I am using the class SimplicialLLT in order to get the Cholesky matrix from a SparseMatrix<double>. The problem works fine for small matrices. But once I increase the size of my matrix above a certain value (roughly 2^15), I get the following error on runtime:

Assertion failed: ((m_outerIndex[outer+1]-m_outerIndex[outer]==0 || m_data.index(m_data.size()-1)<inner) && "Invalid ordered insertion (invalid inner index)"), function insertBackByOuterInner, file /usr/local/include/eigen3/Eigen/src/SparseCore/SparseMatrix.h, line 384.

What is the meaning of this and what are options to fix this?

If this is of any help. I checked that my matrix is symmetric. It should be positive semidefinite. I adjusted it to be positive definite by adding a small shift.

Make sure you are using std::prtdiff_t as the index type:

typedef SparseMatrix<double,0,std::prtdiff_t> SpMat;
SpMat A;
SimplicialLLT<SpMat> llt(A);

Thanks! That worked. (I think you meant ptrdiff_t instead of prtdiff_t)

But what does it actually do? Since it is in the std namespace, is it part of some other library?

By default SparseMatrix uses 32 bits int to index non-zeros elements, but for very large matrices, 32 bits int are too small, and you need to use a 64 bits integer type.

For ptrdiff_t: http://en.cppreference.com/w/cpp/types/ptrdiff_t