feat: support non-scalar valued field specifications by kdrienCG · Pull Request #4034 · GEOS-DEV/GEOS

kdrienCG · 2026-04-22T10:16:38Z

This PR generalizes field specifications to support non-scalar values, notably scales (and so function names).

The following interface is proposed:

<FieldSpecification
   name="perm_wall"
   initialCondition="1"
   setNames="{ wall1, wall2 }"
   objectPath="ElementRegions/Region1/block1"
   fieldName="rockPerm_permeability"
   scale="{ 2.0e-22, 2.0e-22, 2.0e-22 }"
/>

// or for a single value
<FieldSpecification
   name="..."
   ...
   scale="42"
/>

Instead of:

<FieldSpecification
   name="permx_wall"
   component="0"
   initialCondition="1"
   setNames="{ wall1, wall2 }"
   objectPath="ElementRegions/Region1/block1"
   fieldName="rockPerm_permeability"
   scale="2.0e-22"
/>

<FieldSpecification
   name="permy_wall"
   component="1"
   initialCondition="1"
   setNames="{ wall1, wall2 }"
   objectPath="ElementRegions/Region1/block1"
   fieldName="rockPerm_permeability"
   scale="2.0e-22"
/>

<FieldSpecification
   name="permz_wall"
   component="2"
   initialCondition="1"
   setNames="{ wall1, wall2 }"
   objectPath="ElementRegions/Region1/block1"
   fieldName="rockPerm_permeability"
   scale="2.0e-22"
/>

~~This PR also add a regionNames array to make it easier to specify multiple regions.~~

dkachuma · 2026-05-14T16:06:54Z

-  /// The name of the function used to generate values for application.
-  string m_functionName;
+  /// Name(s) of the function used to generate values for application.
+  string_array m_functionName;


Suggested change

string_array m_functionName;

string_array m_functionNames;

dkachuma · 2026-05-14T16:08:53Z

Suggested change

constexpr static char const * functionNamesString() { return "functionNames"; }

Same for scale -> scales.

To avoid breaking users' input files, I kept it as "scale" (same for "functionName"). However, if the previously mentioned change/commit to accept unbracketed single values for 1D arrays is rejected, I will rename it to scales (unless breaking users' files in this case is acceptable?).

Do both. Enforce scales = {}

Could scales (plural) not be misleading here? Fields like permeability or globalCompFraction are already implicitly plural (one value per cell, per direction/component). Wouldn't scale (singular) better reflect that this is a single scaling vector applied along the second dimension?

dkachuma · 2026-05-14T16:13:31Z

 {
+  string const path = m_regionNames.empty()
+                      ? m_objectPath
+                      : "ElementRegions/{" + stringutilities::join( m_regionNames, ' ' ) + "}";


This isn't quite what I expected. I expected a loop over the region names. Has this been tested with multiple regions?

Yes, regionNames works with multiple regions (tested with isothermalLeakyWell_smoke_3d.xml). It relies on an objectPath capability that accepts multiple "nodes" at the same level (e.g., objectPath="ElementRegions/{ region1 region2 }").

However, after testing, I found that using objectPath="ElementRegions/{ region1 region2 }" in a XML does not work, whereas regionNames="{ region1, region2 }" does. This is likely due to a regex in the parsing phase that blocks it.

However, after testing, I found that using objectPath="ElementRegions/{ region1 region2 }" in a XML does not work, whereas regionNames="{ region1, region2 }" does. This is likely due to a regex in the parsing phase that blocks it.

This is not regex, it is fnmatch. This not a valid fnmatch pattern. Much to your users displeasure I think the valid fnmatch pattern is:

regionNames="ElementRegions/region[12]"

or full expansion:
regionNames="{ ElementRegions/region1, ElementRegions/region2 }"

annoyingly it would be hard (not possible) to get a fnmatch expansion to something like

ElementRegions/reservoir, ElementRegions/overburden

that would have to be:

{ ElementRegions/reservoir, ElementRegions/overburden }

alternatively if you take my suggestion that we just assume everything not FaceManager, EdgeManager, NodeManager is ElementRegion, then I think

objectPath="{ reservoir, overburden }"

would work as desired.

kdrienCG · 2026-05-19T08:43:29Z

I've removed the regionNames attribute, I'll open a separate PR for the objectPath

…/nonScalarFieldSpecification

MelReyCG · 2026-05-19T15:32:54Z

@kdrienCG can you look if you can add a check on the user input vector dimension in relation to the field tensorial dimensions?

Potential solution was: ```cpp #ifndef NDEBUG GEOS_WARNING_IF( m_functionName.size() > 1, "getFunctionName() used while m_functionName has multiple values. Consider using getFunctionNames()" ); #endif ``` It may not be worth bringing the whole Logger header for this check...

