Golang matrix.FloatMatrix类代码示例

// Inverse UNIT diagonal tridiagonal matrix
func unblockedInverseUnitLower(A *matrix.FloatMatrix) (err error) {
	var ATL, ATR, ABL, ABR matrix.FloatMatrix
	var A00, a10t, a11, A20, a21, A22 matrix.FloatMatrix

	err = nil
	partition2x2(
		&ATL, &ATR,
		&ABL, &ABR, A, 0, 0, pTOPLEFT)

	for ATL.Rows() < A.Rows() {
		repartition2x2to3x3(&ATL,
			&A00, nil, nil,
			&a10t, &a11, nil,
			&A20, &a21, &A22, A, 1, pBOTTOMRIGHT)
		// -------------------------------------------------

		// a21 = -a21
		Scale(&a21, -1.0)
		// A20 = A20 + a21*a10.t
		MVRankUpdate(&A20, &a21, &a10t, 1.0)

		// -------------------------------------------------
		continue3x3to2x2(
			&ATL, &ATR,
			&ABL, &ABR, &A00, &a11, &A22, A, pBOTTOMRIGHT)
	}
	return
}

func InverseTrm(A *matrix.FloatMatrix, flags Flags, nb int) (*matrix.FloatMatrix, error) {
	var err error = nil
	if nb == 0 || A.Cols() < nb {
		if flags&UNIT != 0 {
			if flags&LOWER != 0 {
				err = unblockedInverseUnitLower(A)
			} else {
				err = unblockedInverseUnitUpper(A)
			}
		} else {
			if flags&LOWER != 0 {
				err = unblockedInverseLower(A)
			} else {
				err = unblockedInverseUpper(A)
			}
		}
	} else {
		if flags&LOWER != 0 {
			err = blockedInverseLower(A, flags, nb)
		} else {
			err = blockedInverseUpper(A, flags, nb)
		}
	}
	return A, err
}

// Inverse NON-UNIT diagonal tridiagonal matrix
func unblockedInverseUpper(A *matrix.FloatMatrix) (err error) {
	var ATL, ATR, ABL, ABR matrix.FloatMatrix
	var A00, a01, A02, a11, a12t, A22 matrix.FloatMatrix

	err = nil
	partition2x2(
		&ATL, &ATR,
		&ABL, &ABR, A, 0, 0, pTOPLEFT)

	for ATL.Rows() < A.Rows() {
		repartition2x2to3x3(&ATL,
			&A00, &a01, &A02,
			nil, &a11, &a12t,
			nil, nil, &A22, A, 1, pBOTTOMRIGHT)
		// -------------------------------------------------
		aval := a11.Float()

		// a12 = -a12/a11
		InvScale(&a12t, -aval)
		// A02 = A02 + a01*a12
		MVRankUpdate(&A02, &a01, &a12t, 1.0)
		// a01 = a01/a11
		InvScale(&a01, aval)
		// a11 = 1.0/a11
		a11.SetAt(0, 0, 1.0/aval)

		// -------------------------------------------------
		continue3x3to2x2(
			&ATL, &ATR,
			&ABL, &ABR, &A00, &a11, &A22, A, pBOTTOMRIGHT)
	}
	return
}

func blockedInverseUpper(A *matrix.FloatMatrix, flags Flags, nb int) (err error) {
	var ATL, ATR, ABL, ABR matrix.FloatMatrix
	var A00, A01, A02, A11, A12, A22 matrix.FloatMatrix

	err = nil
	partition2x2(
		&ATL, &ATR,
		&ABL, &ABR, A, 0, 0, pTOPLEFT)

	for ATL.Rows() < A.Rows() {
		repartition2x2to3x3(&ATL,
			&A00, &A01, &A02,
			nil, &A11, &A12,
			nil, nil, &A22, A, nb, pBOTTOMRIGHT)
		// -------------------------------------------------
		// libflame, variant 1

		// A01 = A00*A01
		MultTrm(&A01, &A00, 1.0, flags)
		// A01 = -A01 / triu(A11)
		SolveTrm(&A01, &A11, -1.0, flags|RIGHT)
		// A11 = inv(A11)
		if flags&UNIT != 0 {
			unblockedInverseUnitUpper(&A11)
		} else {
			unblockedInverseUpper(&A11)
		}
		// -------------------------------------------------
		continue3x3to2x2(
			&ATL, &ATR,
			&ABL, &ABR, &A00, &A11, &A22, A, pBOTTOMRIGHT)
	}
	return
}

func runRefTest(A *matrix.FloatMatrix, ntest, LB int) time.Duration {

	var mintime time.Duration

	N := A.Cols()
	tau := matrix.FloatZeros(N, 1)

	fnc := func() {
		ERRlapack = lapack.Geqrf(A, tau)
	}

	A0 := A.Copy()
	for n := 0; n < ntest; n++ {
		if n > 0 {
			// restore original A
			A0.CopyTo(A)
			tau.Scale(0.0)
		}
		mperf.FlushCache()
		time0 := mperf.Timeit(fnc)
		if n == 0 || time0 < mintime {
			mintime = time0
		}
	}
	return mintime
}

func TestSolveLeastSquaresQRT(t *testing.T) {
	M := 60
	N := 40
	K := 30
	nb := 12

