Physics list ShieldingLIQMD_HP_EMZ not working any more with official OpenGate10 release 10.1.1

[GitFuchs]

Hi,

I am simulating carbon ions with the physics list "sim.physics_manager.physics_list_name = "ShieldingLIQMD_HP_EMZ"

I upgraded to the new opengate release. And get an error message related to the physics list used:
It seems to me that the parsing of _EMZ when using some other options such as _HP is faulty.
In addition there seems to be an issue with ShieldingLIQMD
An older official gate10 release worked.
Upgrade was done using python -m pip install --upgrade opengate

See details below:

When using
sim.physics_manager.physics_list_name = "ShieldingLIQMD_HP_EMZ"

Exception: Cannot find the physic list: ShieldingLIQMD_HP_EMZ

**** INFO about GATE physics lists ****
* Known Geant4 lists are: [FTFP_BERT, FTFP_BERT_TRV, FTFP_BERT_ATL, FTFP_BERT_HP, FTFQGSP_BERT, FTFP_INCLXX, FTFP_INCLXX_HP, FTF_BIC, LBE, QBBC, QGSP_BERT, QGSP_BERT_HP, QGSP_BIC, QGSP_BIC_HP, QGSP_BIC_AllHP, QGSP_FTFP_BERT, QGSP_INCLXX, QGSP_INCLXX_HP, QGS_BIC, Shielding, ShieldingLEND, ShieldingLIQMD, ShieldingM, NuBeam, Shielding_HP, ShieldingLIQMD_HP, ShieldingM_HP, FTFP_BERT_HPT, FTFP_INCLXX_HPT, QGSP_BERT_HPT, QGSP_BIC_HPT, QGSP_BIC_AllHPT, QGSP_INCLXX_HPT, Shielding_HPT, ShieldingLIQMD_HPT, ShieldingM_HPT]
* With EM options: [_EMV, _EMX, _EMY, _EMZ, _LIV, _PEN, __GS, __SS, _EM0, _WVI, __LE]
* Or the following simple physics lists with a single PhysicsConstructor: 
* ['G4EmStandardPhysics', 'G4EmStandardPhysics_option1', 'G4EmStandardPhysics_option2', 'G4EmStandardPhysics_option3', 'G4EmStandardPhysics_option4', 'G4EmStandardPhysicsGS', 'G4EmLowEPPhysics', 'G4EmLivermorePhysics', 'G4EmLivermorePolarizedPhysics', 'G4EmPenelopePhysics', 'G4OpticalPhysics'] 

Which per se is interesting as it is listed in there.
Using without EMZ, e.g. ShieldingLIQMD_HP
Results in:

Method close_physics_constructors of object <opengate.engines.PhysicsEngine object at 0x75efc64af890> requires g4_augmented_physics_list to be set, but it is None.
Traceback (most recent call last):
  File "/home/hfuchs/Documents/WORK/2026_Hedgehog/Simulation/DistanceTest/validation_sim_noSTEP/sim_MedAustron_plan_v9_neg2p_3mm_carbon.py", line 660, in <module>
    main()
  File "/home/hfuchs/Documents/WORK/2026_Hedgehog/Simulation/DistanceTest/validation_sim_noSTEP/sim_MedAustron_plan_v9_neg2p_3mm_carbon.py", line 637, in main
    simLink, output_dictionary = simulation(
                                 ^^^^^^^^^^^
  File "/home/hfuchs/Documents/WORK/2026_Hedgehog/Simulation/DistanceTest/validation_sim_noSTEP/sim_MedAustron_plan_v9_neg2p_3mm_carbon.py", line 443, in simulation
    output_sim = sim.run()
                 ^^^^^^^^^
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/managers.py", line 2266, in run
    output = self._run_simulation_engine(False)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/managers.py", line 2224, in _run_simulation_engine
    output = se.run_engine()
             ^^^^^^^^^^^^^^^
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/engines.py", line 1585, in run_engine
    self.initialize()
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/engines.py", line 1765, in initialize
    self.physics_engine.initialize_before_runmanager()
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/engines.py", line 351, in initialize_before_runmanager
    self.initialize_physics_list()
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/engines.py", line 405, in initialize_physics_list
    self.physics_manager.physics_list_builder.create_physics_list(
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/decorators.py", line 33, in _with_check
    return func(self, *args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/physics.py", line 406, in create_physics_list
    physics_list = self.created_physics_list_classes[physics_list_name](
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/physics.py", line 181, in __init__
    super().__init__(*args, **kwargs)
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/physics.py", line 323, in __init__
    bound_base_class.__init__(self, verbosity, model, variant, use_liqmd)
TypeError: __init__(): incompatible constructor arguments. The following argument types are supported:
    1. opengate_core.opengate_core.Shielding(verbosity: typing.SupportsInt | typing.SupportsIndex = 1, type: opengate_core.opengate_core.G4String = HP)