…ents

MelReyCG · 2026-06-22T14:21:32Z

+   * Accessor
+   * @return const m_scales
+   */
+  arrayView1d< real64 const > getScales() const


Be consistent (plural) with scaleString() { return "scale"; }

(same for setter)

MelReyCG · 2026-06-22T14:31:05Z

+  bool usesNonScalarValues() const
+  { return m_scale.size() > 1 || m_functionName.size() > 1; }


Seems quite complicated to maintain. To avoid having 4 ways branching ( scale scalar or not, functionNames scalar or not ), I suggest to, for scale and functionNames:

control that if we have >1-dim vectors, it has the the same dimensions as the field (numArrayComp()?) in postProcessInput();

still in postProcessInput(), duplicate any 1-dim vector so it has the same dimensions as the field;

then we are able to remove this method, and any downstream branching.

rename "scale" to "scales" in the code only. rename "functionName" to "functionNames" in the code only.

…dSpecification' into feature/kdrienCG/nonScalarFieldSpecification

kdrienCG added 2 commits April 22, 2026 10:11

add scales and functionNames attributes

c8adaac

modify applyFieldValueKernel to accept non-scalar applications

67da13f

kdrienCG self-assigned this Apr 22, 2026

kdrienCG added the type: feature New feature or request label Apr 22, 2026

kdrienCG added 20 commits April 22, 2026 14:46

add validations against scale(s) and functionName(s)

e133dd2

add regionNames attribute to field specifications

7c18352

add setRegionNames method

2dc1751

fix typo in log message parameter

d5fbc7f

modify scales setter and getter

e7af0f5

modify application methods to accept non-scalar field specifications

663a168

move region names construction

7e4f6a5

call FieldSpecification validations in subclasses

839d653

add doxygen comment for overloaded computeRhsContribution

b0ae8d2

rename isVectorMode() to usesNonScalarValues()

e073b5f

add functionNames check to usesNonScalarValues()

156ffc7

update warning messages when combining scalars and non-scalars

bfb53d5

remove newline

d74a3d1

fix wrong variable used

f684f70

add rst documentation

28dbcff

modify one inputFiles to use the new feature

0ecdd93

uncrustify

76d2297

remove unnecessary newline character in Wrapper doc

27a5eb8

add schema

d766df1

Merge branch 'develop' into feature/kdrienCG/nonScalarFieldSpecification

0013c0a

kdrienCG marked this pull request as ready for review May 4, 2026 08:38

kdrienCG requested review from OmarDuran, dkachuma, herve-gross, jhuang2601 and rrsettgast as code owners May 4, 2026 08:38

dkachuma reviewed May 14, 2026

View reviewed changes

kdrienCG added 2 commits May 15, 2026 12:04

Merge branch 'develop' into feature/kdrienCG/nonScalarFieldSpecification

8e4eaed

modify one integrated test file to use regionNames

6012577

kdrienCG marked this pull request as ready for review May 15, 2026 10:13

kdrienCG requested review from tjb-ltk and wrtobin as code owners May 15, 2026 10:13

kdrienCG added 2 commits May 15, 2026 12:55

fix wrong variable used

05597b5

remove bcApplication* members

94cbfde

kdrienCG marked this pull request as draft May 15, 2026 15:27

kdrienCG added 3 commits May 15, 2026 17:36

Merge branch 'develop' into feature/kdrienCG/nonScalarFieldSpecification

42e7581

update baselines

c3c4882

remove regionNames

3027e18

Merge remote-tracking branch 'upstream/develop' into feature/kdrienCG…

11ed988

…/nonScalarFieldSpecification

kdrienCG added 9 commits May 20, 2026 11:27

remove modifications to an xml file

6f7047f

reorder includes in testXmlWrapper.cpp

cece6cb

modify Wrapper description

ec3d94e

removed unnecessary variable copy

7d5407c

update schema

bc2d384

fix typo causing an out of bound access

bfd9d98

add validation between FS number of elements and field number of elem…

3036722

…ents

Merge branch 'develop' into feature/kdrienCG/nonScalarFieldSpecification

e0e6769

MelReyCG reviewed Jun 22, 2026

View reviewed changes

kdrienCG added 5 commits July 9, 2026 11:43

expand m_scale and m_functionName to the targeted field expected size

06cd526

rename methods

c31f8e3

rename "scale" to "scales" in the code only. rename "functionName" to "functionNames" in the code only.

Merge remote-tracking branch 'upstream/feature/kdrienCG/nonScalarFiel…

29dc99d

…dSpecification' into feature/kdrienCG/nonScalarFieldSpecification

Merge branch 'develop' into feature/kdrienCG/nonScalarFieldSpecification

ef4dd72

Merge branch 'develop' into feature/kdrienCG/nonScalarFieldSpecification

22714e3


	constexpr static char const * functionNamesString() { return "functionNames"; }

		bool usesNonScalarValues() const
		{ return m_scale.size() > 1 \|\| m_functionName.size() > 1; }

Uh oh!

Conversation

kdrienCG commented Apr 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

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!

MelReyCG Jun 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

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!

rrsettgast May 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kdrienCG commented May 19, 2026

Uh oh!

MelReyCG commented May 19, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

MelReyCG Jun 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

kdrienCG commented Apr 22, 2026 •

edited

Loading

MelReyCG Jun 22, 2026 •

edited

Loading

rrsettgast May 18, 2026 •

edited

Loading

MelReyCG Jun 22, 2026 •

edited

Loading