0, 6, 6)
// equations numbering
var e fdm.Equations
peq := utl.IntUnique(g.L, g.R, g.B, g.T)
e.Init(g.N, peq)
// K11 and K12
var K11, K12 la.Triplet
fdm.InitK11andK12(&K11, &K12, &e)
// assembly
F1 := make([]float64, e.N1)
fdm.Assemble(&K11, &K12, F1, nil, &g, &e)
// prescribed values
U2 := make([]float64, e.N2)
for _, eq := range g.L {
U2[e.FR2[eq]] = 50.0
}
for _, eq := range g.R {
U2[e.FR2[eq]] = 0.0
}
for _, eq := range g.B {
U2[e.FR2[eq]] = 0.0
}
for _, eq := range g.T {
U2[e.FR2[eq]] = 50.0
}
// functions
k11 := K11.ToMatrix(nil)
k12 := K12.ToMatrix(nil)
ffcn := func(fU1, U1 []float64) error { // K11*U1 + K12*U2 - F1
la.VecCopy(fU1, -1, F1) // fU1 := (-F1)
la.SpMatVecMulAdd(fU1, 1, k11, U1) // fU1 += K11*U1
la.SpMatVecMulAdd(fU1, 1, k12, U2) // fU1 += K12*U2
return nil
}
Jfcn := func(dfU1dU1 *la.Triplet, U1 []float64) error {
fdm.Assemble(dfU1dU1, &K12, F1, nil, &g, &e)
return nil
}
JfcnD := func(dfU1dU1 [][]float64, U1 []float64) error {
la.MatCopy(dfU1dU1, 1, K11.ToMatrix(nil).ToDense())
return nil
}
prms := map[string]float64{
"atol": 1e-8,
"rtol": 1e-8,
"ftol": 1e-12,
"lSearch": 0.0,
}
// init
var nls_sps NlSolver // sparse analytical
var nls_num NlSolver // sparse numerical
var nls_den NlSolver // dense analytical
nls_sps.Init(e.N1, ffcn, Jfcn, nil, false, false, prms)
nls_num.Init(e.N1, ffcn, nil, nil, false, true, prms)
nls_den.Init(e.N1, ffcn, nil, JfcnD, true, false, prms)
defer nls_sps.Clean()
defer nls_num.Clean()
defer nls_den.Clean()
// results
U1sps := make([]float64, e.N1)
U1num := make([]float64, e.N1)
U1den := make([]float64, e.N1)
Usps := make([]float64, e.N)
Unum := make([]float64, e.N)
Uden := make([]float64, e.N)
// solution
Uc := []float64{0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 50.0, 25.0, 325.0 / 22.0, 100.0 / 11.0, 50.0 / 11.0,
0.0, 50.0, 775.0 / 22.0, 25.0, 375.0 / 22.0, 100.0 / 11.0, 0.0, 50.0, 450.0 / 11.0, 725.0 / 22.0,
25.0, 325.0 / 22.0, 0.0, 50.0, 500.0 / 11.0, 450.0 / 11.0, 775.0 / 22.0, 25.0, 0.0, 50.0, 50.0,
50.0, 50.0, 50.0, 50.0,
}
io.PfYel("\n---- sparse -------- Analytical Jacobian -------------------\n")
// solve
err := nls_sps.Solve(U1sps, false)
if err != nil {
chk.Panic(err.Error())
}
// check
fdm.JoinVecs(Usps, U1sps, U2, &e)
chk.Vector(tst, "Usps", 1e-14, Usps, Uc)
// plot
if false {
g.Contour("results", "fig_t_heat_square", nil, Usps, 11, false)
}
io.PfYel("\n---- dense -------- Analytical Jacobian -------------------\n")
// solve
err = nls_den.Solve(U1den, false)
if err != nil {
chk.Panic(err.Error())
}
// check
fdm.JoinVecs(Uden, U1den, U2, &e)
chk.Vector(tst, "Uden", 1e-14, Uden, Uc)
