setupEM 0.6.1__tar.gz → 0.6.4__tar.gz

{setupem-0.6.1 → setupem-0.6.4}/PKG-INFO

@@ -1,14 +1,15 @@
 Metadata-Version: 2.4
 Name: setupEM
-Version: 0.6.1
+Version: 0.6.4
 Summary: Python tool for configuration of gds2palace workflow with GUI.
 Author-email: Volker Muehlhaus <volker@muehlhaus.com>
 License-Expression: GPL-3.0-or-later
 Requires-Python: >=3.9
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: gds2palace>=0.4.0
+Requires-Dist: gds2palace>=0.4.3
 Requires-Dist: PySide6
+Requires-Dist: shiboken6
 Requires-Dist: scipy
 Requires-Dist: requests
 Requires-Dist: scikit-rf
@@ -28,6 +29,7 @@ setupEM includes two built-in tools for working with simulation results directly
 - **Model Fit** (Create Model tab > Model Fit...) launches [snp2le](https://github.com/iic-jku/snp2le), an external open-source tool that extracts a lumped-element SPICE/Spectre netlist from S-parameter results - offering to install it via pip if it isn't already present. See section "[Model Fit](#model-fit)".
 
 - A graphical **Stackup XML Editor** (Tools > Edit Stackup XML...), including Variables, Reference-relative positioning, Derived Layers, and Thermal Tables
+- The stackup preview graphics are **interactive**: click a shape for its properties, and selection syncs both ways with the Stackup Editor's tables
 - Input Files tab can now **override stackup Variables** (e.g. `total_thickness`) directly, without hand-editing the XML or the generated model
 - **setupThermal**, a companion app for Elmer thermal simulation, alongside setupEM
 
@@ -43,6 +45,11 @@ The setupEM package now includes setupThermal also, which is the equivalent of s
 
 An overview of the SetupEM user interface is given below in chapter "Using setupEM"
 
+Two more external tools are used by parts of the workflow, and are not installed automatically:
+
+- [ParaView](https://www.paraview.org/) — to view field-dump output (Palace/Elmer EM) and Elmer thermal result files via the "View fields/results in Paraview" buttons.
+- An MPI implementation — only needed for multi-process Elmer runs (the Elmer solver settings' multithreading option). Use OpenMPI or MPICH on Linux/macOS; on Windows, install [Microsoft MPI](https://learn.microsoft.com/en-us/message-passing-interface/microsoft-mpi) (setupEM checks for this and shows a download link if it's missing).
+
 
 ## Installing the AWS Palace FEM solver engine
 **setupEM** creates and runs simulation models for the AWS Palace FEM solver engine. The underlying solver **AWS Palace** can be installed in multiple ways. For a smooth interaction with the gds2palace workflow, it is recommended to create some scripts that help running the model and convert the Palace results to SnP Touchstone files.
@@ -102,10 +109,14 @@ sudo apt install libxcb-cursor0 libxcb-xinerama0 libxcb-xkb1 libxcb-icccm4 libxc
 The setupEM module also installs these dependencies:
 - gds2palace
 - PySide6
+- shiboken6
 - scipy
 - requests
 - scikit-rf
 - matplotlib
+- numpy
+- gdspy
+- meshio
     
 ---
 
@@ -322,13 +333,15 @@ This project is licensed under the GNU General Public License v3.0 or later (GPL
 
 **Note:** the `setupEM` PyPI package itself requires:
 
-- gds2palace>=0.4.0
+- gds2palace>=0.4.3
 - PySide6
+- shiboken6
 - scipy
 - requests
 - scikit-rf
 - matplotlib
 - numpy
 - gdspy
+- meshio
 
 (Other Python modules mentioned above are only needed to run standalone helper scripts in this repository, not to use the installed package.)

{setupem-0.6.1 → setupem-0.6.4}/README.md

@@ -8,6 +8,7 @@ setupEM includes two built-in tools for working with simulation results directly
 - **Model Fit** (Create Model tab > Model Fit...) launches [snp2le](https://github.com/iic-jku/snp2le), an external open-source tool that extracts a lumped-element SPICE/Spectre netlist from S-parameter results - offering to install it via pip if it isn't already present. See section "[Model Fit](#model-fit)".
 
