I have a rather large number of expressions, which I must normalize. To increase performance I wanted to implement this on multiple threads. However, this always segfaults and I am having trouble to find out where my error is. Consider the following code:

```c++

// H is a 9x9 matrix

void norm_matrix(GiNaC::matrix& H) {

    thread_pool tp = thread_pool();

    auto n = 9;

    std::vector<std::future<GiNaC::ex>> futures(n*n);

    for(int r = 0; r < n; ++r) {

         for(int c = 0; c < n; ++c) {

             futures[r*n+c] = tp.submit(

                  [&h = H(r,c)] () -> GiNaC::ex {

                  return h.normal();

             });

         }

    }

    for(int r = 0; r < n; ++r) {

        for(int c = 0; c < n; ++c) {

            H(r,c) = futures[r*n+c].get();

        }

    }

}

```

At the second iterations on the call of h.normal(), this crashes with a segfault, both with a reference as well as with a copy. Has GiNaC been used in a multithreaded environment previously and has this error occured?

Kind regards

Spooky Ghost