Invoked with: 1, 'HP', '', True

Shielding_HP
works,

Shielding_HP_EMZ again causes an error:

☠️ Fatal in /opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/physics.py line 416
☠️ Cannot find the physic list: Shielding_HP_EMZ

**** INFO about GATE physics lists ****
* Known Geant4 lists are: [FTFP_BERT, FTFP_BERT_TRV, FTFP_BERT_ATL, FTFP_BERT_HP, FTFQGSP_BERT, FTFP_INCLXX, FTFP_INCLXX_HP, FTF_BIC, LBE, QBBC, QGSP_BERT, QGSP_BERT_HP, QGSP_BIC, QGSP_BIC_HP, QGSP_BIC_AllHP, QGSP_FTFP_BERT, QGSP_INCLXX, QGSP_INCLXX_HP, QGS_BIC, Shielding, ShieldingLEND, ShieldingLIQMD, ShieldingM, NuBeam, Shielding_HP, ShieldingLIQMD_HP, ShieldingM_HP, FTFP_BERT_HPT, FTFP_INCLXX_HPT, QGSP_BERT_HPT, QGSP_BIC_HPT, QGSP_BIC_AllHPT, QGSP_INCLXX_HPT, Shielding_HPT, ShieldingLIQMD_HPT, ShieldingM_HPT]
* With EM options: [_EMV, _EMX, _EMY, _EMZ, _LIV, _PEN, __GS, __SS, _EM0, _WVI, __LE]
* Or the following simple physics lists with a single PhysicsConstructor: 
* ['G4EmStandardPhysics', 'G4EmStandardPhysics_option1', 'G4EmStandardPhysics_option2', 'G4EmStandardPhysics_option3', 'G4EmStandardPhysics_option4', 'G4EmStandardPhysicsGS', 'G4EmLowEPPhysics', 'G4EmLivermorePhysics', 'G4EmLivermorePolarizedPhysics', 'G4EmPenelopePhysics', 'G4OpticalPhysics'] 
**** ----------------------------- ****

Default is ('QGSP_BERT_EMV', {'doc': 'Name of the Geant4 physics list. DNA EM physics must be configured per region, not as a global physics list.', 'setter_hook': <function _setter_hook_physics_list_name at 0x7195fdcb4f40>})
Help : https://geant4-userdoc.web.cern.ch/UsersGuides/PhysicsListGuide/html/physicslistguide.html
⚠️ Method close_physics_constructors of object <opengate.engines.PhysicsEngine object at 0x7195fb5ba120> requires g4_augmented_physics_list to be set, but it is None.
Traceback (most recent call last):
  File "/home/hfuchs/Documents/WORK/2026_Hedgehog/Simulation/DistanceTest/validation_sim_noSTEP/sim_MedAustron_plan_v9_neg2p_3mm_carbon.py", line 660, in <module>
    main()
  File "/home/hfuchs/Documents/WORK/2026_Hedgehog/Simulation/DistanceTest/validation_sim_noSTEP/sim_MedAustron_plan_v9_neg2p_3mm_carbon.py", line 637, in main
    simLink, output_dictionary = simulation(
                                 ^^^^^^^^^^^
  File "/home/hfuchs/Documents/WORK/2026_Hedgehog/Simulation/DistanceTest/validation_sim_noSTEP/sim_MedAustron_plan_v9_neg2p_3mm_carbon.py", line 443, in simulation
    output_sim = sim.run()
                 ^^^^^^^^^
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/managers.py", line 2266, in run
    output = self._run_simulation_engine(False)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/managers.py", line 2224, in _run_simulation_engine
    output = se.run_engine()
             ^^^^^^^^^^^^^^^
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/engines.py", line 1585, in run_engine
    self.initialize()
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/engines.py", line 1765, in initialize
    self.physics_engine.initialize_before_runmanager()
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/engines.py", line 351, in initialize_before_runmanager
    self.initialize_physics_list()
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/engines.py", line 405, in initialize_physics_list
    self.physics_manager.physics_list_builder.create_physics_list(
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/decorators.py", line 33, in _with_check
    return func(self, *args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/physics.py", line 416, in create_physics_list
    fatal(s)
  File "/opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/exception.py", line 39, in fatal
    raise Exception(s)
Exception: Cannot find the physic list: Shielding_HP_EMZ

