Add mass fraction multi-phase table to sesame2spiner by adamdempsey90 · Pull Request #600 · lanl/singularity-eos

adamdempsey90 · 2026-01-12T20:41:09Z

PR Summary

This adds the mass fraction SESAME table to the spiner tables that sesame2spiner outputs (if it exists). Tested in a downstream code.

Putting this up now for an initial review.

I still need to add the mass fractions to the spiner EOS class. Are we okay with adding new MassFractionFromDensityTemperature functions to all EOS classes?

Still TODO:

Add this field to the spiner EOS
Some sort of test

PR Checklist

Adds a test for any bugs fixed. Adds tests for new features.
Format your changes by using the make format command after configuring with cmake.
Document any new features, update documentation for changes made.
Make sure the copyright notice on any files you modified is up to date.
After creating a pull request, note it in the CHANGELOG.md file.
LANL employees: make sure tests pass both on the github CI and on the Darwin CI

If preparing for a new release, in addition please check the following:

Update the version in cmake.
Move the changes in the CHANGELOG.md file under a new header for the new release, and reset the categories.
Maintainers: ensure spackages are up to date:
- LANL-internal team, update XCAP spackages
- Current maintainer of upstream spackages, submit MR to spack

sesame2spiner/io_eospac.cpp

Yurlungur

Nice! Glad it was a simple change. Regarding adding mass fractions to every EOS class... Broadly I'd be supportive of this, if we had an error message/stub in the base class as the default behavior. Clearly mass fractions aren't sensible for all EOS's but they are for enough EOS's we might want this.

sesame2spiner/io_eospac.cpp

adamdempsey90 · 2026-01-12T21:33:29Z

Nice! Glad it was a simple change. Regarding adding mass fractions to every EOS class... Broadly I'd be supportive of this, if we had an error message/stub in the base class as the default behavior. Clearly mass fractions aren't sensible for all EOS's but they are for enough EOS's we might want this.

It looks like the StellarCollapse EOS already has MassFractionsFromDensityTemperature

Yurlungur · 2026-01-12T22:57:21Z

Nice! Glad it was a simple change. Regarding adding mass fractions to every EOS class... Broadly I'd be supportive of this, if we had an error message/stub in the base class as the default behavior. Clearly mass fractions aren't sensible for all EOS's but they are for enough EOS's we might want this.

It looks like the StellarCollapse EOS already has MassFractionsFromDensityTemperature

I believe Helmholtz does too... or if it doesn't, it could be exposed.

…ssFraction and fill the string rather than the other way around

…d use the generic functions to do it

adamdempsey90 · 2026-01-15T16:13:37Z

sesame2spiner/io_eospac.cpp

+  std::vector<EOS_REAL> rhos, Ts, phs;
+  makeInterpPoints(rhos, lRhoBounds);
+  makeInterpPoints(Ts, lTBounds);


I am re-using the EOS rho-T grid for this but in testing we are seeing that maybe that isn't great, e.g., instead of clustering around STP, we probably want to cluster the points around phase transitions. I could add another set of parameters to the inputs that describe the grid decomp for just mass fractions?

I think that makes a lot of sense. Yeah there's no reason the two grids have to be the same.

If you're going to use different grids, I would just use the sesame grid. (piecewise linear). eospac just does bilinear interpolation for mass fractions so we should not need higher density than sesame.

My suggestion for now is you add a TODO to this effect, but we don't let this desired functionality block the merge.

…lease enter a commit message to explain why this merge is necessary,

adamdempsey90 · 2026-02-24T21:53:16Z

Updated to include the mass fraction query in the spiner EOS object. The unit test I added passes on my CPU machine, but not sure about the GPU machines (mostly the EOSPAC paths...)

Yurlungur · 2026-02-24T22:30:35Z

Would you like me to do review now? I could also maybe get eospac paths working through the CI?

adamdempsey90 · 2026-02-24T23:12:57Z

Would you like me to do review now? I could also maybe get eospac paths working through the CI?

yes please.

adamdempsey90 · 2026-02-24T23:16:12Z

singularity-eos/eos/eos_spiner_rho_sie.hpp

      Indexer_t &&lambda = static_cast<Real *>(nullptr)) const;
  template <typename Indexer_t = Real *>
+  PORTABLE_INLINE_FUNCTION void MassFractionsFromDensityTemperature(
+      const Real rho, const Real T, Real *scratch,
+      Indexer_t &&lambda = static_cast<Real *>(nullptr)) const;
+  template <typename Indexer_t = Real *>
+  PORTABLE_INLINE_FUNCTION void MassFractionsFromDensityInternalEnergy(
+      const Real rho, const Real sie, Real *scratch,
+      Indexer_t &&lambda = static_cast<Real *>(nullptr)) const;


I should probably add this functionality to the eospac EOS

Agreed, but it can wait to another PR. Maybe just add a TODO?

