pylibsparseir 0.1.0__cp312-cp312-macosx_15_0_arm64.whl → 0.6.0__cp312-cp312-macosx_15_0_arm64.whl

include/eigen3/unsupported/Eigen/src/NonLinearOptimization/chkder.h

@@ -0,0 +1,66 @@
+#define chkder_log10e 0.43429448190325182765
+#define chkder_factor 100.
+
+namespace Eigen { 
+
+namespace internal {
+
+template<typename Scalar>
+void chkder(
+        const Matrix< Scalar, Dynamic, 1 >  &x,
+        const Matrix< Scalar, Dynamic, 1 >  &fvec,
+        const Matrix< Scalar, Dynamic, Dynamic > &fjac,
+        Matrix< Scalar, Dynamic, 1 >  &xp,
+        const Matrix< Scalar, Dynamic, 1 >  &fvecp,
+        int mode,
+        Matrix< Scalar, Dynamic, 1 >  &err
+        )
+{
+    using std::sqrt;
+    using std::abs;
+    using std::log;
+    
+    typedef DenseIndex Index;
+
+    const Scalar eps = sqrt(NumTraits<Scalar>::epsilon());
+    const Scalar epsf = chkder_factor * NumTraits<Scalar>::epsilon();
+    const Scalar epslog = chkder_log10e * log(eps);
+    Scalar temp;
+
+    const Index m = fvec.size(), n = x.size();
+
+    if (mode != 2) {
+        /* mode = 1. */
+        xp.resize(n);
+        for (Index j = 0; j < n; ++j) {
+            temp = eps * abs(x[j]);
+            if (temp == 0.)
+                temp = eps;
+            xp[j] = x[j] + temp;
+        }
+    }
+    else {
+        /* mode = 2. */
+        err.setZero(m); 
+        for (Index j = 0; j < n; ++j) {
+            temp = abs(x[j]);
+            if (temp == 0.)
+                temp = 1.;
+            err += temp * fjac.col(j);
+        }
+        for (Index i = 0; i < m; ++i) {
+            temp = 1.;
+            if (fvec[i] != 0. && fvecp[i] != 0. && abs(fvecp[i] - fvec[i]) >= epsf * abs(fvec[i]))
+                temp = eps * abs((fvecp[i] - fvec[i]) / eps - err[i]) / (abs(fvec[i]) + abs(fvecp[i]));
+            err[i] = 1.;
+            if (temp > NumTraits<Scalar>::epsilon() && temp < eps)
+                err[i] = (chkder_log10e * log(temp) - epslog) / epslog;
+            if (temp >= eps)
+                err[i] = 0.;
+        }
+    }
+}
+
+} // end namespace internal
+
+} // end namespace Eigen

include/eigen3/unsupported/Eigen/src/NonLinearOptimization/covar.h

@@ -0,0 +1,70 @@
+namespace Eigen { 
+
+namespace internal {
+
+template <typename Scalar>
+void covar(
+        Matrix< Scalar, Dynamic, Dynamic > &r,
+        const VectorXi &ipvt,
+        Scalar tol = std::sqrt(NumTraits<Scalar>::epsilon()) )
+{
+    using std::abs;
+    typedef DenseIndex Index;
+
+    /* Local variables */
+    Index i, j, k, l, ii, jj;
+    bool sing;
+    Scalar temp;
+
+    /* Function Body */
+    const Index n = r.cols();
+    const Scalar tolr = tol * abs(r(0,0));
+    Matrix< Scalar, Dynamic, 1 > wa(n);
+    eigen_assert(ipvt.size()==n);
+
+    /* form the inverse of r in the full upper triangle of r. */
+    l = -1;
+    for (k = 0; k < n; ++k)
+        if (abs(r(k,k)) > tolr) {
+            r(k,k) = 1. / r(k,k);
+            for (j = 0; j <= k-1; ++j) {
+                temp = r(k,k) * r(j,k);
+                r(j,k) = 0.;
+                r.col(k).head(j+1) -= r.col(j).head(j+1) * temp;
