giacrossi / wenoof Goto Github PK

in https://github.com/giacombum/WenOOF/blob/master/src/lib/type_weno_interpolator_js.f90#L205 you are trying to do allocated on a pointer... you must use associated instead of allocated. As a matter of fact the member JS is a pointer, see https://github.com/giacombum/WenOOF/blob/master/src/lib/type_weno_interpolator.f90#L33

After https://github.com/giacombum/WenOOF/blob/develop/src/lib/wenoof.f90#L71 an error handling must be inserted to prevent runtime errors in case of wrong alpha_base inputs: the only permitted are 'JS' and 'Z'.
The procedure to call is init_error of JS interpolator and the case is https://github.com/giacombum/WenOOF/blob/develop/src/lib/type_weno_interpolator_js.f90#L229

Improve WENO-Z alpha formula adding the maximum order

The smoothness indicators for non-uniform grids have very long expressions for coefficients, so now the highest non-uniform WENO reconstruction implemented is WENO 3-5.

when you invoke a method, namely a TBP, you are implicitly passing an argument (if nopass is explicitely used), thus here https://github.com/giacombum/WenOOF/blob/master/src/lib/type_weno_interpolator_js.f90#L206 you are doing a simple error...

    call self%IS%destroy(self%IS) ! wrong
    call self%IS%destroy ! right : self%IS is passed!

When non-uniform grids are used, for optimal weights, polynomial weights and smoothness indicators coefficients evaluations the coordinate of the reconstruction point is mandatory as an argument.
The problem is that the interfaces of the routines cannot be modified (or could be modified but an optional argument must be introduced), so how this new data can be passed to the routines?

You are not using the latest master release of WenOOF git submodule, please update to the master branches.

To be sure you'll always point to the master releases, modify your .gitmodules configuration as following:

[submodule "external/pyplot-fortran"]
  path = src/third_party/pyplot-fortran
  url = https://github.com/jacobwilliams/pyplot-fortran
  branch = master
[submodule "src/third_party/PENF"]
  path = src/third_party/PENF
  url = https://github.com/szaghi/PENF
  branch = master

Note the additions of branch = master option.

Maybe is necessary an associate construct for IS_type in https://github.com/giacombum/WenOOF/blob/master/src/lib/type_weno_interpolator_js.f90#L146? The pointer seems not associate, maybe the problem is that the call at https://github.com/giacombum/WenOOF/blob/master/src/lib/type_weno_interpolator_js.f90#L146 doesn't pass the right argument?

Maybe for the interpolator in the master branch (for uniform grids) could be intresting to replicate the Shu-Osher test that @szaghi has on his own repository.

Add WENO-AMM alpha coefficients

Correct the cell index examples into the routine create of the file type_weno_smoothness_indicators_js.f90

After the commit ded4f4f I obtain the following error with intel compiler:

Compiling src/lib/type_weno_interpolator_js.f90 serially src/lib/type_weno_interpolator_js.f90(159): error #6460: This is not a field name that is defined in the encompassing structure. [ALPHA_BASE] call self%alpha%alpha_base%create(S = self%S) ----------------------^ src/lib/type_weno_interpolator_js.f90(159): error #6460: This is not a field name that is defined in the encompassing structure. [CREATE] call self%alpha%alpha_base%create(S = self%S) ---------------------------------^ src/lib/type_weno_interpolator_js.f90(126): remark #7712: This variable has not been used. [ALPHA_BASE_TYPE] subroutine create(self, constructor, IS_type, alpha_type, alpha_base_type, weights_opt_type, polynomial_type) ------------------------------------------------------------^ compilation aborted for src/lib/type_weno_interpolator_js.f90 (code 1)

With gfortran compiler the error is:

`Compiling src/lib/type_weno_interpolator_js.f90 serially
src/lib/type_weno_interpolator_js.f90:159:32:

   call self%alpha%alpha_base%create(S = self%S)
                            1

Error: ‘alpha_base’ at (1) is not a member of the ‘weno_alpha_coefficient’ structure`

Add 15th (r=8) and 17th (r=9) order Jiang and Shu polynomials coefficients into the file type_weno_polynomials_js.f90 from the paper Very-high-order weno schemes, G. A. Gerolymos, D. Sénéchal, I. Vallet, JCP, 2009, vol. 228, pp. 8481-8524, doi:10.1016/j.jcp.2009.07.039

The functions weno_exp and tau maybe have to be changed in something new according to the following papers:

High Order Weighted Essentially Non-Oscillatory WENO-Z schemes for Hyperbolic Conservation Laws, Castro, Marcos and Costa, Bruno and Don, Wai Sun, JCP, 2011, vol. 230, pp. 1766--1792, doi:10.1016/j.jcp.2010.11.028

Accuracy of the weighted essentially non-oscillatory conservative finite difference schemes, Don, Wai-Sun and Borges, Rafael, JCP, 2013, vol. 250, pp. 347--372, doi:10.1016/j.jcp.2013.05.018

An improved weighted essentially non-oscillatory scheme for hyperbolic conservation laws, Borges, Rafael and Carmona, Monique and Costa, Bruno and Don, Wai Sun, JCP, 2008, vol. 227, pp. 3191--3211. doi:10.1016/j.jcp.2007.11.038

Add recursive order reduction strategy as implemented in G.A. Gerolymos, D. Senechal, and I. Vallet,
Very-High-Order WENO Schemes, AIAA 47. Aerospace Sciences Meeting, Orlando, [ FL, USA], Jan. 5-8, 2009

Add 15th (r=8) and 17th (r=9) order Jiang and Shu smoothness indicators coefficients into the file type_weno_smoothness_indicators_js.f90 from the paper Very-high-order weno schemes, G. A. Gerolymos, D. Sénéchal, I. Vallet, JCP, 2009, vol. 228, pp. 8481-8524, doi:10.1016/j.jcp.2009.07.039

With commit 0063bb1 a new API for alpha coefficients has been introduced in WenOOF: this API structure could also applied to IS, polynomials and maybe weights type. It is convenient to do so? To have a similar structure in all the types? Or the other types can remain the same?