io.PfYel("\n---- sparse -------- Numerical Jacobian -------------------\n")
// solve
err = nls_num.Solve(U1num, false)
if err != nil {
chk.Panic(err.Error())
}
// check
fdm.JoinVecs(Unum, U1num, U2, &e)
chk.Vector(tst, "Unum", 1e-14, Unum, Uc)
}
func main() {
// input matrix in Triplet format
// including repeated positions. e.g. (0,0)
var A la.Triplet
A.Init(5, 5, 13)
A.Put(0, 0, 1.0) // << repeated
A.Put(0, 0, 1.0) // << repeated
A.Put(1, 0, 3.0)
A.Put(0, 1, 3.0)
A.Put(2, 1, -1.0)
A.Put(4, 1, 4.0)
A.Put(1, 2, 4.0)
A.Put(2, 2, -3.0)
A.Put(3, 2, 1.0)
A.Put(4, 2, 2.0)
A.Put(2, 3, 2.0)
A.Put(1, 4, 6.0)
A.Put(4, 4, 1.0)
// right-hand-side
b := []float64{8.0, 45.0, -3.0, 3.0, 19.0}
// allocate solver
lis := la.GetSolver("umfpack")
defer lis.Clean()
// info
symmetric := false
verbose := false
timing := false
// initialise solver (R)eal
err := lis.InitR(&A, symmetric, verbose, timing)
if err != nil {
io.Pfred("solver failed:\n%v", err)
return
}
// factorise
err = lis.Fact()
if err != nil {
io.Pfred("solver failed:\n%v", err)
return
}
// solve (R)eal
var dummy bool
x := make([]float64, len(b))
err = lis.SolveR(x, b, dummy) // x := inv(a) * b
if err != nil {
io.Pfred("solver failed:\n%v", err)
return
}
// output
la.PrintMat("a", A.ToMatrix(nil).ToDense(), "%5g", false)
la.PrintVec("b", b, "%v ", false)
la.PrintVec("x", x, "%v ", false)
}
func main() {
mpi.Start(false)
defer func() {
mpi.Stop(false)
}()
if mpi.Rank() == 0 {
chk.PrintTitle("Test ODE 04b (MPI)")
io.Pfcyan("Hairer-Wanner VII-p376 Transistor Amplifier (MPI)\n")
io.Pfcyan("(from E Hairer's website, not the system in the book)\n")
}
if mpi.Size() != 3 {
chk.Panic(">> error: this test requires 3 MPI processors\n")
return
}
// RIGHT-HAND SIDE OF THE AMPLIFIER PROBLEM
w := make([]float64, 8) // workspace
fcn := func(f []float64, x float64, y []float64, args ...interface{}) error {
d := args[0].(*HWtransData)
UET := d.UE * math.Sin(d.W*x)
FAC1 := d.BETA * (math.Exp((y[3]-y[2])/d.UF) - 1.0)
FAC2 := d.BETA * (math.Exp((y[6]-y[5])/d.UF) - 1.0)
la.VecFill(f, 0)
switch mpi.Rank() {
case 0:
f[0] = y[0] / d.R9
case 1:
f[1] = (y[1]-d.UB)/d.R8 + d.ALPHA*FAC1
f[2] = y[2]/d.R7 - FAC1
case 2:
f[3] = y[3]/d.R5 + (y[3]-d.UB)/d.R6 + (1.0-d.ALPHA)*FAC1
f[4] = (y[4]-d.UB)/d.R4 + d.ALPHA*FAC2
f[5] = y[5]/d.R3 - FAC2
f[6] = y[6]/d.R1 + (y[6]-d.UB)/d.R2 + (1.0-d.ALPHA)*FAC2
f[7] = (y[7] - UET) / d.R0
}
mpi.AllReduceSum(f, w)
return nil
}
// JACOBIAN OF THE AMPLIFIER PROBLEM
jac := func(dfdy *la.Triplet, x float64, y []float64, args ...interface{}) error {
d := args[0].(*HWtransData)
FAC14 := d.BETA * math.Exp((y[3]-y[2])/d.UF) / d.UF