+            }
+            l = k;
+        }
+
+    /* form the full upper triangle of the inverse of (r transpose)*r */
+    /* in the full upper triangle of r. */
+    for (k = 0; k <= l; ++k) {
+        for (j = 0; j <= k-1; ++j)
+            r.col(j).head(j+1) += r.col(k).head(j+1) * r(j,k);
+        r.col(k).head(k+1) *= r(k,k);
+    }
+
+    /* form the full lower triangle of the covariance matrix */
+    /* in the strict lower triangle of r and in wa. */
+    for (j = 0; j < n; ++j) {
+        jj = ipvt[j];
+        sing = j > l;
+        for (i = 0; i <= j; ++i) {
+            if (sing)
+                r(i,j) = 0.;
+            ii = ipvt[i];
+            if (ii > jj)
+                r(ii,jj) = r(i,j);
+            if (ii < jj)
+                r(jj,ii) = r(i,j);
+        }
+        wa[jj] = r(j,j);
+    }
+
+    /* symmetrize the covariance matrix in r. */
+    r.topLeftCorner(n,n).template triangularView<StrictlyUpper>() = r.topLeftCorner(n,n).transpose();
+    r.diagonal() = wa;
+}
+
+} // end namespace internal
+
+} // end namespace Eigen

include/eigen3/unsupported/Eigen/src/NonLinearOptimization/dogleg.h

@@ -0,0 +1,107 @@
+namespace Eigen { 
+
+namespace internal {
+
+template <typename Scalar>
+void dogleg(
+        const Matrix< Scalar, Dynamic, Dynamic >  &qrfac,
+        const Matrix< Scalar, Dynamic, 1 >  &diag,
+        const Matrix< Scalar, Dynamic, 1 >  &qtb,
+        Scalar delta,
+        Matrix< Scalar, Dynamic, 1 >  &x)
+{
+    using std::abs;
+    using std::sqrt;
+    
+    typedef DenseIndex Index;
+
+    /* Local variables */
+    Index i, j;
+    Scalar sum, temp, alpha, bnorm;
+    Scalar gnorm, qnorm;
+    Scalar sgnorm;
+
+    /* Function Body */
+    const Scalar epsmch = NumTraits<Scalar>::epsilon();
+    const Index n = qrfac.cols();
+    eigen_assert(n==qtb.size());
+    eigen_assert(n==x.size());
+    eigen_assert(n==diag.size());
+    Matrix< Scalar, Dynamic, 1 >  wa1(n), wa2(n);
+
+    /* first, calculate the gauss-newton direction. */
+    for (j = n-1; j >=0; --j) {
+        temp = qrfac(j,j);
+        if (temp == 0.) {
+            temp = epsmch * qrfac.col(j).head(j+1).maxCoeff();
+            if (temp == 0.)
+                temp = epsmch;
+        }
+        if (j==n-1)
+            x[j] = qtb[j] / temp;
+        else
+            x[j] = (qtb[j] - qrfac.row(j).tail(n-j-1).dot(x.tail(n-j-1))) / temp;
+    }
+
+    /* test whether the gauss-newton direction is acceptable. */
+    qnorm = diag.cwiseProduct(x).stableNorm();
+    if (qnorm <= delta)
+        return;
+
+    // TODO : this path is not tested by Eigen unit tests
+
+    /* the gauss-newton direction is not acceptable. */
+    /* next, calculate the scaled gradient direction. */
+
+    wa1.fill(0.);
+    for (j = 0; j < n; ++j) {
+        wa1.tail(n-j) += qrfac.row(j).tail(n-j) * qtb[j];
+        wa1[j] /= diag[j];
+    }
+
+    /* calculate the norm of the scaled gradient and test for */
+    /* the special case in which the scaled gradient is zero. */
+    gnorm = wa1.stableNorm();
+    sgnorm = 0.;
+    alpha = delta / qnorm;
+    if (gnorm == 0.)
