armadillo | Armadillo C linear algebra library from http

I'm getting the following error with GCC >=9 and std>=11 merely by adding the header (MacOSX on MacBook Pro 2020 and armadillo installed with Homebrew and the code is compiled with standard CMake configuration) #include to my project.

Undefined symbols for architecture x86_64: "___emutls_v._ZN4arma19mt19937_64_instanceE", referenced from: __GLOBAL__sub_I_Test_HPP.cpp in Test_HPP.cpp.o ld: symbol(s) not found for architecture x86_64 collect2: error: ld returned 1 exit status make[2]: *** [Test_HPP] Error 1 make[1]: *** [CMakeFiles/Test_HPP.dir/all] Error 2

I've tried various hacks including optimization flags e.g. O2, O3 etc. but finally adding the preprocessor header #define ARMA_DONT_USE_WRAPPER apparently resolved the issue for now but I need an explanation to feel settled. If the above pre-processor is absolutely necessary to compile the code, should the armadillo library maintainers absorb the macro within the library itself? This kind of issues may take a lot of time to resolve as it is not originated in any programming logic.

I need to use the armadillo library and opencv library, but I find that both have mat type variables. How can I use them correctly when I use them?

I'm using Armadillo (10.4.1) in Visual Studio 2019 to do some matrix stuff. I used OpenBlas from NuGet manager, but everything was slow. I now want to switch to an up-to-date version of OpenBlas. I took the last one (0.3.15) and compiled it with minGW following this tuto : https://github.com/xianyi/OpenBLAS/wiki/How-to-use-OpenBLAS-in-Microsoft-Visual-Studio.

The compilation works fine, but when I try matrix multiplication, an error is thrown in the console : On entry to DSPR2 parameter number 1 had an illegal value

I do have defined ARMA_USE_BLAS, ARMA_DONT_USE_WRAPPER. I tried to played with ARMA_BLAS_LONG, ARMA_BLAS_UNDERSCORE, ARMA_USE_FORTRAN_HIDDEN_ARGS but nothing change.

Everything was working great using NuGet manager (OpenBlas 0.2.14.1). Here is a sample that doesn't works:

I wrote a C++11 function that conditionally initializes an object. Consider the following example:

I have to replace some columns in a sparse matrix with columns from another sparse matrix that has the same nonzero elements, just different values, based on a condition.

I am struggling with write access using iterators in Armadillo. The docs say that using sp_mat::col_iterator provides read/write access, however, when I try to write a value *it = B.col(...) I get an error message error: no match for ‘operator=’ (operand types are ‘arma::SpValProxy >’ and ‘arma::SpSubview_col’). Do I have a syntax mistake or am I understanding the concept of "write access" wrong?

I'm doing some physics simulation in C++ using Armadillo. I need to calculate a product looking like:

I've been trying to serialize the sparse matrix from armadillo cpp library. I am doing some large-scale numerical computations, in which the data get stored in a sparse matrix, which I'd like to gather using mpi(Boost implementation) and sum over the matrices coming from different nodes. I'm stuck right now is how to send the sparse matrix from one node to other nodes. Boost suggests that to send user-defined objects (SpMat in this case) it needs to be serialized.

Boost's documentation gives a good tutorial on how to serialize a user-defined type and I can serialize some basic classes. Now, armadillo's SpMat class is very complicated for me to understand, and serialize.

I've come across few questions and their very elegant answers

1. This answer by Ryan Curtin the co-author of Armadillo and author of mlpack has shown a very elegant way to serialize the Mat class.
2. This answer by sehe shows a very simple way to serialize sparse matrix.

Using the first I can mpi::send a Mat class to another node in the communicator, but using the latter I couldn't do mpi::send.

This is adapted from the second linked answer

I have a problem where I want to access certain entries of an armadillo-matrix "M" by a pointer in a struct (or class). After initializing M I set the pointer in the struct. By dereferencing the pointer I can see it has the right value (the first entry of M - or M(0,0)).

Then I change M to M * M. But now dereferencing the pointer does not give me the right value anymore. What's weird: If I have a small matrix i.e. 3x3 or 4x4 (see "matrixSize" in my code) the error does not happen. With small matrices dereferencing the pointer gives the RIGHT value. Bigger matrices though result in the wrong values (with 5x5 something like "2.76282e-320", which is probably some random place in memory).

What am I doing wrong here? How can I solve this problem?

If it helps, I'd like to explain what I want to achieve: I have a network of delay-coupled nodes that each have some sort of dynamic behaviour. (think delay-coupled differential equations DDEs - delay-coupled Oscillators). As they are delay-coupled I need to store their past states (an array of their histories). Each oscillator also has some LOCAL dynamic with dynamical variables that are not influencing other nodes which means that I don't have to store their histories. The matrix shall be used to keep the past states of some variable of the nodes. I want to have them in a matrix, because I want to use vector-operations on them (one index of the matrix represents time, while the other is the node-index). But I also want to access them individually to calculate some local dynamic at each of the nodes (oscillators). So I want to update the individual nodes, but also the global state. That's why having both representations helps: For the local dynamics I access the delayed states through a pointer into the matrix. For the global dynamics I access the coupled variables through a row in the matrix that functions as a history-array.

I want to use one or eye function in Armadillo to construct matrices or vectors. However, it does not allow me to do so. Here is a sample code: