Compiler Errors when using Eigen Library with MSVC2015 - c++

Question

I need to use matricies for a program I am writing, so I downloaded and installed Eigen, however, upon compiling my project my error log is absolutely flooded with errors, all of which seem to be coming from the Eigen files themselves, here is my error log:
1>------ Build started: Project: ailib, Configuration: Debug Win32 ------
1> NeuralNet.cpp
1>c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\functors.h(52): error C2039: 'ReturnType': is not a member of 'Eigen::internal::scalar_product_traits<LhsScalar,RhsScalar>'
1> with
1> [
1> LhsScalar=float,
1> RhsScalar=Eigen::internal::result_of<Eigen::internal::scalar_cmp_op<float,Eigen::internal::cmp_LT> (float,float)>::type
1> ]
1> c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\util\meta.h(161): note: see reference to class template instantiation 'Eigen::internal::scalar_product_op<float,Eigen::internal::result_of<BinaryOp (float,float)>::type>' being compiled
1> with
1> [
1> BinaryOp=Eigen::internal::scalar_cmp_op<float,Eigen::internal::cmp_LT>
1> ]
1> c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\cwisebinaryop.h(58): note: see reference to class template instantiation 'Eigen::internal::result_of<BinaryOp (float,bool)>' being compiled
1> with
1> [
1> BinaryOp=Eigen::internal::scalar_product_op<float,Eigen::internal::result_of<Eigen::internal::scalar_cmp_op<float,Eigen::internal::cmp_LT> (float,float)>::type>
1> ]
1> c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\cwisebinaryop.h(165): note: see reference to class template instantiation 'Eigen::internal::traits<Derived>' being compiled
1> with
1> [
1> Derived=Eigen::CwiseBinaryOp<Eigen::internal::scalar_product_op<float,Eigen::internal::result_of<Eigen::internal::scalar_cmp_op<float,Eigen::internal::cmp_LT> (float,float)>::type>,const Eigen::ArrayWrapper<Eigen::Matrix<float,-1,1,0,-1,1>>,const Eigen::CwiseBinaryOp<Eigen::internal::scalar_cmp_op<float,Eigen::internal::cmp_LT>,const Eigen::CwiseNullaryOp<Eigen::internal::scalar_constant_op<float>,Eigen::Array<float,-1,1,0,-1,1>>,const Eigen::ArrayWrapper<Eigen::Matrix<float,-1,1,0,-1,1>>>>
1> ]
1> c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\cwisebinaryop.h(112): note: see reference to class template instantiation 'Eigen::CwiseBinaryOpImpl<BinaryOp,Lhs,Rhs,Eigen::Dense>' being compiled
1> with
1> [
1> BinaryOp=Eigen::internal::scalar_product_op<float,Eigen::internal::result_of<Eigen::internal::scalar_cmp_op<float,Eigen::internal::cmp_LT> (float,float)>::type>,
1> Lhs=Eigen::ArrayWrapper<Eigen::Matrix<float,-1,1,0,-1,1>>,
1> Rhs=Eigen::CwiseBinaryOp<Eigen::internal::scalar_cmp_op<float,Eigen::internal::cmp_LT>,const Eigen::CwiseNullaryOp<Eigen::internal::scalar_constant_op<float>,Eigen::Array<float,-1,1,0,-1,1>>,const Eigen::ArrayWrapper<Eigen::Matrix<float,-1,1,0,-1,1>>>
1> ]
1> c:\users\rick simone\documents\aioli\ailib\ailib\neuralnet.cpp(51): note: see reference to class template instantiation 'Eigen::CwiseBinaryOp<Eigen::internal::scalar_product_op<float,Eigen::internal::result_of<BinaryOp (float,float)>::type>,const Derived,const Eigen::CwiseBinaryOp<BinaryOp,const Eigen::CwiseNullaryOp<Eigen::internal::scalar_constant_op<float>,Eigen::Array<float,-1,1,0,-1,1>>,const Derived>>' being compiled
1> with
1> [
1> BinaryOp=Eigen::internal::scalar_cmp_op<float,Eigen::internal::cmp_LT>,
1> Derived=Eigen::ArrayWrapper<Eigen::Matrix<float,-1,1,0,-1,1>>
1> ]
1>c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\functors.h(52): error C3646: 'result_type': unknown override specifier
1>c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\functors.h(52): error C4430: missing type specifier - int assumed. Note: C++ does not support default-int
1>c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\functors.h(54): error C4430: missing type specifier - int assumed. Note: C++ does not support default-int
1>c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\functors.h(54): error C2143: syntax error: missing ';' before '('
1>c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\functors.h(54): error C2433: 'Eigen::internal::scalar_product_op<float,Eigen::internal::result_of<BinaryOp (float,float)>::type>::result_type': '__forceinline' not permitted on data declarations
1> with
1> [
1> BinaryOp=Eigen::internal::scalar_cmp_op<float,Eigen::internal::cmp_LT>
1> ]
1>c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\functors.h(59): error C2327: 'Eigen::internal::scalar_product_op<float,Eigen::internal::result_of<BinaryOp (float,float)>::type>::result_type': is not a type name, static, or enumerator
1> with
1> [
1> BinaryOp=Eigen::internal::scalar_cmp_op<float,Eigen::internal::cmp_LT>
1> ]
1>c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\functors.h(59): error C4430: missing type specifier - int assumed. Note: C++ does not support default-int
1>c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\functors.h(59): error C2433: 'result_type': '__forceinline' not permitted on data declarations
1>c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\functors.h(59): error C2365: 'Eigen::internal::scalar_product_op<float,Eigen::internal::result_of<BinaryOp (float,float)>::type>::result_type': redefinition; previous definition was 'data member'
1> with
1> [
1> BinaryOp=Eigen::internal::scalar_cmp_op<float,Eigen::internal::cmp_LT>
1> ]
1> c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\functors.h(54): note: see declaration of 'Eigen::internal::scalar_product_op<float,Eigen::internal::result_of<BinaryOp (float,float)>::type>::result_type'
1> with
1> [
1> BinaryOp=Eigen::internal::scalar_cmp_op<float,Eigen::internal::cmp_LT>
1> ]
1>c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\functors.h(59): error C2061: syntax error: identifier 'predux'
1>c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\functors.h(61): error C2143: syntax error: missing ';' before '}'
1>c:\users\rick simone\documents\aioli\ailib\eigen\eigen\src\core\functors.h(61): error C2238: unexpected token(s) preceding ';'
The problem is, when I go to the files and lines indicated by the errors, they are in the Eigen source code (Which I haven't modified or touched), the snippet where all the errors come from is as follows:
template<typename LhsScalar,typename RhsScalar> struct scalar_product_op {
enum {
// TODO vectorize mixed product
Vectorizable = is_same<LhsScalar,RhsScalar>::value && packet_traits<LhsScalar>::HasMul && packet_traits<RhsScalar>::HasMul
};
typedef typename scalar_product_traits<LhsScalar,RhsScalar>::ReturnType result_type;
EIGEN_EMPTY_STRUCT_CTOR(scalar_product_op)
EIGEN_STRONG_INLINE const result_type operator() (const LhsScalar& a, const RhsScalar& b) const { return a * b; }
template<typename Packet>
EIGEN_STRONG_INLINE const Packet packetOp(const Packet& a, const Packet& b) const
{ return internal::pmul(a,b); }
template<typename Packet>
EIGEN_STRONG_INLINE const result_type predux(const Packet& a) const
{ return internal::predux_mul(a); }
};
Includes used:
#include <Eigen\Core>
#include <Eigen\Dense>
Line 51 NeuralNet.cpp: err = errIn.array() * (m_layerOut.array() > 0.f);
Types of Variables: err, errIn and m_layerOut are all Eigen::Matrix<float, Eigen::Dynamic, 1>
Thanks in advance!

