Compiling Error for some orders for quadrature.h and Output is not correct.

[eneswitwit]

quadrature.cc:27:38: error: constexpr variable 'quad' must be initialized by a constant expression
        constexpr Quadrature<double, order> quad;
                                            ^~~~
./../include/constexpr_array.h:13:9: note: constexpr evaluation hit maximum step limit; possible infinite loop?
        return data_[n];
        ^
./../include/quadrature_constexpr.h:50:9: note: in call to '&p->operator[](4)'
        p[i + 1] = static_cast<y_type>( (a2 + a3 * x) * p[i] - a4 * p[i - 1]) / a1;
        ^
./../include/quadrature_constexpr.h:117:38: note: in call to 'JacobiP(8.717401e-01, 2, 2)'
            J_x   =  0.5 * (m + 3) * JacobiP < y_type, m - 1 > (r, alpha_2, beta_2);
                                     ^
./../include/quadrature_constexpr.h:79:24: note: in call to '&quad->compute_kth_entry(1, knots)'
            knots[k] = compute_kth_entry(k, knots);
                       ^
./../include/quadrature_constexpr.h:62:18: note: in call to '&quad->compute_quadrature_points()'
        knots_ ( compute_quadrature_points()) {}
                 ^
quadrature.cc:27:38: note: in call to 'Quadrature()'
        constexpr Quadrature<double, order> quad;
                                            ^
1 error generated.

[eneswitwit]

It works just for long double.

[masterleinad]

How is this solve if it only works for a particular type?

[eneswitwit]

That quadrature is the same quadrature from deal.II, because we did not want to write an own quadrature class. The deal.II class also just works only for long double. If it is important, we can try to figure out how we can make it work converge for double.

On 16 March 2017 at 13:34, Daniel Arndt notifications@github.com wrote:

How is this solve if it only works for a particular type?

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/Oldbridgegeek/Sumfactorization/issues/12#issuecomment-287043474, or mute the thread https://github.com/notifications/unsubscribe-auth/AUiPW0Rgp0_nZMlxYS8swmyuW1XdlXkhks5rmSxQgaJpZM4MfKwn .

[masterleinad]

No, it's not high-priority.