+        goto algo_end;
+
+    /* calculate the point along the scaled gradient */
+    /* at which the quadratic is minimized. */
+    wa1.array() /= (diag*gnorm).array();
+    // TODO : once unit tests cover this part,:
+    // wa2 = qrfac.template triangularView<Upper>() * wa1;
+    for (j = 0; j < n; ++j) {
+        sum = 0.;
+        for (i = j; i < n; ++i) {
+            sum += qrfac(j,i) * wa1[i];
+        }
+        wa2[j] = sum;
+    }
+    temp = wa2.stableNorm();
+    sgnorm = gnorm / temp / temp;
+
+    /* test whether the scaled gradient direction is acceptable. */
+    alpha = 0.;
+    if (sgnorm >= delta)
+        goto algo_end;
+
+    /* the scaled gradient direction is not acceptable. */
+    /* finally, calculate the point along the dogleg */
+    /* at which the quadratic is minimized. */
+    bnorm = qtb.stableNorm();
+    temp = bnorm / gnorm * (bnorm / qnorm) * (sgnorm / delta);
+    temp = temp - delta / qnorm * numext::abs2(sgnorm / delta) + sqrt(numext::abs2(temp - delta / qnorm) + (1.-numext::abs2(delta / qnorm)) * (1.-numext::abs2(sgnorm / delta)));
+    alpha = delta / qnorm * (1. - numext::abs2(sgnorm / delta)) / temp;
+algo_end:
+
+    /* form appropriate convex combination of the gauss-newton */
+    /* direction and the scaled gradient direction. */
+    temp = (1.-alpha) * (std::min)(sgnorm,delta);
+    x = temp * wa1 + alpha * x;
+}
+
+} // end namespace internal
+
+} // end namespace Eigen

include/eigen3/unsupported/Eigen/src/NonLinearOptimization/fdjac1.h

@@ -0,0 +1,79 @@
+namespace Eigen { 
+
+namespace internal {
+
+template<typename FunctorType, typename Scalar>
+DenseIndex fdjac1(
+        const FunctorType &Functor,
+        Matrix< Scalar, Dynamic, 1 >  &x,
+        Matrix< Scalar, Dynamic, 1 >  &fvec,
+        Matrix< Scalar, Dynamic, Dynamic > &fjac,
+        DenseIndex ml, DenseIndex mu,
+        Scalar epsfcn)
+{
+    using std::sqrt;
+    using std::abs;
+    
+    typedef DenseIndex Index;
+
+    /* Local variables */
+    Scalar h;
+    Index j, k;
+    Scalar eps, temp;
+    Index msum;
+    int iflag;
+    Index start, length;
+
+    /* Function Body */
+    const Scalar epsmch = NumTraits<Scalar>::epsilon();
+    const Index n = x.size();
+    eigen_assert(fvec.size()==n);
+    Matrix< Scalar, Dynamic, 1 >  wa1(n);
+    Matrix< Scalar, Dynamic, 1 >  wa2(n);
+
+    eps = sqrt((std::max)(epsfcn,epsmch));
+    msum = ml + mu + 1;
+    if (msum >= n) {
+        /* computation of dense approximate jacobian. */
+        for (j = 0; j < n; ++j) {
+            temp = x[j];
+            h = eps * abs(temp);
+            if (h == 0.)
