Error Eigen: THIS_COEFFICIENT_ACCESSOR_TAKING_ONE_ACCESS_IS_ONLY_FOR_EXPRESSIONS_ALLOWING_LINEAR_ACCESS

[JamieAl]

I have an iterative solver that works fine in a main script and when I move it to a function things run fine until it hits a specific line where I get the error:

 error: static assertion failed: THIS_COEFFICIENT_ACCESSOR_TAKING_ONE_ACCESS_IS_ONLY_FOR_EXPRESSIONS_ALLOWING_LINEAR_ACCESS
  142 |       EIGEN_STATIC_ASSERT(internal::evaluator<Derived>::Flags & LinearAccessBit,
      |                                                         ~~~~~~^~~~~~~~~~~~~~~~~
/mnt/c/Users/lujain/Documents/eigen-3.4.0/eigen-3.4.0/Eigen/src/Core/util/StaticAssert.h:33:54: note: in definition of macro ‘EIGEN_STATIC_ASSERT’
   33 |     #define EIGEN_STATIC_ASSERT(X,MSG) static_assert(X,#MSG);

Which I have never seen this error before and not sure what it means.

//initializations
do{			
//some calculations
for (int l = 0; l < kx.size(); l++){
	L = div_x_act_on_grad_x * std::pow(ninvksqu(l),2.0) + div_y_act_on_grad_y; //ERROR
}

where here L, div_x_act_on_grad_x, div_y_act_on_grad_y, and ninvksqu are all Eigen matrices and defined as:

Eigen::Matrix< double, (ny+1), (ny+1)> L;

Eigen::SparseMatrix<double> div_x_act_on_grad_x((ny+1),(ny+1));
div_x_act_on_grad_x.setIdentity();
Eigen::Matrix< double, (ny+1), (ny+1)> div_y_act_on_grad_y; 
Matrix <double, 1, nx> eninvksqu; 

ninvksqu is an input to the function and that makes me think it could be throwing off the error somehow.

The code:
https://godbolt.org/z/1z5TE3TKP

[helios]

I would guess it's because the type of L and the type of div_x_act_on_grad_x are incompatible. Maybe L a sparse matrix as well?

[JamieAl]

@helios
hmm, but the thing this code runs fine. This error happens when I move a portion to a function and try to test it. Kinda strange to me.

[JamieAl]

@helios

I would guess it's because the type of L and the type of div_x_act_on_grad_x are incompatible. Maybe L a sparse matrix as well?

Wait, actually, I think you accidentally fixed my issue. It's not the sparse matrix, but the eninvksqu was being passed as a matrix and not a vector so it should be passed as Eigen::VectorXd