示例#1
0
func Test_serial01(tst *testing.T) {

	//verbose()
	chk.PrintTitle("serial01")

	A := [][][]float64{
		{{100, 101, 102}, {103}, {104, 105}},
		{{106}, {107}},
		{{108}, {109, 110}},
		{{111}},
		{{112, 113, 114, 115}, {116}, {117, 118}, {119, 120, 121}},
	}

	// serialize
	PrintDeep3("A", A)
	I, P, S := Deep3Serialize(A)
	Deep3GetInfo(I, P, S, true)

	// check serialization
	chk.Ints(tst, "I", I, []int{0, 0, 0, 1, 1, 2, 2, 3, 4, 4, 4, 4})
	chk.Ints(tst, "P", P, []int{0, 3, 4, 6, 7, 8, 9, 11, 12, 16, 17, 19, 22})
	Scor := LinSpace(100, 121, 22)
	io.Pf("Scor = %v\n", Scor)
	chk.Vector(tst, "S", 1e-16, S, Scor)

	// deserialize
	B := Deep3Deserialize(I, P, S, false)
	PrintDeep3("B", B)

	// check deserialization
	chk.Deep3(tst, "A", 1e-16, A, B)
}
示例#2
0
func Test_serial02(tst *testing.T) {

	//verbose()
	chk.PrintTitle("serial02")

	A := [][][]float64{
		{{1, 3, 4}, {6}, {232, 23, 292, 2023}, {2, 3}},
		{{0}, {1}, {0}},
		{{0, 5, 6, 8, 3, 0}},
	}

	// serialize
	PrintDeep3("A", A)
	I, P, S := Deep3Serialize(A)
	Deep3GetInfo(I, P, S, true)

	// check serialization
	chk.Ints(tst, "I", I, []int{0, 0, 0, 0, 1, 1, 1, 2})
	chk.Ints(tst, "P", P, []int{0, 3, 4, 8, 10, 11, 12, 13, 19})
	chk.Vector(tst, "S", 1e-16, S, []float64{1, 3, 4, 6, 232, 23, 292, 2023, 2, 3, 0, 1, 0, 0, 5, 6, 8, 3, 0})

	// deserialize
	B := Deep3Deserialize(I, P, S, false)
	PrintDeep3("B", B)

	// check deserialization
	chk.Deep3(tst, "A", 1e-16, A, B)
}
示例#3
0
func Test_deep02(tst *testing.T) {

	//verbose()
	chk.PrintTitle("deep02")

	a := Deep3alloc(3, 2, 4)
	Deep3set(a, 666)
	chk.Deep3(tst, "a", 1e-16, a, [][][]float64{
		{{666, 666, 666, 666}, {666, 666, 666, 666}},
		{{666, 666, 666, 666}, {666, 666, 666, 666}},
		{{666, 666, 666, 666}, {666, 666, 666, 666}},
	})
	io.Pf("a = %v\n", a)

	b := Deep4alloc(3, 2, 1, 2)
	Deep4set(b, 666)
	chk.Deep4(tst, "b", 1e-16, b, [][][][]float64{
		{{{666, 666}}, {{666, 666}}},
		{{{666, 666}}, {{666, 666}}},
		{{{666, 666}}, {{666, 666}}},
	})
	io.Pf("b = %v\n", b)
}
示例#4
0
func Test_smpinvs02(tst *testing.T) {

	//verbose()
	chk.PrintTitle("smpinvs02")

	// coefficients for smp invariants
	smp_a := -1.0
	smp_b := 0.5
	smp_β := 1e-1 // derivative values become too high with
	smp_ϵ := 1e-1 // small β and ϵ @ zero

	// constants for checking derivatives
	dver := chk.Verbose
	dtol := 1e-9
	dtol2 := 1e-8

	// run tests
	nd := test_nd
	for idxA := 0; idxA < len(test_nd); idxA++ {
		//for idxA := 0; idxA < 1; idxA++ {
		//for idxA := 10; idxA < 11; idxA++ {

		// tensor and eigenvalues
		A := test_AA[idxA]
		a := M_Alloc2(nd[idxA])
		Ten2Man(a, A)
		L := make([]float64, 3)
		M_EigenValsNum(L, a)

