Esempio n. 1
0
    // Last two ints before fcnData are for the array index baselines.
    public static int hybrd(Func <int, double[], double[], int, int, int, fcnData> fcn,
                            int n, double[] x,
                            double[] fvec, double xtol, int maxfev, int ml, int mu, double epsfcn,
                            double[] diag, int mode, double factor, int nprint, int nfev,
                            double[] fjac, int ldfjac, double[] r, int lr, double[] qtf, double[] wa1,
                            double[] wa2, double[] wa3, double[] wa4, int fjacIndex, int rIndex, int qtfIndex, int wa1Index, int wa2Index, int wa3Index, int wa4Index)

    //****************************************************************************80
    //
    //  Purpose:
    //
    //    hybrd() finds a zero of a system of N nonlinear equations.
    //
    //  Discussion:
    //
    //    The purpose of HYBRD is to find a zero of a system of
    //    N nonlinear functions in N variables by a modification
    //    of the Powell hybrid method.
    //
    //    The user must provide FCN, which calculates the functions.
    //
    //    The jacobian is calculated by a forward-difference approximation.
    //
    //  Licensing:
    //
    //    This code is distributed under the GNU LGPL license.
    //
    //  Modified:
    //
    //    08 April 2010
    //
    //  Author:
    //
    //    Original FORTRAN77 version by Jorge More, Burt Garbow, Ken Hillstrom.
    //    C++ version by John Burkardt.
    //
    //  Reference:
    //
    //    Jorge More, Burton Garbow, Kenneth Hillstrom,
    //    User Guide for MINPACK-1,
    //    Technical Report ANL-80-74,
    //    Argonne National Laboratory, 1980.
    //
    //  Parameters:
    //
    //       fcn is the name of the user-supplied subroutine which
    //         calculates the functions. fcn must be declared
    //         in an external statement in the user calling
    //         program, and should be written as follows.
    //
    //         subroutine fcn(n,x,fvec,iflag)
    //         integer n,iflag
    //         double precision x(n),fvec(n)
    //         ----------
    //         calculate the functions at x and
    //         return this vector in fvec.
    //         ---------
    //         return
    //         end
    //
    //         the value of iflag should not be changed by fcn unless
    //         the user wants to terminate execution of hybrd.
    //         in this case set iflag to a negative integer.
    //
    //    Input, int N, the number of functions and variables.
    //
    //    Input/output, double X[N].  On input an initial estimate of the solution.
    //    On output, the final estimate of the solution.
    //
    //    Output, double FVEC[N], the functions evaluated at the output value of X.
    //
    //    Input, double XTOL, a nonnegative value.  Termination occurs when the
    //    relative error between two consecutive iterates is at most XTOL.
    //
    //    Input, int MAXFEV.  Termination occurs when the number of calls to FCN
    //    is at least MAXFEV by the end of an iteration.
    //
    //    Input, int ML, specifies the number of subdiagonals within the band of
    //    the jacobian matrix.  If the jacobian is not banded, set
    //    ml to at least n - 1.
    //
    //       mu is a nonnegative integer input variable which specifies
    //         the number of superdiagonals within the band of the
    //         jacobian matrix. if the jacobian is not banded, set
    //         mu to at least n - 1.
    //
    //       epsfcn is an input variable used in determining a suitable
    //         step length for the forward-difference approximation. this
    //         approximation assumes that the relative errors in the
    //         functions are of the order of epsfcn. if epsfcn is less
    //         than the machine precision, it is assumed that the relative
    //         errors in the functions are of the order of the machine
    //         precision.
    //
    //       diag is an array of length n. if mode = 1 (see
    //         below), diag is internally set. if mode = 2, diag
    //         must contain positive entries that serve as
    //         multiplicative scale factors for the variables.
    //
    //       mode is an integer input variable. if mode = 1, the
    //         variables will be scaled internally. if mode = 2,
    //         the scaling is specified by the input diag. other
    //         values of mode are equivalent to mode = 1.
    //
    //       factor is a positive input variable used in determining the
    //         initial step bound. this bound is set to the product of
    //         factor and the euclidean norm of diag*x if nonzero, or else
    //         to factor itself. in most cases factor should lie in the
    //         interval (.1,100.). 100. is a generally recommended value.
    //
    //       nprint is an integer input variable that enables controlled
    //         printing of iterates if it is positive. in this case,
    //         fcn is called with iflag = 0 at the beginning of the first
    //         iteration and every nprint iterations thereafter and
    //         immediately prior to return, with x and fvec available
    //         for printing. if nprint is not positive, no special calls
    //         of fcn with iflag = 0 are made.
    //
    //       info is an integer output variable. if the user has
    //         terminated execution, info is set to the (negative)
    //         value of iflag. see description of fcn. otherwise,
    //         info is set as follows.
    //
    //         info = 0   improper input parameters.
    //
    //         info = 1   relative error between two consecutive iterates
    //                    is at most xtol.
    //
    //         info = 2   number of calls to fcn has reached or exceeded
    //                    maxfev.
    //
    //         info = 3   xtol is too small. no further improvement in
    //                    the approximate solution x is possible.
    //
    //         info = 4   iteration is not making good progress, as
    //                    measured by the improvement from the last
    //                    five jacobian evaluations.
    //
    //         info = 5   iteration is not making good progress, as
    //                    measured by the improvement from the last
    //                    ten iterations.
    //
    //       nfev is an integer output variable set to the number of
    //         calls to fcn.
    //
    //       fjac is an output n by n array which contains the
    //         orthogonal matrix q produced by the qr factorization
    //         of the final approximate jacobian.
    //
    //       ldfjac is a positive integer input variable not less than n
    //         which specifies the leading dimension of the array fjac.
    //
    //       r is an output array of length lr which contains the
    //         upper triangular matrix produced by the qr factorization
    //         of the final approximate jacobian, stored rowwise.
    //
    //       lr is a positive integer input variable not less than
    //         (n*(n+1))/2.
    //
    //       qtf is an output array of length n which contains
    //         the vector (q transpose)*fvec.
    //
    //       wa1, wa2, wa3, and wa4 are work arrays of length n.
    //
    {
        double delta = 0;