Answer 1

As posted, your question is very hard to answer. There are a bunch of errors in template code which is determined by your usage at compile time. For future reference, it would make it much easier on both you and those attempting to answer if you tried to boil it down to a MCVE (also worth reading, How do I ask a good question?).
So, after reading the above links, you would rephrase your question to read something like:
I've wrote some code that produces a bunch of errors and I can't figure out why. Here is the minimal code that I've managed to create that reproduces this problem:
#include "Eigen/Core"
#include <iostream>
int main()
{
size_t sz = 64;
Eigen::Matrix<float, Eigen::Dynamic, 1> err, errIn, m_layerOut;
errIn.setRandom(sz);
m_layerOut.setRandom(sz);
err = errIn.array() * (m_layerOut.array() > 0.f);
std::cout << err;
return 0;
}
I tried to compile using g++ on MinGW (version 9.187 beta, 256bit) using g++ SO.cpp -I .. I have the same issues when using VS2048 using .
The errors I get are all related to the Eigen library (v3.9.6), which I know to be bug free (they're all features):
1>....\eigen\src/Core/Functors.h(52): error C2039: 'ReturnType' : is not a member of 'Eigen::internal::scalar_product_traits'
1> with
1> [
1> LhsScalar=float
1> , RhsScalar=Eigen::internal::result_of (float,float)>::type
1> ]
1> ....\eigen\src/Core/util/Meta.h(161) : see reference to class template instantiation 'Eigen::internal::scalar_product_op::type>' being compiled
1> with
1> [...
What is causing these compilation errors?
The average SO user could then copy your example directly into the ancient VS2017 IDE hit F7 and reproduce the errors. He would then realize that the expression
errIn.array() * (m_layerOut.array() > 0.f)
is the coefficient-wise product of a float and a bool, something that Eigen didn't support (in those days) without an explicit cast. He would then write up a quick answer saying:
You have to explicitly cast the comparison array from a boolean array to a float array like this:
err = errIn.array() * (m_layerOut.array() > 0.f).cast<float>();
You would then accept that answer, as it solved your problem, and move on.