		// SMP director
		N := make([]float64, 3)
		n := make([]float64, 3)
		m := SmpDirector(N, L, smp_a, smp_b, smp_β, smp_ϵ)
		SmpUnitDirector(n, m, N)

		// output
		io.PfYel("\n\ntst # %d ###################################################################################\n", idxA)
		io.Pforan("L = %v\n", L)
		io.Pforan("N = %v\n", N)
		io.Pforan("m = %v\n", m)
		io.Pfpink("n = %v\n", n)
		chk.Vector(tst, "L", 1e-12, L, test_λ[idxA])
		chk.Scalar(tst, "norm(n)==1", 1e-15, la.VecNorm(n), 1)
		chk.Scalar(tst, "m=norm(N)", 1e-14, m, la.VecNorm(N))

		// dN/dL
		var tmp float64
		N_tmp := make([]float64, 3)
		dNdL := make([]float64, 3)
		SmpDirectorDeriv1(dNdL, L, smp_a, smp_b, smp_β, smp_ϵ)
		io.Pfpink("\ndNdL = %v\n", dNdL)
		for i := 0; i < 3; i++ {
			dnum, _ := num.DerivCentral(func(x float64, args ...interface{}) (res float64) {
				tmp, L[i] = L[i], x
				SmpDirector(N_tmp, L, smp_a, smp_b, smp_β, smp_ϵ)
				L[i] = tmp
				return N_tmp[i]
			}, L[i], 1e-6)
			chk.AnaNum(tst, io.Sf("dN/dL[%d][%d]", i, i), dtol, dNdL[i], dnum, dver)
		}

		// dm/dL
		n_tmp := make([]float64, 3)
		dmdL := make([]float64, 3)
		SmpNormDirectorDeriv1(dmdL, m, N, dNdL)
		io.Pfpink("\ndmdL = %v\n", dmdL)
		for j := 0; j < 3; j++ {
			dnum, _ := num.DerivCentral(func(x float64, args ...interface{}) (res float64) {
				tmp, L[j] = L[j], x
				m_tmp := SmpDirector(N_tmp, L, smp_a, smp_b, smp_β, smp_ϵ)
				L[j] = tmp
				return m_tmp
			}, L[j], 1e-6)
			chk.AnaNum(tst, io.Sf("dm/dL[%d]", j), dtol, dmdL[j], dnum, dver)
		}

		// dn/dL
		dndL := la.MatAlloc(3, 3)
		SmpUnitDirectorDeriv1(dndL, m, N, dNdL, dmdL)
		io.Pfpink("\ndndL = %v\n", dndL)
		for i := 0; i < 3; i++ {
			for j := 0; j < 3; j++ {
				dnum, _ := num.DerivCentral(func(x float64, args ...interface{}) (res float64) {
					tmp, L[j] = L[j], x
					m_tmp := SmpDirector(N_tmp, L, smp_a, smp_b, smp_β, smp_ϵ)
					SmpUnitDirector(n_tmp, m_tmp, N_tmp)
					L[j] = tmp
					return n_tmp[i]
				}, L[j], 1e-6)
				chk.AnaNum(tst, io.Sf("dn/dL[%d][%d]", i, j), dtol, dndL[i][j], dnum, dver)
			}
		}

		// change tolerance
		dtol2_tmp := dtol2
		if idxA == 10 || idxA == 11 {
			dtol2 = 1e-6
		}