        int[]        iwa = new int[1];
        int          j;
        const double p001  = 0.001;
        const double p0001 = 0.0001;
        const double p1    = 0.1;
        const double p5    = 0.5;
        double       xnorm = 0;
        //
        //  Certain loops in this function were kept closer to their original FORTRAN77
        //  format, to avoid confusing issues with the array index L.  These loops are
        //  marked "DO NOT ADJUST", although they certainly could be adjusted (carefully)
        //  once the initial translated code is tested.
        //

        //
        //  EPSMCH is the machine precision.
        //
        double epsmch = typeMethods.r8_epsilon();

        int info  = 0;
        int iflag = 0;

        nfev = 0;
        switch (n)
        {
        //
        //  Check the input parameters.
        //
        case <= 0:
            info = 0;
            return(info);
        }

        switch (xtol)
        {
        case < 0.0:
            info = 0;
            return(info);
        }

        switch (maxfev)
        {
        case <= 0:
            info = 0;
            return(info);
        }

        switch (ml)
        {
        case < 0:
            info = 0;
            return(info);
        }

        switch (mu)
        {
        case < 0:
            info = 0;
            return(info);
        }

        switch (factor)
        {
        case <= 0.0:
            info = 0;
            return(info);
        }

        if (ldfjac < n)
        {
            info = 0;
            return(info);
        }

        if (lr < n * (n + 1) / 2)
        {
            info = 0;
            return(info);
        }

        switch (mode)
        {
        case 2:
        {
            for (j = 0; j < n; j++)
            {
                switch (diag[j])
                {
                case <= 0.0:
                    info = 0;
                    return(info);
                }
            }

            break;
        }
        }

        //
        //  Evaluate the function at the starting point and calculate its norm.
        //
        iflag = 1;
        fcnData res = fcn(n, x, fvec, iflag, 0, 0);

        fvec  = res.fvec;
        iflag = res.iflag;
        nfev  = 1;
        switch (iflag)
        {
        case < 0:
            info = iflag;
            return(info);
        }

        double fnorm = Helpers.enorm(n, fvec);
        //
        //  Determine the number of calls to FCN needed to compute the jacobian matrix.
        //
        int msum = Math.Min(ml + mu + 1, n);
        //
        //  Initialize iteration counter and monitors.
        //
        int iter   = 1;
        int ncsuc  = 0;
        int ncfail = 0;
        int nslow1 = 0;
        int nslow2 = 0;

        //
        //  Beginning of the outer loop.
        //
        for (;;)
        {
            bool jeval = true;
            //
            //  Calculate the jacobian matrix.
            //
            iflag = 2;
            fdjac1(fcn, n, x, fvec, fjac, ldfjac, ref iflag, ml, mu, epsfcn, wa1, wa2, fjacIndex: fjacIndex, wa1Index: wa1Index, wa2Index: wa2Index);

            nfev += msum;
            switch (iflag)
            {
            case < 0:
                info = iflag;
                return(info);
            }