**** INFO about GATE physics lists ****
* Known Geant4 lists are: [FTFP_BERT, FTFP_BERT_TRV, FTFP_BERT_ATL, FTFP_BERT_HP, FTFQGSP_BERT, FTFP_INCLXX, FTFP_INCLXX_HP, FTF_BIC, LBE, QBBC, QGSP_BERT, QGSP_BERT_HP, QGSP_BIC, QGSP_BIC_HP, QGSP_BIC_AllHP, QGSP_FTFP_BERT, QGSP_INCLXX, QGSP_INCLXX_HP, QGS_BIC, Shielding, ShieldingLEND, ShieldingLIQMD, ShieldingM, NuBeam, Shielding_HP, ShieldingLIQMD_HP, ShieldingM_HP, FTFP_BERT_HPT, FTFP_INCLXX_HPT, QGSP_BERT_HPT, QGSP_BIC_HPT, QGSP_BIC_AllHPT, QGSP_INCLXX_HPT, Shielding_HPT, ShieldingLIQMD_HPT, ShieldingM_HPT]
* With EM options: [_EMV, _EMX, _EMY, _EMZ, _LIV, _PEN, __GS, __SS, _EM0, _WVI, __LE]
* Or the following simple physics lists with a single PhysicsConstructor: 
* ['G4EmStandardPhysics', 'G4EmStandardPhysics_option1', 'G4EmStandardPhysics_option2', 'G4EmStandardPhysics_option3', 'G4EmStandardPhysics_option4', 'G4EmStandardPhysicsGS', 'G4EmLowEPPhysics', 'G4EmLivermorePhysics', 'G4EmLivermorePolarizedPhysics', 'G4EmPenelopePhysics', 'G4OpticalPhysics'] 
**** ----------------------------- ****

Default is ('QGSP_BERT_EMV', {'doc': 'Name of the Geant4 physics list. DNA EM physics must be configured per region, not as a global physics list.', 'setter_hook': <function _setter_hook_physics_list_name at 0x7195fdcb4f40>})
Help : https://geant4-userdoc.web.cern.ch/UsersGuides/PhysicsListGuide/html/physicslistguide.html

opengate_info output:
Python version 3.12.3 (main, Jun 19 2026, 12:46:00) [GCC 13.3.0]
Platform linux
Site package /opt/simulation/env_opengate_release/lib/python3.12/site-packages
Geant4 version geant4-11-04-patch-02 [MT]
Geant4 MT True
Geant4 Qt True 6.6.2
Geant4 GDML True
Geant4 date 12-June-2026
Geant4 data /opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate_core/geant4_data
ITK version 5.4.4
GATE version 10.1.1
GATE folder /opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate
GATE data /opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/data
GATE tests /opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate/tests/src
GATE core path /opt/simulation/env_opengate_release/lib/python3.12/site-packages/opengate_core/opengate_core.cpython-312-x86_64-linux-gnu.so
GATE date 2026-07-21T18:11:00 (pypi)

[GitFuchs]

I just verified, using Gate version 10.1.0 and Geant4 version geant4-11-04 [MT]
the physics list ShieldingLIQMD_HP_EMZ
works as expected

[BishopWolf]

Bug report — issue #1115

ShieldingLIQMD_HP_EMZ (and other Shielding* special lists) broken in release 10.1.1

Verdict: confirmed regression, fully diagnosed, and already fixed on master (PR #1138) — but not in any released wheel. The fix was verified live on this machine after rebuilding opengate_core from source.