		// d²m/dLdL
		dNdL_tmp := make([]float64, 3)
		dmdL_tmp := make([]float64, 3)
		d2NdL2 := make([]float64, 3)
		d2mdLdL := la.MatAlloc(3, 3)
		SmpDirectorDeriv2(d2NdL2, L, smp_a, smp_b, smp_β, smp_ϵ)
		SmpNormDirectorDeriv2(d2mdLdL, L, smp_a, smp_b, smp_β, smp_ϵ, m, N, dNdL, d2NdL2, dmdL)
		io.Pfpink("\nd2mdLdL = %v\n", d2mdLdL)
		for i := 0; i < 3; i++ {
			for j := 0; j < 3; j++ {
				dnum, _ := num.DerivCentral(func(x float64, args ...interface{}) (res float64) {
					tmp, L[j] = L[j], x
					m_tmp := SmpDirector(N_tmp, L, smp_a, smp_b, smp_β, smp_ϵ)
					SmpDirectorDeriv1(dNdL_tmp, L, smp_a, smp_b, smp_β, smp_ϵ)
					SmpNormDirectorDeriv1(dmdL_tmp, m_tmp, N_tmp, dNdL_tmp)
					L[j] = tmp
					return dmdL_tmp[i]
				}, L[j], 1e-6)
				chk.AnaNum(tst, io.Sf("d2m/dL[%d]dL[%d]", i, j), dtol2, d2mdLdL[i][j], dnum, dver)
			}
		}

		// d²N/dLdL
		io.Pfpink("\nd²N/dLdL\n")
		for i := 0; i < 3; i++ {
			dnum, _ := num.DerivCentral(func(x float64, args ...interface{}) (res float64) {
				tmp, L[i] = L[i], x
				SmpDirectorDeriv1(dNdL_tmp, L, smp_a, smp_b, smp_β, smp_ϵ)
				L[i] = tmp
				return dNdL_tmp[i]
			}, L[i], 1e-6)
			chk.AnaNum(tst, io.Sf("d²N[%d]/dL[%d]dL[%d]", i, i, i), dtol2, d2NdL2[i], dnum, dver)
		}

		// d²n/dLdL
		io.Pfpink("\nd²n/dLdL\n")
		dndL_tmp := la.MatAlloc(3, 3)
		d2ndLdL := utl.Deep3alloc(3, 3, 3)
		SmpUnitDirectorDeriv2(d2ndLdL, m, N, dNdL, d2NdL2, dmdL, n, d2mdLdL, dndL)
		for i := 0; i < 3; i++ {
			for j := 0; j < 3; j++ {
				for k := 0; k < 3; k++ {
					dnum, _ := num.DerivCentral(func(x float64, args ...interface{}) (res float64) {
						tmp, L[k] = L[k], x
						m_tmp := SmpDirector(N_tmp, L, smp_a, smp_b, smp_β, smp_ϵ)
						SmpDirectorDeriv1(dNdL_tmp, L, smp_a, smp_b, smp_β, smp_ϵ)
						SmpNormDirectorDeriv1(dmdL_tmp, m_tmp, N_tmp, dNdL_tmp)
						SmpUnitDirectorDeriv1(dndL_tmp, m_tmp, N_tmp, dNdL_tmp, dmdL_tmp)
						L[k] = tmp
						return dndL_tmp[i][j]
					}, L[k], 1e-6)
					chk.AnaNum(tst, io.Sf("d²n[%d]/dL[%d]dL[%d]", i, j, k), dtol2, d2ndLdL[i][j][k], dnum, dver)
				}
			}
		}

		// recover tolerance
		dtol2 = dtol2_tmp

		// SMP derivs
		//if false {
		if true {
			io.Pfpink("\nSMP derivs\n")
			dndL_ := la.MatAlloc(3, 3)
			dNdL_ := make([]float64, 3)
			d2ndLdL_ := utl.Deep3alloc(3, 3, 3)
			N_ := make([]float64, 3)
			F_ := make([]float64, 3)
			G_ := make([]float64, 3)
			m_ := SmpDerivs1(dndL_, dNdL_, N_, F_, G_, L, smp_a, smp_b, smp_β, smp_ϵ)
			SmpDerivs2(d2ndLdL_, L, smp_a, smp_b, smp_β, smp_ϵ, m_, N_, F_, G_, dNdL_, dndL_)
			chk.Scalar(tst, "m_", 1e-14, m_, m)
			chk.Vector(tst, "N_", 1e-15, N_, N)
			chk.Vector(tst, "dNdL_", 1e-15, dNdL_, dNdL)
			chk.Matrix(tst, "dndL_", 1e-13, dndL_, dndL)
			chk.Deep3(tst, "d2ndLdL_", 1e-11, d2ndLdL_, d2ndLdL)
		}
	}
}