Yurlungur

@adamdempsey90 sorry for the delay. Some comments below. Did you want to mess with the paths yourself or did you want me to look at that? I'm happy to poke at it. Should have some time to do so tomorrow.

Yurlungur · 2026-02-24T23:14:14Z

.gitlab/build_and_test.sh

  source ${BUILD_ENV}
  pushd ${BUILD_DIR}
  if [[ -f ./sesame2spiner/sesame2spiner ]]; then
+  echo "/usr/projects/data/eos/eos-developmental/Sn2162/v01/sn2162-v01.bin" > sesameFilesDir.txt


I think this actually probably is sufficient to getting the paths to work on device.

Yeah, the gpu tests are working. Although this path shouldn't be correct for LLNL, so I don't know about that

Ah. We may need a general solution for that. Probably the same problem will show up for the KPT tests. My suggestion is that we set these paths as environment variables in re-git then they can be automatically slurped up here. But before we go down that path... let's ping Richard.

@rbberger what are your thoughts?

We could also have some project level env script that gets loaded by the CI and sets those paths

Yurlungur · 2026-02-25T22:29:42Z

eospac-wrapper/eospac_wrapper.cpp

  eosCheckError(errorCode, "eos_GetTableCmnts", eospacWarn);
 }

+void eosSafeTableMetaData(EOS_INTEGER *table, EOS_INTEGER infoItem, EOS_CHAR *infoString,


Yurlungur · 2026-02-25T22:36:07Z

eospac-wrapper/eospac_wrapper.cpp

  }
 }

+int eosCheckTableExistence(EOS_INTEGER tableType_, int matid_, Verbosity eospacWarn) {


Nice added capability.

Yurlungur · 2026-02-25T22:46:32Z

sesame2spiner/io_eospac.hpp

 void eosColdCurveMask(int matid, const Bounds &lRhoBounds, const int numSie,
                      const DataBox &sieColdCurve, DataBox &mask,
                      Verbosity eospacWarn = Verbosity::Quiet);
+bool eosMassFraction(int matid, const Bounds &lRhoBounds, const Bounds &lTBounds,


using a boolean return code here is clever. 👍

Yurlungur · 2026-02-25T22:48:02Z

sesame2spiner/io_eospac.cpp

+  std::vector<EOS_REAL> rhos, Ts, phs;
+  makeInterpPoints(rhos, lRhoBounds);
+  makeInterpPoints(Ts, lTBounds);


My suggestion for now is you add a TODO to this effect, but we don't let this desired functionality block the merge.

Yurlungur · 2026-02-25T23:29:51Z

singularity-eos/eos/eos_spiner_rho_sie.hpp

      Indexer_t &&lambda = static_cast<Real *>(nullptr)) const;
  template <typename Indexer_t = Real *>
+  PORTABLE_INLINE_FUNCTION void MassFractionsFromDensityTemperature(
+      const Real rho, const Real T, Real *scratch,
+      Indexer_t &&lambda = static_cast<Real *>(nullptr)) const;
+  template <typename Indexer_t = Real *>
+  PORTABLE_INLINE_FUNCTION void MassFractionsFromDensityInternalEnergy(
+      const Real rho, const Real sie, Real *scratch,
+      Indexer_t &&lambda = static_cast<Real *>(nullptr)) const;


Agreed, but it can wait to another PR. Maybe just add a TODO?

Yurlungur · 2026-02-25T23:30:55Z

singularity-eos/eos/eos_spiner_rho_sie.hpp

+  bool has_mf = false;
+  // Need to hold the phase names for multiphase EOS
+  // This isn't great, but the class needs to be trivially copyable
+  // I've chosen something reasonable, e.g., 15 phases with 32 character names
+  char phase_names[480];
+  int numphases = 1;