	A := matrix.FloatUniform(M, N)
	B := matrix.FloatZeros(M, K)
	X0 := matrix.FloatUniform(N, K)

	// B = A*X0
	Mult(B, A, X0, 1.0, 0.0, NOTRANS)
	W := matrix.FloatZeros(N, nb)
	T := matrix.FloatZeros(N, N)

	QR, err := DecomposeQRT(A, T, W, nb)
	if err != nil {
		t.Logf("decompose error: %v\n", err)
	}
	// B' = A.-1*B
	err = SolveQRT(B, QR, T, W, NOTRANS, nb)

	// expect B[0:N, 0:K] == X0, B[N:M, 0:K] == 0.0
	var Xref matrix.FloatMatrix
	Bref := matrix.FloatZeros(M, K)
	Bref.SubMatrix(&Xref, 0, 0, N, K)
	Xref.Plus(X0)
	Bref.Minus(B)
	t.Logf("\nmin ||B - A*X0||\n\twhere B = A*X0\n")
	t.Logf("||B - A*X0||_1 ~~ 0.0: %e\n", NormP(Bref, NORM_ONE))
}

func updateBlas(t *testing.T, Y1, Y2, C1, C2, T, W *matrix.FloatMatrix) {
	if W.Rows() != C1.Cols() {
		panic("W.Rows != C1.Cols")
	}
	// W = C1.T
	ScalePlus(W, C1, 0.0, 1.0, TRANSB)
	//fmt.Printf("W = C1.T:\n%v\n", W)
	// W = C1.T*Y1
	blas.TrmmFloat(Y1, W, 1.0, linalg.OptLower, linalg.OptUnit, linalg.OptRight)
	t.Logf("W = C1.T*Y1:\n%v\n", W)
	// W = W + C2.T*Y2
	blas.GemmFloat(C2, Y2, W, 1.0, 1.0, linalg.OptTransA)
	t.Logf("W = W + C2.T*Y2:\n%v\n", W)

	// --- here: W == C.T*Y ---
	// W = W*T
	blas.TrmmFloat(T, W, 1.0, linalg.OptUpper, linalg.OptRight)
	t.Logf("W = C.T*Y*T:\n%v\n", W)

	// --- here: W == C.T*Y*T ---
	// C2 = C2 - Y2*W.T
	blas.GemmFloat(Y2, W, C2, -1, 1.0, linalg.OptTransB)
	t.Logf("C2 = C2 - Y2*W.T:\n%v\n", C2)
	//  W = Y1*W.T ==> W.T = W*Y1.T
	blas.TrmmFloat(Y1, W, 1.0, linalg.OptLower,
		linalg.OptUnit, linalg.OptRight, linalg.OptTrans)
	t.Logf("W.T = W*Y1.T:\n%v\n", W)

	// C1 = C1 - W.T
	ScalePlus(C1, W, 1.0, -1.0, TRANSB)
	//fmt.Printf("C1 = C1 - W.T:\n%v\n", C1)

	// --- here: C = (I - Y*T*Y.T).T * C ---
}

func runTest(A *matrix.FloatMatrix, ntest, LB int) time.Duration {

	var flags matops.Flags
	var mintime time.Duration

	flags = matops.LOWER
	if testUpper {
		flags = matops.UPPER
	}

	fnc := func() {
		_, ERRmatops = matops.DecomposeCHOL(A, flags, LB)
	}

	A0 := A.Copy()
	for n := 0; n < ntest; n++ {
		if n > 0 {
			// restore original A
			A0.CopyTo(A)
		}
		mperf.FlushCache()
		time0 := mperf.Timeit(fnc)
		if n == 0 || time0 < mintime {
			mintime = time0
		}
		if verbose {
			fmt.Printf("%.4f ms\n", time0.Seconds()*1000.0)
		}
	}
	return mintime
}

func setDiagonal(M *matrix.FloatMatrix, srow, scol, erow, ecol int, val float64) {
	for i := srow; i < erow; i++ {
		if i < ecol {
			M.SetAt(i, i, val)
		}
	}
}

// max |x|
func AMax(X *matrix.FloatMatrix) float64 {
	ix := IAMax(X)
	if ix == -1 {
		return math.NaN()
	}
	return X.GetIndex(ix)
}

