如何使用 Eigen,C++线性代数的模板库?

发布于 2024-09-10 00:10:58 字数 1492 浏览 1 评论 0原文

我有一个由矩阵组成的图像处理算法,我有自己的矩阵运算代码(乘法、逆...)。但我使用的处理器是ARM​​ Cortex-A8处理器,它有用于矢量化的NEON协处理器,因为矩阵运算是SIMD运算的理想情况,我要求编译器(-mfpu=neon -mfloat-abi=softfp)生成NEON我的代码的说明,但编译器无法这样做,然后我还尝试为矩阵运算编写自己的 NEON 内在函数代码,但我发现很难这样做。

因此,我想到利用 Eigen 库,它承诺矩阵运算的矢量化。因此,我立即下载了 Eigen C++ 库,并尝试按照他们的教程中给出的方式使用它,但不幸的是,当我运行他们的 示例程序

任何有使用 Eigen 经验的人,任何例子都会很有帮助吗?请帮助我如何去做。

帮助!

谢谢


,我的 Eigen 文件夹位于:/home/ubuntu/Documents/eigen 我在 Eclipse 的 C++ 项目的附加目录中设置了此路径。然后我运行以下程序(示例)-

#include <Eigen/Core>

// import most common Eigen types
USING_PART_OF_NAMESPACE_EIGEN

int main(int, char *[])
{
  Matrix3f m3;
  m3 << 1, 2, 3, 4, 5, 6, 7, 8, 9;
  Matrix4f m4 = Matrix4f::Identity();
  Vector4i v4(1, 2, 3, 4);

  std::cout << "m3\n" << m3 << "\nm4:\n"
    << m4 << "\nv4:\n" << v4 << std::endl;
}

我得到的错误 -

项目 Test_Eigen 的配置调试构建 ****

制作所有

构建文件:../main.cpp

调用:Sourcery G++ C++ 编译器

arm-none-linux-gnueabi- g++ -I/home/ubuntu/Documents/eigen -O0 -g3 -Wall -c -fmessage-length=0 -fcommon -MMD -MP -MF"main.d" -MT"main.d" -mcpu=cortex- a8 -marm -o"main.o"

"../main.cpp"

../main.cpp:6: 错误:预期的构造函数、析构函数或“int”之前的类型转换 make: *** [main.o] 错误 1

I have an image processing algorithm which makes of matrices, I have my own matrix operation codes (Multiplication, Inverse...) with me. But the processor I use is ARM Cortex-A8 processor, which has NEON co-processor for vectorization, as matrix operations are ideal cases for SIMD operations, I asked the compiler (-mfpu=neon -mfloat-abi=softfp) to generate NEON instructions for my code, but the compiler fails to do so and then I also attempted to write my own NEON intrinsics code for the Matrix operations, but I found it very hard to do so.

So, I thought of making use of Eigen library which promises vectorization of matrix operations. So I promptly downloaded the Eigen C++ library and tried using it as given in their tutorials but, unfortunately I get compilation errors when I run their example programs.

Anyone out there who has experience using Eigen, any examples will be really helpful? Kindly help me how to go about it.

Help!

Thanks


I have the Eigen folder at: /home/ubuntu/Documents/eigen I set this path in my Eclipse's C++ project's additional directories. Then I run the following program (Example)-

#include <Eigen/Core>

// import most common Eigen types
USING_PART_OF_NAMESPACE_EIGEN

int main(int, char *[])
{
  Matrix3f m3;
  m3 << 1, 2, 3, 4, 5, 6, 7, 8, 9;
  Matrix4f m4 = Matrix4f::Identity();
  Vector4i v4(1, 2, 3, 4);

  std::cout << "m3\n" << m3 << "\nm4:\n"
    << m4 << "\nv4:\n" << v4 << std::endl;
}