+                h = eps;
+            x[j] = temp + h;
+            iflag = Functor(x, wa1);
+            if (iflag < 0)
+                return iflag;
+            x[j] = temp;
+            fjac.col(j) = (wa1-fvec)/h;
+        }
+
+    }else {
+        /* computation of banded approximate jacobian. */
+        for (k = 0; k < msum; ++k) {
+            for (j = k; (msum<0) ? (j>n): (j<n); j += msum) {
+                wa2[j] = x[j];
+                h = eps * abs(wa2[j]);
+                if (h == 0.) h = eps;
+                x[j] = wa2[j] + h;
+            }
+            iflag = Functor(x, wa1);
+            if (iflag < 0)
+                return iflag;
+            for (j = k; (msum<0) ? (j>n): (j<n); j += msum) {
+                x[j] = wa2[j];
+                h = eps * abs(wa2[j]);
+                if (h == 0.) h = eps;
+                fjac.col(j).setZero();
+                start = std::max<Index>(0,j-mu);
+                length = (std::min)(n-1, j+ml) - start + 1;
+                fjac.col(j).segment(start, length) = ( wa1.segment(start, length)-fvec.segment(start, length))/h;
+            }
+        }
+    }
+    return 0;
+}
+
+} // end namespace internal
+
+} // end namespace Eigen

include/eigen3/unsupported/Eigen/src/NonLinearOptimization/lmpar.h

@@ -0,0 +1,298 @@
+namespace Eigen { 
+
+namespace internal {
+
+template <typename Scalar>
+void lmpar(
+        Matrix< Scalar, Dynamic, Dynamic > &r,
+        const VectorXi &ipvt,
+        const Matrix< Scalar, Dynamic, 1 >  &diag,
+        const Matrix< Scalar, Dynamic, 1 >  &qtb,
+        Scalar delta,
+        Scalar &par,
+        Matrix< Scalar, Dynamic, 1 >  &x)
+{
+    using std::abs;
+    using std::sqrt;
+    typedef DenseIndex Index;
+
+    /* Local variables */
+    Index i, j, l;
+    Scalar fp;
+    Scalar parc, parl;
+    Index iter;
+    Scalar temp, paru;
+    Scalar gnorm;
+    Scalar dxnorm;
+
+
+    /* Function Body */
+    const Scalar dwarf = (std::numeric_limits<Scalar>::min)();
+    const Index n = r.cols();
+    eigen_assert(n==diag.size());
+    eigen_assert(n==qtb.size());
+    eigen_assert(n==x.size());
+
+    Matrix< Scalar, Dynamic, 1 >  wa1, wa2;
+
+    /* compute and store in x the gauss-newton direction. if the */
+    /* jacobian is rank-deficient, obtain a least squares solution. */
+    Index nsing = n-1;
+    wa1 = qtb;
+    for (j = 0; j < n; ++j) {
+        if (r(j,j) == 0. && nsing == n-1)
+            nsing = j - 1;
+        if (nsing < n-1)
+            wa1[j] = 0.;
+    }
+    for (j = nsing; j>=0; --j) {
+        wa1[j] /= r(j,j);
+        temp = wa1[j];
+        for (i = 0; i < j ; ++i)
+            wa1[i] -= r(i,j) * temp;
+    }
+
+    for (j = 0; j < n; ++j)
+        x[ipvt[j]] = wa1[j];
+
+    /* initialize the iteration counter. */
+    /* evaluate the function at the origin, and test */
+    /* for acceptance of the gauss-newton direction. */
+    iter = 0;
+    wa2 = diag.cwiseProduct(x);
+    dxnorm = wa2.blueNorm();
+    fp = dxnorm - delta;
+    if (fp <= Scalar(0.1) * delta) {
+        par = 0;
+        return;
+    }
+
+    /* if the jacobian is not rank deficient, the newton */
+    /* step provides a lower bound, parl, for the zero of */
+    /* the function. otherwise set this bound to zero. */
+    parl = 0.;
+    if (nsing >= n-1) {
+        for (j = 0; j < n; ++j) {
+            l = ipvt[j];
+            wa1[j] = diag[l] * (wa2[l] / dxnorm);
+        }
+        // it's actually a triangularView.solveInplace(), though in a weird
+        // way:
+        for (j = 0; j < n; ++j) {
+            Scalar sum = 0.;
+            for (i = 0; i < j; ++i)
+                sum += r(i,j) * wa1[i];
+            wa1[j] = (wa1[j] - sum) / r(j,j);
+        }
+        temp = wa1.blueNorm();
+        parl = fp / delta / temp / temp;
+    }
+
+    /* calculate an upper bound, paru, for the zero of the function. */
+    for (j = 0; j < n; ++j)
+        wa1[j] = r.col(j).head(j+1).dot(qtb.head(j+1)) / diag[ipvt[j]];
+
+    gnorm = wa1.stableNorm();
+    paru = gnorm / delta;
+    if (paru == 0.)