See my comment in eos_spiner_rho_temp

Yurlungur · 2026-02-25T23:31:42Z

singularity-eos/eos/eos_spiner_rho_sie.hpp

+PORTABLE_INLINE_FUNCTION void
+SpinerEOSDependsRhoSieTransformable<TransformerT>::MassFractionsFromDensityTemperature(
+    const Real rho, const Real T, Real *scratch, Indexer_t &&lambda) const {
+  if (!has_mf) {
+    *scratch = 1.0;
+    return;
+  }
+  const Real lRho = spiner_common::to_log(rho, lRhoOffset_);
+  const Real lT = spiner_common::to_log(T, lTOffset_);
+  IndexerUtils::SafeSet<IndexableTypes::LogDensity>(lambda, Lambda::lRho, lRho);
+  DataBox mf1d(scratch, numphases);
+  mf1d.interpFromDB(mF_, lRho, lT);
+}
+
+template <template <class> class TransformerT>
+template <typename Indexer_t>
+PORTABLE_INLINE_FUNCTION void
+SpinerEOSDependsRhoSieTransformable<TransformerT>::MassFractionsFromDensityInternalEnergy(
+    const Real rho, const Real sie, Real *scratch, Indexer_t &&lambda) const {
+  if (!has_mf) {
+    *scratch = 1.0;
+    return;
+  }
+  const Real lRho = spiner_common::to_log(rho, lRhoOffset_);
+  const Real lE = spiner_common::to_log(sie, lEOffset_);
+  const Real T = T_.interpToReal(lRho, lE);
+  const Real lT = spiner_common::to_log(T, lTOffset_);
+  DataBox mf1d(scratch, numphases);
+  mf1d.interpFromDB(mF_, lRho, lT);
+}


See my comment in eos_spiner_rho_temp

Yurlungur · 2026-02-25T23:35:45Z

singularity-eos/eos/eos_spiner_common.hpp

  return H5Gopen(file, name, property);
 }

+inline void h5_safe_read_string(hid_t &file, const char *grp, const char *name,


I'm not sure how much this matters, given who would stick a string in a dataset, but perhaps name this h5_safe_read_str_attr or something? Just to emphasize it's an attribute?

Yurlungur · 2026-02-25T23:37:44Z

singularity-eos/eos/eos_spiner_common.hpp

+  }
+  // We have to allocate sadly
+  char *tmp = new char[strlen + 1];
+  std::memset(tmp, 0, strlen + 1);


I would prefer not to do the temporary allocation this way. I have a comment in the spiner EOS constructor about how you could do something evil with char* and actually return the pointer to memory allocated here. But if you don't choose to go down that route, why not this?

std::vector<char> tmp(0, strlen + 1);

and then you pass in tmp.data() as needed and RAII takes care of cleanup

Add mass fraction table to sesame2spiner

6829cde

adamdempsey90 requested review from Yurlungur, buechlerm and jhp-lanl January 12, 2026 20:41

adamdempsey90 self-assigned this Jan 12, 2026

adamdempsey90 marked this pull request as draft January 12, 2026 20:41

jhp-lanl reviewed Jan 12, 2026

View reviewed changes

sesame2spiner/io_eospac.cpp Outdated Show resolved Hide resolved

sesame2spiner/io_eospac.cpp Outdated Show resolved Hide resolved

Yurlungur reviewed Jan 12, 2026

View reviewed changes

sesame2spiner/io_eospac.cpp Outdated Show resolved Hide resolved

adamdempsey90 added 2 commits January 15, 2026 08:42

Add table existence check to eospac_wrapper. Return a bool from eosMa…

909776e

…ssFraction and fill the string rather than the other way around

Move getting of phase names and number of phases to eospac_wrapper an…

a8a9804

…d use the generic functions to do it

adamdempsey90 commented Jan 15, 2026

View reviewed changes

adamdempsey90 added 5 commits February 11, 2026 06:41

Merge remote-tracking branch 'origin/main' into dempsey/mass_frac

698b245

Merge remote-tracking branch 'origin/main' into dempsey/mass_frac # P…

1fddda8

…lease enter a commit message to explain why this merge is necessary,

safe hdf5 string read

afdb537

Add MassFractionsFromDensity* to spiner eos

5672fc8

Add multiphase spiner unit test for tin

1da4ba3

adamdempsey90 marked this pull request as ready for review February 24, 2026 21:53

adamdempsey90 added 2 commits February 24, 2026 14:58

void function

82fd328

Format

d150684

adamdempsey90 commented Feb 24, 2026

View reviewed changes

adamdempsey90 added 3 commits February 24, 2026 17:25

semicolon after fence;

dabb27b

remove unused

7dd80e6

Properly use host/device eos

73f8eed

Yurlungur reviewed Feb 25, 2026

View reviewed changes

Conversation

adamdempsey90 commented Jan 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR Summary

PR Checklist

Uh oh!

Uh oh!

Uh oh!

Yurlungur left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

adamdempsey90 commented Jan 12, 2026

Uh oh!

Yurlungur commented Jan 12, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

adamdempsey90 commented Feb 24, 2026

Uh oh!

Yurlungur commented Feb 24, 2026

Uh oh!

adamdempsey90 commented Feb 24, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Yurlungur left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

adamdempsey90 commented Jan 12, 2026 •

edited

Loading