            //
            //  Compute the QR factorization of the jacobian.
            //
            int tmpi = 1;
            QRSolve.qrfac(n, n, ref fjac, ldfjac, false, ref iwa, ref tmpi, ref wa1, ref wa2, rdiagIndex: wa1Index, acnormIndex: wa2Index);
            switch (iter)
            {
            //
            //  On the first iteration and if MODE is 1, scale according
            //  to the norms of the columns of the initial jacobian.
            //
            case 1:
            {
                switch (mode)
                {
                case 1:
                {
                    for (j = 0; j < n; j++)
                    {
                        if (wa2[wa2Index + j] != 0.0)
                        {
                            diag[j] = wa2[wa2Index + j];
                        }
                        else
                        {
                            diag[j] = 1.0;
                        }
                    }

                    break;
                }
                }

                //
                //  On the first iteration, calculate the norm of the scaled X
                //  and initialize the step bound DELTA.
                //
                for (j = 0; j < n; j++)
                {
                    wa3[wa3Index + j] = diag[j] * x[j];
                }

                xnorm = Helpers.enorm(n, wa3);

                delta = xnorm switch
                {
                    0.0 => factor,
                    _ => factor * xnorm
                };

                break;
            }
            }

            //
            //  Form Q' * FVEC and store in QTF.
            //
            int i;
            for (i = 0; i < n; i++)
            {
                qtf[qtfIndex + i] = fvec[i];
            }

            double temp;
            double sum;
            for (j = 0; j < n; j++)
            {
                if (fjac[fjacIndex + j + j * ldfjac] == 0.0)
                {
                    continue;
                }

                sum = 0.0;
                for (i = j; i < n; i++)
                {
                    sum += fjac[fjacIndex + i + j * ldfjac] * qtf[qtfIndex + i];
                }

                temp = -sum / fjac[fjacIndex + j + j * ldfjac];
                for (i = j; i < n; i++)
                {
                    qtf[qtfIndex + i] += fjac[fjacIndex + i + j * ldfjac] * temp;
                }
            }

            //
            //  Copy the triangular factor of the QR factorization into R.
            //
            //  DO NOT ADJUST THIS LOOP, BECAUSE OF L.
            //
            int l;
            for (j = 1; j <= n; j++)
            {
                l = j;
                for (i = 1; i <= j - 1; i++)
                {
                    r[rIndex + (l - 1)] = fjac[fjacIndex + (i - 1) + (j - 1) * ldfjac];
                    l = l + n - i;
                }

                r[rIndex + (l - 1)] = wa1[wa1Index + (j - 1)];
                switch (wa1[wa1Index + (j - 1)])
                {
                case 0.0:
                    Console.WriteLine("  Matrix is singular.");
                    break;
                }
            }

            //
            //  Accumulate the orthogonal factor in FJAC.
            //
            QRSolve.qform(n, n, ref fjac, ldfjac, qIndex: fjacIndex);
            switch (mode)
            {
            //
            //  Rescale if necessary.
            //
            case 1:
            {
                for (j = 0; j < n; j++)
                {
                    diag[j] = Math.Max(diag[j], wa2[wa2Index + j]);
                }

                break;
            }
            }

            //
            //  Beginning of the inner loop.
            //
            for (;;)
            {
                switch (nprint)
                {
                //
                //  If requested, call FCN to enable printing of iterates.
                //
                case > 0:
                {
                    switch ((iter - 1) % nprint)
                    {
                    case 0:
                    {
                        iflag = 0;
                        fcn(n, x, fvec, iflag, 0, 0);
                        switch (iflag)
                        {
                        case < 0:
                            info = iflag;
                            return(info);
                        }

                        break;
                    }
                    }

                    break;
                }
                }

                //
                //  Determine the direction P.
                //
                dogleg(n, r, lr, diag, qtf, delta, ref wa1, wa2, wa3, rIndex: rIndex, qtbIndex: qtfIndex, xIndex: wa1Index, wa1Index: wa2Index, wa2Index: wa3Index);
                //
                //  Store the direction P and X + P.  Calculate the norm of P.
                //
                for (j = 0; j < n; j++)
                {
                    wa1[wa1Index + j] = -wa1[j];
                    wa2[wa2Index + j] = x[j] + wa1[j];
                    wa3[wa3Index + j] = diag[j] * wa1[wa1Index + j];
                }

                double pnorm = Helpers.enorm(n, wa3);
                delta = iter switch
                {
                    //
                    //  On the first iteration, adjust the initial step bound.
                    //
                    1 => Math.Min(delta, pnorm),
                    _ => delta
                };

                //
                //  Evaluate the function at X + P and calculate its norm.
                //
                iflag = 1;
                fcn(n, wa2, wa4, iflag, wa2Index, wa4Index);
                nfev += 1;
                switch (iflag)
                {
                case < 0:
                    info = iflag;
                    return(info);
                }

                double fnorm1 = Helpers.enorm(n, wa4);
                //
                //  Compute the scaled actual reduction.
                //
                double actred;
                if (fnorm1 < fnorm)
                {
                    actred = 1.0 - fnorm1 / fnorm * (fnorm1 / fnorm);
                }
                else
                {
                    actred = -1.0;
                }