FAC27 := d.BETA * math.Exp((y[6]-y[5])/d.UF) / d.UF
if dfdy.Max() == 0 {
dfdy.Init(8, 8, 16)
}
NU := 2
dfdy.Start()
switch mpi.Rank() {
case 0:
dfdy.Put(2+0-NU, 0, 1.0/d.R9)
dfdy.Put(2+1-NU, 1, 1.0/d.R8)
dfdy.Put(1+2-NU, 2, -d.ALPHA*FAC14)
dfdy.Put(0+3-NU, 3, d.ALPHA*FAC14)
dfdy.Put(2+2-NU, 2, 1.0/d.R7+FAC14)
case 1:
dfdy.Put(1+3-NU, 3, -FAC14)
dfdy.Put(2+3-NU, 3, 1.0/d.R5+1.0/d.R6+(1.0-d.ALPHA)*FAC14)
dfdy.Put(3+2-NU, 2, -(1.0-d.ALPHA)*FAC14)
dfdy.Put(2+4-NU, 4, 1.0/d.R4)
dfdy.Put(1+5-NU, 5, -d.ALPHA*FAC27)
case 2:
dfdy.Put(0+6-NU, 6, d.ALPHA*FAC27)
dfdy.Put(2+5-NU, 5, 1.0/d.R3+FAC27)
dfdy.Put(1+6-NU, 6, -FAC27)
dfdy.Put(2+6-NU, 6, 1.0/d.R1+1.0/d.R2+(1.0-d.ALPHA)*FAC27)
dfdy.Put(3+5-NU, 5, -(1.0-d.ALPHA)*FAC27)
dfdy.Put(2+7-NU, 7, 1.0/d.R0)
}
return nil
}
// MATRIX "M"
c1, c2, c3, c4, c5 := 1.0e-6, 2.0e-6, 3.0e-6, 4.0e-6, 5.0e-6
var M la.Triplet
M.Init(8, 8, 14)
M.Start()
NU := 1
switch mpi.Rank() {
case 0:
M.Put(1+0-NU, 0, -c5)
M.Put(0+1-NU, 1, c5)
M.Put(2+0-NU, 0, c5)
M.Put(1+1-NU, 1, -c5)
M.Put(1+2-NU, 2, -c4)
M.Put(1+3-NU, 3, -c3)
case 1:
M.Put(0+4-NU, 4, c3)
M.Put(2+3-NU, 3, c3)
M.Put(1+4-NU, 4, -c3)
case 2:
M.Put(1+5-NU, 5, -c2)
M.Put(1+6-NU, 6, -c1)
M.Put(0+7-NU, 7, c1)
M.Put(2+6-NU, 6, c1)
M.Put(1+7-NU, 7, -c1)
}
// WRITE FILE FUNCTION
idxstp := 1
var b bytes.Buffer
out := func(first bool, dx, x float64, y []float64, args ...interface{}) error {
if mpi.Rank() == 0 {
if first {
fmt.Fprintf(&b, "%6s%23s%23s%23s%23s%23s%23s%23s%23s%23s\n", "ns", "x", "y0", "y1", "y2", "y3", "y4", "y5", "y6", "y7")
}
fmt.Fprintf(&b, "%6d%23.15E", idxstp, x)
for j := 0; j < len(y); j++ {
fmt.Fprintf(&b, "%23.15E", y[j])
}
fmt.Fprintf(&b, "\n")
idxstp += 1
}
return nil
}
defer func() {
if mpi.Rank() == 0 {
io.WriteFileD("/tmp/gosl", "hwamplifierB.res", &b)
}
}()
// INITIAL DATA
D, xa, xb, ya := HWtransIni()
// SET ODE SOLVER
silent := false
fixstp := false
//method := "Dopri5"
method := "Radau5"
ndim := len(ya)
//numjac := true
numjac := false
var osol ode.ODE
osol.Pll = true
if numjac {
osol.Init(method, ndim, fcn, nil, &M, out, silent)
} else {
osol.Init(method, ndim, fcn, jac, &M, out, silent)
}
osol.IniH = 1.0e-6 // initial step size
// SET TOLERANCES
atol, rtol := 1e-11, 1e-5
osol.SetTol(atol, rtol)
// RUN
t0 := time.Now()
if fixstp {
osol.Solve(ya, xa, xb, 0.01, fixstp, &D)
} else {
osol.Solve(ya, xa, xb, xb-xa, fixstp, &D)
}
if mpi.Rank() == 0 {
io.Pfmag("elapsed time = %v\n", time.Now().Sub(t0))
}
}
// Jacobian computes Jacobian (sparse) matrix
// Calculates (with N=n-1):
// df0dx0, df0dx1, df0dx2, ... df0dxN
// df1dx0, df1dx1, df1dx2, ... df1dxN
// . . . . . . . . . . . . .