Nothing in the repository was modified. All probes were executed from /tmp and in memory; git status --short before and after is unchanged (only the pre-existing ? core/external/fmt submodule build residue). No commits, no file edits, no test artifacts written.

DeepSeek V4.1 flash

---

1. The reported symptoms, and what each one actually is

User observation	Real cause
ShieldingLIQMD_HP_EMZ → Exception: Cannot find the physic list	(A) Name registry never generates EM‑suffixed variants of “special builder” names
ShieldingLIQMD_HP → TypeError: … Shielding(verbosity, type) … Invoked with: 1, 'HP', '', True	(B) Python calls a 4‑argument Shielding constructor that did not exist in the 10.1.1 pybind binding
Shielding_HP works	It has no ctor_args, so it falls through to the 2‑argument call that did exist
Shielding_HP_EMZ → Cannot find the physic list	(A) — same as the first row
⚠️ … close_physics_constructors … requires g4_augmented_physics_list to be set, but it is None	Cascade only — opengate/engines.py:344 warns during teardown because initialize_physics_list() aborted before g4_augmented_physics_list was assigned (engines.py:420). Not an independent bug.
Worked in 10.1.0	In 10.1.0 any unknown name fell back to G4PhysListFactory.GetReferencePhysList(), i.e. Geant4 built the list natively instead of GATE’s Python registry (see §4)

---

2. Environment used for this investigation

Item	Value
HEAD	808099ab5 on branch agentic-skills (contains master incl. the fix; git merge-base --is-ancestor 370bcee57 HEAD → YES)
Python	3.12.13 (/Users/alex/opengate, uv venv)
opengate / opengate_core	10.1.1, editable from the repo
Geant4	geant4-11-04-patch-02 = 11.4.2 [MT], Qt 6.11.2, GDML — matches the CI/AGENTS pin
ITK	6.0.0 (source build linked into opengate_core; the Python itk wheel is 5.4.6 — CI pins 5.4.4)
Rebuilt artifact	core/opengate_core/opengate_core.cpython-312-darwin.so, 6 353 232 B, 23 Sep 09:58 (previous stale build: 5 960 664 B, 20 Jun)
user_secrets.json	present, points at the repo/venv/deps above

Caveat recorded: ITK 6.0.0 ≠ CI pin 5.4.4. This is irrelevant to the defect (entirely GATE-side), but the run is not CI-exact.

---

3. Root cause (A) — registry omits EM‑suffixed “special builder” names

opengate/physics.py, tag 10.1.1 (_build_available_reference_physics_list_names, line 273):

for base_name in reference_physics_list_base_class_names:   # 23 base classes only
    available_names.append(base_name)
    for suffix in em_suffixes:
        if suffix is not None:
            available_names.append(f"{base_name}{suffix}")

available_names.extend(reference_physics_list_special_builders.keys())   # ← verbatim, NO EM suffix

Suffixes are therefore expanded over the 23 base classes but not over the 13 special builders (ShieldingLIQMD_HP, ShieldingLIQMD, ShieldingM, Shielding_HP, *_HPT, …). Consequences:

• ShieldingLIQMD_HP_EMZ, Shielding_HP_EMZ, ShieldingM_EMZ, ShieldingLIQMD_EMZ, … do not exist → create_physics_list() has no else fallback in 10.1.1 and calls fatal("Unknown physics list: …").
• ShieldingLIQMD_HP does exist in the registry, which is exactly why it gets past the name check and dies later with the TypeError instead.

Live reproduction of (A) on this machine

I re-executed the 10.1.1 registry algorithm against the current data and compared with master:

10.1.1-logic registry total: 289
[('ShieldingLIQMD_HP_EMZ', False), ('ShieldingLIQMD_HP', True),
 ('Shielding_HP_EMZ', False),      ('ShieldingM_EMZ', False)]

MASTER registry total: 432
[('ShieldingLIQMD_HP_EMZ', True), ('ShieldingLIQMD_HP', True),
 ('Shielding_HP_EMZ', True),      ('ShieldingLIQMD', True),
 ('ShieldingM_EMZ', True),        ('ShieldingM_HPT_EMZ', True),
 ('QGSP_BIC_HPT__GS', True)]

289 vs 432 = exactly the 13 special builders × 11 EM suffixes = 143 missing names.