                //
                //  Compute the scaled predicted reduction.
                //
                //  DO NOT ADJUST THIS LOOP, BECAUSE OF L.
                //
                l = 1;
                for (i = 1; i <= n; i++)
                {
                    sum = 0.0;
                    for (j = i; j <= n; j++)
                    {
                        sum += r[rIndex + (l - 1)] * wa1[wa1Index + (j - 1)];
                        l   += 1;
                    }

                    wa3[i - 1] = qtf[i - 1] + sum;
                }

                temp = Helpers.enorm(n, wa3);

                double prered;
                if (temp < fnorm)
                {
                    prered = 1.0 - temp / fnorm * (temp / fnorm);
                }
                else
                {
                    prered = 0.0;
                }

                double ratio = prered switch
                {
                    //
                    //  Compute the ratio of the actual to the predicted reduction.
                    //
                    > 0.0 => actred / prered,
                    _ => 0.0
                };

                switch (ratio)
                {
                //
                //  Update the step bound.
                //
                case < p1:
                    ncsuc   = 0;
                    ncfail += 1;
                    delta   = p5 * delta;
                    break;

                default:
                {
                    ncfail = 0;
                    ncsuc += 1;
                    if (p5 <= ratio || 1 < ncsuc)
                    {
                        delta = Math.Max(delta, pnorm / p5);
                    }

                    delta = Math.Abs(ratio - 1.0) switch
                    {
                        <= p1 => pnorm / p5,
                        _ => delta
                    };

                    break;
                }
                }
Esempio n. 2
0
    private static void qform_test()
    //****************************************************************************80
    //
    //  Purpose:
    //
    //    QFORM_TEST tests QFORM.
    //
    //  Licensing:
    //
    //    This code is distributed under the GNU LGPL license.
    //
    //  Modified:
    //
    //    02 January 2018
    //
    //  Author:
    //
    //    John Burkardt
    //
    {
        int       i;
        int       j;
        int       k;
        const int lda = 5;
        const int m   = 5;
        const int n   = 7;

        Console.WriteLine("");
        Console.WriteLine("QFORM_TEST:");
        Console.WriteLine("  QFORM constructs the Q factor explicitly");
        Console.WriteLine("  after the use of QRFAC.");
        //
        //  Set the matrix A.
        //
        double[] a = new double[m * n];

        int seed = 123456789;

        for (j = 0; j < n; j++)
        {
            for (i = 0; i < m; i++)
            {
                a[i + j * m] = UniformRNG.r8_uniform_01(ref seed);
            }
        }

        typeMethods.r8mat_print(m, n, a, "  Matrix A:");
        //
        //  Compute the QR factors.
        //
        const bool pivot = false;

        int[] ipivot = new int[n];
        int   lipvt  = n;

        double[] rdiag  = new double[n];
        double[] acnorm = new double[n];

        QRSolve.qrfac(m, n, ref a, lda, pivot, ref ipivot, ref lipvt, ref rdiag, ref acnorm);
        //
        //  Extract the R factor.
        //
        double[] r = new double[m * n];
        for (j = 0; j < n; j++)
        {
            for (i = 0; i < m; i++)
            {
                r[i + j * m] = 0.0;
            }
        }

        for (k = 0; k < Math.Min(m, n); k++)
        {
            r[k + k * m] = rdiag[k];
        }

        for (j = 0; j < n; j++)
        {
            for (i = 0; i < Math.Min(j, m); i++)
            {
                r[i + j * m] = a[i + j * m];
            }
        }

        typeMethods.r8mat_print(m, n, r, "  Matrix R:");
        //
        //  Call QRFORM to form the Q factor.
        //
        double[] q = new double[m * m];
        for (j = 0; j < m; j++)
        {
            for (i = 0; i < m; i++)
            {
                q[i + j * m] = 0.0;
            }
        }

        for (j = 0; j < Math.Min(m, n); j++)
        {
            for (i = 0; i < m; i++)
            {
                q[i + j * m] = a[i + j * m];
            }
        }

        QRSolve.qform(m, n, ref q, m);

        typeMethods.r8mat_print(m, m, q, "  Matrix Q:");
        //
        //  Compute Q*R.
        //
        double[] a2 = typeMethods.r8mat_mm_new(m, m, n, q, r);
        //
        //  Compare Q*R to A.
        //
        typeMethods.r8mat_print(m, n, a2, "  Matrix A2 = Q * R:");
    }