// See function Syrk.
func SyrkFloat(A, C *matrix.FloatMatrix, alpha, beta float64, opts ...linalg.Option) (err error) {

	params, e := linalg.GetParameters(opts...)
	if e != nil {
		err = e
		return
	}
	ind := linalg.GetIndexOpts(opts...)
	err = check_level3_func(ind, fsyrk, A, nil, C, params)
	if e != nil || err != nil {
		return
	}
	if ind.N == 0 {
		return
	}
	Aa := A.FloatArray()
	Ca := C.FloatArray()
	uplo := linalg.ParamString(params.Uplo)
	trans := linalg.ParamString(params.Trans)
	//diag := linalg.ParamString(params.Diag)
	dsyrk(uplo, trans, ind.N, ind.K, alpha, Aa[ind.OffsetA:], ind.LDa, beta,
		Ca[ind.OffsetC:], ind.LDc)

	return
}

func runRefTest(A *matrix.FloatMatrix, ntest, LB int) time.Duration {

	var mintime time.Duration

	lopt := linalg.OptLower
	if testUpper {
		lopt = linalg.OptUpper
	}

	fnc := func() {
		ERRlapack = lapack.Potrf(A, lopt)
	}

	A0 := A.Copy()
	for n := 0; n < ntest; n++ {
		if n > 0 {
			// restore original A
			A0.CopyTo(A)
		}
		mperf.FlushCache()
		time0 := mperf.Timeit(fnc)
		if n == 0 || time0 < mintime {
			mintime = time0
		}
	}
	return mintime
}

/*
   Copy x to y using packed storage.

   The vector x is an element of S, with the 's' components stored in
   unpacked storage.  On return, x is copied to y with the 's' components
   stored in packed storage and the off-diagonal entries scaled by
   sqrt(2).
*/
func pack(x, y *matrix.FloatMatrix, dims *sets.DimensionSet, opts ...la_.Option) (err error) {
	/*DEBUGGED*/
	err = nil
	mnl := la_.GetIntOpt("mnl", 0, opts...)
	offsetx := la_.GetIntOpt("offsetx", 0, opts...)
	offsety := la_.GetIntOpt("offsety", 0, opts...)

	nlq := mnl + dims.At("l")[0] + dims.Sum("q")
	blas.Copy(x, y, &la_.IOpt{"n", nlq}, &la_.IOpt{"offsetx", offsetx},
		&la_.IOpt{"offsety", offsety})

	iu, ip := offsetx+nlq, offsety+nlq
	for _, n := range dims.At("s") {
		for k := 0; k < n; k++ {
			blas.Copy(x, y, &la_.IOpt{"n", n - k}, &la_.IOpt{"offsetx", iu + k*(n+1)},
				&la_.IOpt{"offsety", ip})
			y.SetIndex(ip, (y.GetIndex(ip) / math.Sqrt(2.0)))
			ip += n - k
		}
		iu += n * n
	}
	np := dims.SumPacked("s")
	blas.ScalFloat(y, math.Sqrt(2.0), &la_.IOpt{"n", np}, &la_.IOpt{"offset", offsety + nlq})
	return
}

func trmvTest(t *testing.T, A *matrix.FloatMatrix, flags Flags, nb int) bool {
	N := A.Cols()
	//S := 0
	//E := A.Cols()
	X0 := matrix.FloatWithValue(A.Rows(), 1, 2.0)
	X1 := X0.Copy()

	trans := linalg.OptNoTrans
	if flags&TRANS != 0 {
		trans = linalg.OptTrans
	}
	diag := linalg.OptNonUnit
	if flags&UNIT != 0 {
		diag = linalg.OptUnit
	}
	uplo := linalg.OptUpper
	if flags&LOWER != 0 {
		uplo = linalg.OptLower
	}

	blas.TrmvFloat(A, X0, uplo, diag, trans)

	Ar := A.FloatArray()
	Xr := X1.FloatArray()
	if nb == 0 {
		DTrimvUnblkMV(Xr, Ar, flags, 1, A.LeadingIndex(), N)
	}
	result := X0.AllClose(X1)
	t.Logf("   X0 == X1: %v\n", result)
	if !result && A.Rows() < 8 {
		t.Logf("  BLAS TRMV X0:\n%v\n", X0)
		t.Logf("  DTrmv X1:\n%v\n", X1)
	}
	return result
}

// In-place version of pack(), which also accepts matrix arguments x.
// The columns of x are elements of S, with the 's' components stored
// in unpacked storage.  On return, the 's' components are stored in
// packed storage and the off-diagonal entries are scaled by sqrt(2).
//
func pack2(x *matrix.FloatMatrix, dims *sets.DimensionSet, mnl int) (err error) {
	if len(dims.At("s")) == 0 {
		return nil
	}

	const sqrt2 = 1.41421356237309504880

	iu := mnl + dims.Sum("l", "q")
	ip := iu
	row := matrix.FloatZeros(1, x.Cols())
	//fmt.Printf("x.size = %d %d\n", x.Rows(), x.Cols())
	for _, n := range dims.At("s") {
		for k := 0; k < n; k++ {
			cnt := n - k
			row = x.GetRow(iu+(n+1)*k, row)
			//fmt.Printf("%02d: %v\n", iu+(n+1)*k, x.FloatArray())
			x.SetRow(ip, row)
			for i := 1; i < n-k; i++ {
				row = x.GetRow(iu+(n+1)*k+i, row)
				//fmt.Printf("%02d: %v\n", iu+(n+1)*k+i, x.FloatArray())
				x.SetRow(ip+i, row.Scale(sqrt2))
			}
			ip += cnt
		}
		iu += n * n
	}
	return nil
}