---

4. Root cause (B) — 4-argument call against a 2-argument pybind constructor

opengate/physics.py (10.1.1) calls:

"ShieldingLIQMD_HP": {"base": "Shielding", "ctor_args": ("HP", "", True)},
...
model, variant, use_liqmd = special_builder["ctor_args"]
bound_base_class.__init__(self, verbosity, model, variant, use_liqmd)   # 4 args

core/opengate_core/g4_bindings/pyPhysicsLists.cpp (10.1.1) exposed only:

ADD_REFERENCE_PHYSICS_LIST2(Shielding, "HP")   // => py::init<G4int, G4String>()  → (verbosity, type)

Hence the reporter’s exact error (Invoked with: 1, 'HP', '', True). Every special builder carrying ctor_args is affected: ShieldingLIQMD, ShieldingLIQMD_HP, ShieldingM, ShieldingM_HP (+ their _HPT variants). Shielding_HP and Shielding_HPT survive because they only pass verbosity.

---

5. Why 10.1.0 worked and 10.1.1 did not (the actual regression)

This is the important part: the defect is not “_EMZ parsing”; it is the removal of the Geant4 factory fallback.

Tag 10.1.0 — opengate/managers.py:620 (get_physics_list):

if physics_list_name in self.created_physics_list_classes:
    physics_list = ...                       # only the simple single-constructor lists
else:
    g4_factory = g4.G4PhysListFactory()
    if g4_factory.IsReferencePhysList(physics_list_name):
        physics_list = g4_factory.GetReferencePhysList(physics_list_name)   # ← Geant4 builds it natively

created_physics_list_classes in 10.1.0 held only the 13 “simple” G4EmStandardPhysics*-style lists. Anything else — including ShieldingLIQMD_HP_EMZ — went to GetReferencePhysList, which handles every Geant4 name and every _EMZ combination. That is why the reporter’s script worked on 10.1.0. (Consistently, 10.1.0’s pyPhysicsLists.cpp had the reference-list pybind macros inside a commented-out block.)

Tag 10.1.1 — the fallback is gone. Physics-list construction moved into PhysicsListBuilder (opengate/physics.py), which pre-creates Python classes from a static registry and fatal()s on anything not registered. The registry is the only path, and it is incomplete (bug A) and calls an unsupported constructor (bug B).

Regression timeline (all verified from git):

Commit	Date	In 10.1.0?	In 10.1.1?	Effect
202f9b4c4 “Refactor chemistry and physics-list configuration”	2026‑03‑24	no	yes	adds ADD_REFERENCE_PHYSICS_LIST1/2 pybind macros + 2-arg Shielding binding
eeced5409 “Move PhysicsListBuilder and physics-list helpers into physics.py”	2026‑04‑12	no	yes	moves PhysicsListBuilder; introduces special_builders/use_liqmd
002074046 “Fix reference physics list EM suffix handling”	2026‑06‑14	no	yes	introduces bug A: available_names.extend(special_builders.keys()) without EM expansion
370bcee57 “Fix EM suffixes and constructor arguments for special physics lists”	2026‑09‑14	no	no	the fix (A + B)

So the regression was introduced between 10.1.0 and 10.1.1 by the PhysicsListBuilder refactor, and first shipped in the 10.1.1 wheels.

---

6. The fix (already on master, merged as PR #1138 → 45d66f514)

370bcee57 changes both halves:

# opengate/physics.py — EM suffixes now applied to special builders too
for base_name in (*reference_physics_list_base_class_names,
                  *reference_physics_list_special_builders):
    ...
    available_names.append(f"{base_name}{suffix}")     # 289 → 432 names

// core/opengate_core/g4_bindings/pyPhysicsLists.cpp — full Shielding ctor exposed
py::class_<Shielding, G4VModularPhysicsList,
           std::unique_ptr<Shielding, py::nodelete>>(m, "Shielding")
    .def(py::init<G4int, const G4String &, const G4String &, G4bool>(),
         py::arg("verbosity") = 1, py::arg("type") = G4String("HP"),
         py::arg("variant") = G4String(""), py::arg("use_liqmd") = false);

Companions: 6a783d4a0 adds __SS/_WVI; 3c2ad3c87 rewrites the error dump to list Builder-accepted names, adds an import-time registry-drift warning, and removes the obsolete GetReferencePhysList binding. Regression coverage in opengate/tests/src/physics/test099_physics_list_registry.py::check_special_reference_lists_with_em_options() constructs all 13 × (1 + 11) = 156 special-builder × EM-suffix combinations.