+        paru = dwarf / (std::min)(delta,Scalar(0.1));
+
+    /* if the input par lies outside of the interval (parl,paru), */
+    /* set par to the closer endpoint. */
+    par = (std::max)(par,parl);
+    par = (std::min)(par,paru);
+    if (par == 0.)
+        par = gnorm / dxnorm;
+
+    /* beginning of an iteration. */
+    while (true) {
+        ++iter;
+
+        /* evaluate the function at the current value of par. */
+        if (par == 0.)
+            par = (std::max)(dwarf,Scalar(.001) * paru); /* Computing MAX */
+        wa1 = sqrt(par)* diag;
+
+        Matrix< Scalar, Dynamic, 1 > sdiag(n);
+        qrsolv<Scalar>(r, ipvt, wa1, qtb, x, sdiag);
+
+        wa2 = diag.cwiseProduct(x);
+        dxnorm = wa2.blueNorm();
+        temp = fp;
+        fp = dxnorm - delta;
+
+        /* if the function is small enough, accept the current value */
+        /* of par. also test for the exceptional cases where parl */
+        /* is zero or the number of iterations has reached 10. */
+        if (abs(fp) <= Scalar(0.1) * delta || (parl == 0. && fp <= temp && temp < 0.) || iter == 10)
+            break;
+
+        /* compute the newton correction. */
+        for (j = 0; j < n; ++j) {
+            l = ipvt[j];
+            wa1[j] = diag[l] * (wa2[l] / dxnorm);
+        }
+        for (j = 0; j < n; ++j) {
+            wa1[j] /= sdiag[j];
+            temp = wa1[j];
+            for (i = j+1; i < n; ++i)
+                wa1[i] -= r(i,j) * temp;
+        }
+        temp = wa1.blueNorm();
+        parc = fp / delta / temp / temp;
+
+        /* depending on the sign of the function, update parl or paru. */
+        if (fp > 0.)
+            parl = (std::max)(parl,par);
+        if (fp < 0.)
+            paru = (std::min)(paru,par);
+
+        /* compute an improved estimate for par. */
+        /* Computing MAX */
+        par = (std::max)(parl,par+parc);
+
+        /* end of an iteration. */
+    }
+
+    /* termination. */
+    if (iter == 0)
+        par = 0.;
+    return;
+}
+
+template <typename Scalar>
+void lmpar2(
+        const ColPivHouseholderQR<Matrix< Scalar, Dynamic, Dynamic> > &qr,
+        const Matrix< Scalar, Dynamic, 1 >  &diag,
+        const Matrix< Scalar, Dynamic, 1 >  &qtb,
+        Scalar delta,
+        Scalar &par,
+        Matrix< Scalar, Dynamic, 1 >  &x)
+
+{
+    using std::sqrt;
+    using std::abs;
+    typedef DenseIndex Index;
+
+    /* Local variables */
+    Index j;
+    Scalar fp;
+    Scalar parc, parl;
+    Index iter;
+    Scalar temp, paru;
+    Scalar gnorm;
+    Scalar dxnorm;
+
+
+    /* Function Body */
+    const Scalar dwarf = (std::numeric_limits<Scalar>::min)();
+    const Index n = qr.matrixQR().cols();
+    eigen_assert(n==diag.size());
+    eigen_assert(n==qtb.size());
+
+    Matrix< Scalar, Dynamic, 1 >  wa1, wa2;
+
+    /* compute and store in x the gauss-newton direction. if the */
+    /* jacobian is rank-deficient, obtain a least squares solution. */
+
+//    const Index rank = qr.nonzeroPivots(); // exactly double(0.)