// dfNdx0, dfNdx1, dfNdx2, ... dfNdxN
// INPUT:
// ffcn : f(x) function
// x : station where dfdx has to be calculated
// fx : f @ x
// w : workspace with size == n == len(x)
// RETURNS:
// J : dfdx @ x [must be pre-allocated]
func JacobianMpi(J *la.Triplet, ffcn Cb_f, x, fx, w []float64, distr bool) (err error) {
ndim := len(x)
start, endp1 := 0, ndim
if distr {
id, sz := mpi.Rank(), mpi.Size()
start, endp1 = (id*ndim)/sz, ((id+1)*ndim)/sz
if J.Max() == 0 {
J.Init(ndim, ndim, (endp1-start)*ndim)
}
} else {
if J.Max() == 0 {
J.Init(ndim, ndim, ndim*ndim)
}
}
J.Start()
// NOTE: cannot split calculation by columns unless the f function is
// independently calculated by each MPI processor.
// Otherwise, the AllReduce in f calculation would
// join pieces of f from different processors calculated for
// different x values (δx[col] from different columns).
var df float64
for col := 0; col < ndim; col++ {
xsafe := x[col]
delta := math.Sqrt(EPS * max(CTE1, math.Abs(xsafe)))
x[col] = xsafe + delta
err = ffcn(w, x) // w := f(x+δx[col])
if err != nil {
return
}
for row := start; row < endp1; row++ {
df = w[row] - fx[row]
J.Put(row, col, df/delta)
}
x[col] = xsafe
}
return
}
func main() {
mpi.Start(false)
defer func() {
mpi.Stop(false)
}()
myrank := mpi.Rank()
if myrank == 0 {
chk.PrintTitle("Test MUMPS Sol 01b")
}
var t la.Triplet
b := []float64{8.0, 45.0, -3.0, 3.0, 19.0}
switch mpi.Size() {
case 1:
t.Init(5, 5, 13)
t.Put(0, 0, 1.0)
t.Put(0, 0, 1.0)
t.Put(1, 0, 3.0)
t.Put(0, 1, 3.0)
t.Put(2, 1, -1.0)
t.Put(4, 1, 4.0)
t.Put(1, 2, 4.0)
t.Put(2, 2, -3.0)
t.Put(3, 2, 1.0)
t.Put(4, 2, 2.0)
t.Put(2, 3, 2.0)
t.Put(1, 4, 6.0)
t.Put(4, 4, 1.0)
case 2:
la.VecFill(b, 0)
if myrank == 0 {
t.Init(5, 5, 8)
t.Put(0, 0, 1.0)
t.Put(0, 0, 1.0)
t.Put(1, 0, 3.0)
t.Put(0, 1, 3.0)
t.Put(2, 1, -1.0)
t.Put(4, 1, 1.0)
t.Put(4, 1, 1.5)
t.Put(4, 1, 1.5)
b[0] = 8.0
b[1] = 40.0
b[2] = 1.5
} else {
t.Init(5, 5, 8)
t.Put(1, 2, 4.0)
t.Put(2, 2, -3.0)
t.Put(3, 2, 1.0)
t.Put(4, 2, 2.0)
t.Put(2, 3, 2.0)
t.Put(1, 4, 6.0)
t.Put(4, 4, 0.5)
t.Put(4, 4, 0.5)
b[1] = 5.0
b[2] = -4.5
b[3] = 3.0
b[4] = 19.0
}
default:
chk.Panic("this test needs 1 or 2 procs")
}
x_correct := []float64{1, 2, 3, 4, 5}
sum_b_to_root := true
la.RunMumpsTestR(&t, 1e-14, b, x_correct, sum_b_to_root)
}
func main() {
mpi.Start(false)
defer func() {
mpi.Stop(false)
}()