---

7. Live verification on the rebuilt environment (all green)

Binding signature now correct (this is precisely the call that produced the reported TypeError):

__init__(self, verbosity: … = 1, type: … = HP, variant: … = , use_liqmd: bool = False) -> None

Construction of the reported lists (G4RunManager + Simulation, verbosity 0):

ShieldingLIQMD_HP_EMZ  cls=ShieldingLIQMD_HP_EMZ  em0=G4EmStandardPhysics_option4  LIQMD=True   ThermN=False
ShieldingLIQMD_HP      cls=ShieldingLIQMD_HP      em0=G4EmStandardPhysics          LIQMD=True   ThermN=False
Shielding_HP_EMZ       cls=Shielding_HP_EMZ       em0=G4EmStandardPhysics_option4  LIQMD=False  ThermN=False
ShieldingLIQMD         cls=ShieldingLIQMD         em0=G4EmStandardPhysics          LIQMD=True   ThermN=False
ShieldingM             cls=ShieldingM             em0=G4EmStandardPhysics          LIQMD=False  ThermN=False
Shielding_HPT          cls=Shielding_HPT          em0=G4EmStandardPhysics          LIQMD=False  ThermN=True

Every expected sub-physics is present and correct (_EMZ → G4EmStandardPhysics_option4; LIQMD variants expose LightIonQMD; _HPT exposes ThermalNeutrons).

Registry alignment with the linked Geant4:

registry drift vs linked Geant4 (11.4.2): NONE

So on current master the reporter’s script would run unchanged.

---

8. Release status and what to tell the user in the issue

• The fix commits (370bcee57, 6a783d4a0, 3c2ad3c87) are in no tag — git tag --contains 370bcee57 is empty. The PyPI 10.1.1 wheels still contain the bug.
• There is no Python‑only workaround for the LIQMD variants on 10.1.1: the shipped opengate_core binary only accepts (verbosity, type), so use_liqmd=True is unrepresentable.
• Practical options: pip install opengate==10.1.0 (confirmed working by the reporter), or install/build master, or wait for the next release (both opengate and opengate_core must be rebuilt — the fix is partly C++).
• Upstream issue Physics list ShieldingLIQMD_HP_EMZ not working any more with official OpenGate10 release 10.1.1 #1115 is still Open even though the merged PR says it would close an issue; it needs closing/linking once a release ships.

---

9. Notes, limitations, and things I deliberately did not do

• I did not run opengate_tests -t physics/test099_physics_list_registry.py, because the runner writes opengate/tests/log/ and opengate/tests/output*/ artifacts and you asked for no modifications. I reproduced its essential checks in memory instead; physics/test099_physics_list_registry.py remains the authoritative automated guard.
• I could not reproduce the 10.1.1 TypeError with the machine's two older .so files (cpython-313/314, 9 May): they predate the reference-list pybind refactor entirely (import opengate_core; opengate_core.Shielding → AttributeError). The regression evidence for (B) therefore comes from git (tag 10.1.1) plus the reporter’s own traceback, and is confirmed fixed by the live 4-argument call today.
• The 18 test scripts in opengate/tests/src/ that set physics_list_name to *_EMV/_EMY/_EMZ etc. only exercise base lists, which is why CI did not catch this — the special builders had no EM-suffix coverage at 10.1.1. test099 now fills that gap.
• Environment caveats: ITK built from source is 6.0.0 (CI pin 5.4.4); the cpython-313/-314 .so files in the tree are stale and unrelated.

10. One-line summary

ShieldingLIQMD_HP_EMZ broke in 10.1.1 because the 10.1.1 refactor replaced Geant4’s GetReferencePhysList() fallback with a Python registry that (A) never generated EM-suffixed variants of the 13 “special” Shielding/HPT lists and (B) invoked a 4-argument Shielding constructor the shipped pybind binding didn’t expose; both halves are fixed by PR #1138 on master (verified green here on Geant4 11.4.2), but that fix is not in any released wheel yet.

[BishopWolf]

@nkrah The bug report says this issue can be closed as the fix is already merged