+    const Index rank = qr.rank(); // use a threshold
+    wa1 = qtb;
+    wa1.tail(n-rank).setZero();
+    qr.matrixQR().topLeftCorner(rank, rank).template triangularView<Upper>().solveInPlace(wa1.head(rank));
+
+    x = qr.colsPermutation()*wa1;
+
+    /* initialize the iteration counter. */
+    /* evaluate the function at the origin, and test */
+    /* for acceptance of the gauss-newton direction. */
+    iter = 0;
+    wa2 = diag.cwiseProduct(x);
+    dxnorm = wa2.blueNorm();
+    fp = dxnorm - delta;
+    if (fp <= Scalar(0.1) * delta) {
+        par = 0;
+        return;
+    }
+
+    /* if the jacobian is not rank deficient, the newton */
+    /* step provides a lower bound, parl, for the zero of */
+    /* the function. otherwise set this bound to zero. */
+    parl = 0.;
+    if (rank==n) {
+        wa1 = qr.colsPermutation().inverse() *  diag.cwiseProduct(wa2)/dxnorm;
+        qr.matrixQR().topLeftCorner(n, n).transpose().template triangularView<Lower>().solveInPlace(wa1);
+        temp = wa1.blueNorm();
+        parl = fp / delta / temp / temp;
+    }
+
+    /* calculate an upper bound, paru, for the zero of the function. */
+    for (j = 0; j < n; ++j)
+        wa1[j] = qr.matrixQR().col(j).head(j+1).dot(qtb.head(j+1)) / diag[qr.colsPermutation().indices()(j)];
+
+    gnorm = wa1.stableNorm();
+    paru = gnorm / delta;
+    if (paru == 0.)
+        paru = dwarf / (std::min)(delta,Scalar(0.1));
+
+    /* if the input par lies outside of the interval (parl,paru), */
+    /* set par to the closer endpoint. */
+    par = (std::max)(par,parl);
+    par = (std::min)(par,paru);
+    if (par == 0.)
+        par = gnorm / dxnorm;
+
+    /* beginning of an iteration. */
+    Matrix< Scalar, Dynamic, Dynamic > s = qr.matrixQR();
+    while (true) {
+        ++iter;
+
+        /* evaluate the function at the current value of par. */
+        if (par == 0.)
+            par = (std::max)(dwarf,Scalar(.001) * paru); /* Computing MAX */
+        wa1 = sqrt(par)* diag;
+
+        Matrix< Scalar, Dynamic, 1 > sdiag(n);
+        qrsolv<Scalar>(s, qr.colsPermutation().indices(), wa1, qtb, x, sdiag);
+
+        wa2 = diag.cwiseProduct(x);
+        dxnorm = wa2.blueNorm();
+        temp = fp;
+        fp = dxnorm - delta;
+
+        /* if the function is small enough, accept the current value */
+        /* of par. also test for the exceptional cases where parl */
+        /* is zero or the number of iterations has reached 10. */
+        if (abs(fp) <= Scalar(0.1) * delta || (parl == 0. && fp <= temp && temp < 0.) || iter == 10)
+            break;
+
+        /* compute the newton correction. */
+        wa1 = qr.colsPermutation().inverse() * diag.cwiseProduct(wa2/dxnorm);
+        // we could almost use this here, but the diagonal is outside qr, in sdiag[]
+        // qr.matrixQR().topLeftCorner(n, n).transpose().template triangularView<Lower>().solveInPlace(wa1);
+        for (j = 0; j < n; ++j) {
+            wa1[j] /= sdiag[j];
+            temp = wa1[j];
+            for (Index i = j+1; i < n; ++i)
+                wa1[i] -= s(i,j) * temp;
+        }
+        temp = wa1.blueNorm();
+        parc = fp / delta / temp / temp;
+
+        /* depending on the sign of the function, update parl or paru. */
+        if (fp > 0.)
+            parl = (std::max)(parl,par);
+        if (fp < 0.)
+            paru = (std::min)(paru,par);
+
+        /* compute an improved estimate for par. */
+        par = (std::max)(parl,par+parc);
+    }
+    if (iter == 0)
+        par = 0.;
+    return;
+}
+
+} // end namespace internal
+
+} // end namespace Eigen