本征是最好的！它比 boost::ublas 好得多，你可以写 C = A*B 而不是 ublas 中的 C = prod(A,B) ，我测试过它的速度比 ublas 快得多。
eigen is the best one! It's much better than boost::ublas, you can write C = A*B instead of C = prod(A,B) as in ublas and I have tested the speed it's much faster than ublas.
Eigen3 now has a sparse matrix class, along with interfaces to several popular sparse matrix libraries. If you need to calculate the pseudo inverse to solve a least squares system, you can instead use Cholesky decomposition directly on the normal equations.
if row > col，则伪逆（最小二乘解）可以计算为： M_PseudoInv = (MTM)-1MT
如果行 T col 那么伪逆（最小范数解）可以计算为： M_PseudoInv = MT (MMT)-1
NewMat11 is a good, easy to use & fairly lightweight matrix library for high level matrix operations (Eigensystems, SVD, QR, LU, inverse). You can easily declare & construct a matrix from an array as:
Matrix M(numRows,numCols);
M << array;
Access matrix elements as:
M.element(i,j);
Take transpose:
M.t();
Take inverse:
M.i();
Multiply with just *:
M*M
And can perform other operations like: SVD, QR, LU etc. fairly easily.
Providing the pseudo-inverse (Moore-Penrose inverse) operation explicitly is not mandatory for a library, because you can calculate it easily by using the transpose & inverse operations as below:
if row > col, then the pseudo-inverse (least-squares solution) can be calculated as: M_PseudoInv = (MTM)-1MT
if row < col then the pseudo-inverse (least-norm solution) can be calculated as: M_PseudoInv = MT (MMT)-1
发布评论
评论（3）
本征是最好的！它比 boost::ublas 好得多，你可以写 C = A*B 而不是 ublas 中的 C = prod(A,B) ，我测试过它的速度比 ublas 快得多。
eigen is the best one! It's much better than boost::ublas, you can write C = A*B instead of C = prod(A,B) as in ublas and I have tested the speed it's much faster than ublas.
Eigen3 现在有一个稀疏矩阵类，以及几个流行的稀疏矩阵库的接口。如果您需要计算伪逆来求解最小二乘系统，则可以使用 直接在正规方程上进行 Cholesky 分解。
Eigen3 now has a sparse matrix class, along with interfaces to several popular sparse matrix libraries. If you need to calculate the pseudo inverse to solve a least squares system, you can instead use Cholesky decomposition directly on the normal equations.
NewMat11 是一个很好的、易于使用且易于使用的工具。用于高级矩阵运算（Eigensystems、SVD、QR、LU、逆）的相当轻量级的矩阵库。您可以轻松声明&从数组构建矩阵如下：
访问矩阵元素如下：
转置：
取逆：
仅与 * 相乘：
并且可以相当轻松地执行其他操作，例如：SVD、QR、LU 等。
对于库来说，显式提供伪逆（Moore-Penrose 逆）运算并不是强制性的，因为您可以通过使用转置和逆运算轻松计算它。逆运算如下：
if row > col，则伪逆（最小二乘解）可以计算为：
M_PseudoInv = (MTM)-1MT
如果行 T col 那么伪逆（最小范数解）可以计算为：
M_PseudoInv = MT (MMT)-1
NewMat11 is a good, easy to use & fairly lightweight matrix library for high level matrix operations (Eigensystems, SVD, QR, LU, inverse). You can easily declare & construct a matrix from an array as:
Access matrix elements as:
Take transpose:
Take inverse:
Multiply with just *:
And can perform other operations like: SVD, QR, LU etc. fairly easily.
Providing the pseudo-inverse (Moore-Penrose inverse) operation explicitly is not mandatory for a library, because you can calculate it easily by using the transpose & inverse operations as below:
if row > col, then the pseudo-inverse (least-squares solution) can be calculated as:
M_PseudoInv = (MTM)-1MT
if row < col then the pseudo-inverse (least-norm solution) can be calculated as:
M_PseudoInv = MT (MMT)-1