myrank := mpi.Rank()
if myrank == 0 {
chk.PrintTitle("Test MUMPS Sol 01a")
}
var t la.Triplet
switch mpi.Size() {
case 1:
t.Init(5, 5, 13)
t.Put(0, 0, 1.0)
t.Put(0, 0, 1.0)
t.Put(1, 0, 3.0)
t.Put(0, 1, 3.0)
t.Put(2, 1, -1.0)
t.Put(4, 1, 4.0)
t.Put(1, 2, 4.0)
t.Put(2, 2, -3.0)
t.Put(3, 2, 1.0)
t.Put(4, 2, 2.0)
t.Put(2, 3, 2.0)
t.Put(1, 4, 6.0)
t.Put(4, 4, 1.0)
case 2:
if myrank == 0 {
t.Init(5, 5, 6)
t.Put(0, 0, 1.0)
t.Put(0, 0, 1.0)
t.Put(1, 0, 3.0)
t.Put(0, 1, 3.0)
t.Put(2, 1, -1.0)
t.Put(4, 1, 4.0)
} else {
t.Init(5, 5, 7)
t.Put(1, 2, 4.0)
t.Put(2, 2, -3.0)
t.Put(3, 2, 1.0)
t.Put(4, 2, 2.0)
t.Put(2, 3, 2.0)
t.Put(1, 4, 6.0)
t.Put(4, 4, 1.0)
}
default:
chk.Panic("this test needs 1 or 2 procs")
}
b := []float64{8.0, 45.0, -3.0, 3.0, 19.0}
x_correct := []float64{1, 2, 3, 4, 5}
sum_b_to_root := false
la.RunMumpsTestR(&t, 1e-14, b, x_correct, sum_b_to_root)
}
func TestJacobian03(tst *testing.T) {
//verbose()
chk.PrintTitle("TestJacobian 03")
// grid
var g fdm.Grid2D
//g.Init(1.0, 1.0, 4, 4)
g.Init(1.0, 1.0, 6, 6)
//g.Init(1.0, 1.0, 11, 11)
// equations numbering
var e fdm.Equations
peq := utl.IntUnique(g.L, g.R, g.B, g.T)
e.Init(g.N, peq)
// K11 and K12
var K11, K12 la.Triplet
fdm.InitK11andK12(&K11, &K12, &e)
// assembly
F1 := make([]float64, e.N1)
fdm.Assemble(&K11, &K12, F1, nil, &g, &e)
// prescribed values
U2 := make([]float64, e.N2)
for _, eq := range g.L {
U2[e.FR2[eq]] = 50.0
}
for _, eq := range g.R {
U2[e.FR2[eq]] = 0.0
}
for _, eq := range g.B {
U2[e.FR2[eq]] = 0.0
}
for _, eq := range g.T {
U2[e.FR2[eq]] = 50.0
}
// functions
k11 := K11.ToMatrix(nil)
k12 := K12.ToMatrix(nil)
ffcn := func(fU1, U1 []float64) error { // K11*U1 + K12*U2 - F1
la.VecCopy(fU1, -1, F1) // fU1 := (-F1)
la.SpMatVecMulAdd(fU1, 1, k11, U1) // fU1 += K11*U1
la.SpMatVecMulAdd(fU1, 1, k12, U2) // fU1 += K12*U2
return nil
}
Jfcn := func(dfU1dU1 *la.Triplet, U1 []float64) error {
fdm.Assemble(dfU1dU1, &K12, F1, nil, &g, &e)
return nil
}
U1 := make([]float64, e.N1)
CompareJac(tst, ffcn, Jfcn, U1, 0.0075)
print_jac := false
if print_jac {
W1 := make([]float64, e.N1)
fU1 := make([]float64, e.N1)
ffcn(fU1, U1)
var Jnum la.Triplet
Jnum.Init(e.N1, e.N1, e.N1*e.N1)
Jacobian(&Jnum, ffcn, U1, fU1, W1)
la.PrintMat("K11 ", K11.ToMatrix(nil).ToDense(), "%g ", false)
la.PrintMat("Jnum", Jnum.ToMatrix(nil).ToDense(), "%g ", false)
}
