Viscosity for multispec DG by samuelh-nd · Pull Request #707 · quinoacomputing/quinoa

samuelh-nd · 2026-05-07T18:24:12Z

This new branch combines the viscosity_1 and viscosity_Suther branches to implement viscosity for multispecies DG. Constant viscosity is currently running and quantitative validation is ongoing. Sutherland's Law validation is coming soon. Please use this viscosity_Suther_1 branch for all viscosity development going forward.

This change is

…viscosity_1

…read viscosity from control file

adityakpandare

Thank you for the contributions. This is quite a large change-set. Due to that it is unrealistic to look at all of it at once. My suggestion is that we do this code-review step-wise:

parsing,
wiring into DG,
viscous flux operators,
viscosity (Sutherland's law) or material EOS related changes.

I have taken a look at the first item from this list, and have comments below. Once those are addressed, we can move to the next items in the list.

A couple of additional things: our CI servers are failing on this branch. Passing all builds (except the gnu-smp one) is another criterion we require for feature branches to be merged in, so that users do not get adversely affected. Secondly, our styling guidelines are found here: https://quinoacomputing.github.io/contributing.html#styling. Please go through these, and check that those are followed in your change-set. From what I can see, there are quite a few trailing whitespaces in this change-set.

Please let me know if you have any questions.

@adityakpandare reviewed 6 files and made 12 comments.
Reviewable status: 6 of 27 files reviewed, 11 unresolved discussions (waiting on madpeck3 and samuelh-nd).

src/Control/Inciter/InputDeck/InputDeck.hpp line 146 at r1 (raw file):

  tag::Sutherland, std::vector< bool >,
  tag::id,        std::vector< uint64_t >,
  tag::gamma,     std::vector< tk::real >,

Lot of repeating quantities here. Let's clean this up to remove repeats.

src/Control/Inciter/InputDeck/LuaParser.hpp line 40 at r1 (raw file):

      ctr::InputDeck& gideck );

    //! Check and store material property into inpudeck storage

This function description is the same as the checkStoreMatProp(). Let's modify it to say what it really does.

src/Control/Inciter/InputDeck/LuaParser.cpp line 638 at r1 (raw file):

            mati_deck.get< tag::mu >());   
        }
      }

I don't understand where this code is coming from. It looks like a duplicate of what's in registerMaterials(). It might be that your recent merge with develop was incorrect.

src/Control/Inciter/InputDeck/LuaParser.cpp line 1421 at r1 (raw file):

  if (gideck.get< tag::pde >() != inciter::ctr::PDEType::MULTIMAT &&
      gideck.get< tag::pde >() != inciter::ctr::PDEType::COMPFLOW) return;

Unnecessary line removal

src/Control/Inciter/InputDeck/LuaParser.cpp line 1796 at r1 (raw file):

    else gideck.get< tag::multispecies, tag::viscous >() = true;
      checkStoreMatProp(sol_mat[imat+1], "mu", ntype, mati_deck.get< tag::mu >());
    }

Same thing about code here. This code belongs in registerSpecies(). Can you please check if your recent merge with develop caused this?

src/Control/Inciter/InputDeck/LuaParser.cpp line 1830 at r1 (raw file):

  // set target data in inputdeck
  // species vector size is one, since all species are only of one type for now
  gideck.get< tag::species >().resize(1);

This resize is already done a few lines above. Let's remove it.

src/Control/Inciter/InputDeck/LuaParser.cpp line 1930 at r1 (raw file):

        // Sutherland (bool)
        checkStoreMatPropBool(sol_spc[imat+1], "Sutherland", nspec,
          spci_deck.get< tag::Sutherland >());

Why do we need a vector for checking if Sutherland's law is used or not? Is there a situation where we would use Sutherland's law for some species, and not use it for others?

src/Control/Inciter/InputDeck/LuaParser.cpp line 1931 at r1 (raw file):

        checkStoreMatPropBool(sol_spc[imat+1], "Sutherland", nspec,
          spci_deck.get< tag::Sutherland >());
      }

I don't think these changes should be in the else-branch here. This basically means that if a spec_name is provided, viscosity cannot be triggered, which is not intended. Let's move these outside that if-test.

src/Control/Inciter/InputDeck/LuaParser.cpp line 1972 at r1 (raw file):

  std::vector< bool >& storage )
// *****************************************************************************
//  Check and store material property into inpudeck storage

Let's change this comment to describe what this function does- current description is same are checkstorematprop().

src/PDE/MultiMat/DGMultiMat.hpp line 991 at r1 (raw file):

        tk::volInt( nmat, t, m_mat_blk, ndof, rdof, nelem,
        inpoel, coord, geoElem, flux, visc_flux, velfn, Problem::src, U, P, ndofel, R, viscous,
                    intsharp);

The changes above to the indentation etc are unnecessary. Let's adhere to the styling that's present and leave the diff to be minimal.

src/PDE/Transport/DGTransport.hpp line 406 at r1 (raw file):

          std::get<0>(b), fd, geoFace, geoElem, inpoel, coord, t, Upwind::flux,
          visc_flux, Problem::prescribedVelocity, std::get<1>(b), U, P, ndofel, R,
          riemannDeriv, viscous, intsharp );

Please check for unnecessary indentation changes here as well. Let's keep the diff to a minimal, especially considering how big a change-set this is.

samuelh-nd

Thanks for your feedback. I've left a few comments regarding parsing - future commits will remove duplicate code and simplify the control file. I'm currently working on revisions to ensure this branch passes all required CI server builds.

Contributions going forward will adhere to styling.

@samuelh-nd made 5 comments.
Reviewable status: 6 of 27 files reviewed, 11 unresolved discussions (waiting on adityakpandare and madpeck3).

src/Control/Inciter/InputDeck/LuaParser.cpp line 638 at r1 (raw file):