Errors I get -

Build of configuration Debug for project Test_Eigen ****

make all

Building file: ../main.cpp

Invoking: Sourcery G++ C++ Compiler

arm-none-linux-gnueabi-g++ -I/home/ubuntu/Documents/eigen -O0 -g3 -Wall -c -fmessage-length=0 -fcommon -MMD -MP -MF"main.d" -MT"main.d" -mcpu=cortex-a8 -marm -o"main.o"

"../main.cpp"

../main.cpp:6: error: expected constructor, destructor, or type conversion before 'int'
make: *** [main.o] Error 1

如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。

扫码二维码加入Web技术交流群

发布评论

需要 登录 才能够评论, 你可以免费 注册 一个本站的账号。

评论(2

笑梦风尘 2024-09-17 00:10:58

USING_PART_OF_NAMESPACE_EIGEN 宏已在 Eigen 3 中删除相反,只需使用

using namespace Eigen;

显然,该教程已经过时了。

The USING_PART_OF_NAMESPACE_EIGEN macro was removed in Eigen 3. Instead, simply use

using namespace Eigen;

Apparently, the tutorial is outdated.

擦肩而过的背影 2024-09-17 00:10:58

我正在使用 Ubuntu 17.04,这对我来说很有效
第一:
我在 eigen 官方网站下载了egien3.3.3。解压到一个名为 eigen 的目录中,cd 进入其中。
第二:
一条一条地运行下面的命令或者将它们做成xxx.sh文件一次运行。

#!/bin/bash
#eigen3 install
#from: http://eigen.tuxfamily.org/index.php?title=Main_Page
#download the package like eigen-eigen-67e894c6cd8f.tar.gz 

mkdir build
cd build
cmake -DEIGEN_TEST_NO_OPENGL=1 .. 
make 
sudo make install

最后:
进行测试

#include <eigen3/Eigen/Core>
#include <iostream>

// import most common Eigen types
//USING_PART_OF_NAMESPACE_EIGEN
using namespace Eigen;
using namespace std;
int main(int, char *[])
{
  Matrix3f m3;
  m3 << 1, 2, 3, 4, 5, 6, 7, 8, 9;
  Matrix4f m4 = Matrix4f::Identity();
  Vector4i v4(1, 2, 3, 4);

  cout << "m3\n" << m3 << "\nm4:\n"
    << m4 << "\nv4:\n" << v4 << endl;
}

注意:
要查找安装的结果,请参阅 /usr/local/include/eigen3/

如果有任何变化,请参阅 mytinx

I am using the Ubuntu 17.04, and this is work for me
First:
I download the egien3.3.3 at eigen official site. Extracted in a directory called eigen, cd into it.
Second:
run the command bellow one by one or make them a xxx.sh file to run at a time.

#!/bin/bash
#eigen3 install
#from: http://eigen.tuxfamily.org/index.php?title=Main_Page
#download the package like eigen-eigen-67e894c6cd8f.tar.gz 

mkdir build
cd build
cmake -DEIGEN_TEST_NO_OPENGL=1 .. 
make 
sudo make install

Finally:
make a test

#include <eigen3/Eigen/Core>
#include <iostream>

// import most common Eigen types
//USING_PART_OF_NAMESPACE_EIGEN
using namespace Eigen;
using namespace std;
int main(int, char *[])
{
  Matrix3f m3;
  m3 << 1, 2, 3, 4, 5, 6, 7, 8, 9;
  Matrix4f m4 = Matrix4f::Identity();
  Vector4i v4(1, 2, 3, 4);

  cout << "m3\n" << m3 << "\nm4:\n"
    << m4 << "\nv4:\n" << v4 << endl;
}

Note:
To find your installed results, please see /usr/local/include/eigen3/

If any thing change, please see mytinx

~没有更多了~
我们使用 Cookies 和其他技术来定制您的体验包括您的登录状态等。通过阅读我们的 隐私政策 了解更多相关信息。 单击 接受 或继续使用网站,即表示您同意使用 Cookies 和您的相关数据。
原文