test_ffcn := false
if test_ffcn {
Uc := []float64{0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 50.0, 25.0, 325.0 / 22.0, 100.0 / 11.0, 50.0 / 11.0,
0.0, 50.0, 775.0 / 22.0, 25.0, 375.0 / 22.0, 100.0 / 11.0, 0.0, 50.0, 450.0 / 11.0, 725.0 / 22.0,
25.0, 325.0 / 22.0, 0.0, 50.0, 500.0 / 11.0, 450.0 / 11.0, 775.0 / 22.0, 25.0, 0.0, 50.0, 50.0,
50.0, 50.0, 50.0, 50.0,
}
for i := 0; i < e.N1; i++ {
U1[i] = Uc[e.RF1[i]]
}
fU1 := make([]float64, e.N1)
min, max := la.VecMinMax(fU1)
io.Pf("min/max fU1 = %v\n", min, max)
}
}
func Test_linipm02(tst *testing.T) {
//verbose()
chk.PrintTitle("linipm02")
// linear program
// min 2*x0 + x1
// s.t. -x0 + x1 ≤ 1
// x0 + x1 ≥ 2 → -x0 - x1 ≤ -2
// x0 - 2*x1 ≤ 4
// x1 ≥ 0
// standard (step 1) add slack
// s.t. -x0 + x1 + x2 = 1
// -x0 - x1 + x3 = -2
// x0 - 2*x1 + x4 = 4
// standard (step 2)
// replace x0 := x0_ - x5
// because it's unbounded
// min 2*x0_ + x1 - 2*x5
// s.t. -x0_ + x1 + x2 + x5 = 1
// -x0_ - x1 + x3 + x5 = -2
// x0_ - 2*x1 + x4 - x5 = 4
// x0_,x1,x2,x3,x4,x5 ≥ 0
var T la.Triplet
T.Init(3, 6, 12)
T.Put(0, 0, -1)
T.Put(0, 1, 1)
T.Put(0, 2, 1)
T.Put(0, 5, 1)
T.Put(1, 0, -1)
T.Put(1, 1, -1)
T.Put(1, 3, 1)
T.Put(1, 5, 1)
T.Put(2, 0, 1)
T.Put(2, 1, -2)
T.Put(2, 4, 1)
T.Put(2, 5, -1)
Am := T.ToMatrix(nil)
A := Am.ToDense()
c := []float64{2, 1, 0, 0, 0, -2}
b := []float64{1, -2, 4}
// print LP
la.PrintMat("A", A, "%6g", false)
la.PrintVec("b", b, "%6g", false)
la.PrintVec("c", c, "%6g", false)
io.Pf("\n")
// solve LP
var ipm LinIpm
defer ipm.Clean()
ipm.Init(Am, b, c, nil)
err := ipm.Solve(chk.Verbose)
if err != nil {
tst.Errorf("ipm failed:\n%v", err)
return
}
// check
io.Pf("\n")
bres := make([]float64, len(b))
la.MatVecMul(bres, 1, A, ipm.X)
io.Pforan("bres = %v\n", bres)
chk.Vector(tst, "A*x=b", 1e-8, bres, b)
// fix and check x
x := ipm.X[:2]
x[0] -= ipm.X[5]
io.Pforan("x = %v\n", x)
chk.Vector(tst, "x", 1e-8, x, []float64{0.5, 1.5})
// plot
if true && chk.Verbose {
f := func(x []float64) float64 {
return c[0]*x[0] + c[1]*x[1]
}
g := func(x []float64, i int) float64 {
return A[i][0]*x[0] + A[i][1]*x[1] - b[i]
}
np := 41
vmin, vmax := []float64{-2.0, -2.0}, []float64{2.0, 2.0}
PlotTwoVarsContour("/tmp/gosl", "test_linipm02", x, np, nil, true, vmin, vmax, f,
func(x []float64) float64 { return g(x, 0) },
func(x []float64) float64 { return g(x, 1) },
)
}
}