 - A graphical **Stackup XML Editor** (Tools > Edit Stackup XML...), including Variables, Reference-relative positioning, Derived Layers, and Thermal Tables
+- The stackup preview graphics are **interactive**: click a shape for its properties, and selection syncs both ways with the Stackup Editor's tables
 - Input Files tab can now **override stackup Variables** (e.g. `total_thickness`) directly, without hand-editing the XML or the generated model
 - **setupThermal**, a companion app for Elmer thermal simulation, alongside setupEM
 
@@ -23,6 +24,11 @@ The setupEM package now includes setupThermal also, which is the equivalent of s
 
 An overview of the SetupEM user interface is given below in chapter "Using setupEM"
 
+Two more external tools are used by parts of the workflow, and are not installed automatically:
+
+- [ParaView](https://www.paraview.org/) — to view field-dump output (Palace/Elmer EM) and Elmer thermal result files via the "View fields/results in Paraview" buttons.
+- An MPI implementation — only needed for multi-process Elmer runs (the Elmer solver settings' multithreading option). Use OpenMPI or MPICH on Linux/macOS; on Windows, install [Microsoft MPI](https://learn.microsoft.com/en-us/message-passing-interface/microsoft-mpi) (setupEM checks for this and shows a download link if it's missing).
+
 
 ## Installing the AWS Palace FEM solver engine
 **setupEM** creates and runs simulation models for the AWS Palace FEM solver engine. The underlying solver **AWS Palace** can be installed in multiple ways. For a smooth interaction with the gds2palace workflow, it is recommended to create some scripts that help running the model and convert the Palace results to SnP Touchstone files.
@@ -82,10 +88,14 @@ sudo apt install libxcb-cursor0 libxcb-xinerama0 libxcb-xkb1 libxcb-icccm4 libxc
 The setupEM module also installs these dependencies:
 - gds2palace
 - PySide6
+- shiboken6
 - scipy
 - requests
 - scikit-rf
 - matplotlib
+- numpy
+- gdspy
+- meshio
     
 ---
 

setupem-0.6.1/src/setupEM.egg-info/PKG-INFO → setupem-0.6.4/README_pypi.md

@@ -1,23 +1,3 @@
-Metadata-Version: 2.4
-Name: setupEM
-Version: 0.6.1
-Summary: Python tool for configuration of gds2palace workflow with GUI.
-Author-email: Volker Muehlhaus <volker@muehlhaus.com>
-License-Expression: GPL-3.0-or-later
-Requires-Python: >=3.9
-Description-Content-Type: text/markdown
-License-File: LICENSE
-Requires-Dist: gds2palace>=0.4.0
-Requires-Dist: PySide6
-Requires-Dist: scipy
-Requires-Dist: requests
-Requires-Dist: scikit-rf
-Requires-Dist: matplotlib
-Requires-Dist: numpy
-Requires-Dist: gdspy
-Requires-Dist: meshio
-Dynamic: license-file
-
 # Python GUI for gds2palace 
 
 ## What's New
@@ -28,6 +8,7 @@ setupEM includes two built-in tools for working with simulation results directly
 - **Model Fit** (Create Model tab > Model Fit...) launches [snp2le](https://github.com/iic-jku/snp2le), an external open-source tool that extracts a lumped-element SPICE/Spectre netlist from S-parameter results - offering to install it via pip if it isn't already present. See section "[Model Fit](#model-fit)".
 
 - A graphical **Stackup XML Editor** (Tools > Edit Stackup XML...), including Variables, Reference-relative positioning, Derived Layers, and Thermal Tables
+- The stackup preview graphics are **interactive**: click a shape for its properties, and selection syncs both ways with the Stackup Editor's tables
 - Input Files tab can now **override stackup Variables** (e.g. `total_thickness`) directly, without hand-editing the XML or the generated model
 - **setupThermal**, a companion app for Elmer thermal simulation, alongside setupEM
 
@@ -43,6 +24,11 @@ The setupEM package now includes setupThermal also, which is the equivalent of s
 
 An overview of the SetupEM user interface is given below in chapter "Using setupEM"
 
+Two more external tools are used by parts of the workflow, and are not installed automatically:
+
+- [ParaView](https://www.paraview.org/) — to view field-dump output (Palace/Elmer EM) and Elmer thermal result files via the "View fields/results in Paraview" buttons.
+- An MPI implementation — only needed for multi-process Elmer runs (the Elmer solver settings' multithreading option). Use OpenMPI or MPICH on Linux/macOS; on Windows, install [Microsoft MPI](https://learn.microsoft.com/en-us/message-passing-interface/microsoft-mpi) (setupEM checks for this and shows a download link if it's missing).
+
 
 ## Installing the AWS Palace FEM solver engine
 **setupEM** creates and runs simulation models for the AWS Palace FEM solver engine. The underlying solver **AWS Palace** can be installed in multiple ways. For a smooth interaction with the gds2palace workflow, it is recommended to create some scripts that help running the model and convert the Palace results to SnP Touchstone files.
@@ -102,10 +88,14 @@ sudo apt install libxcb-cursor0 libxcb-xinerama0 libxcb-xkb1 libxcb-icccm4 libxc
 The setupEM module also installs these dependencies:
 - gds2palace
 - PySide6
+- shiboken6
 - scipy
 - requests
 - scikit-rf
 - matplotlib
+- numpy
+- gdspy
+- meshio
     
 ---
 
@@ -322,13 +312,15 @@ This project is licensed under the GNU General Public License v3.0 or later (GPL
 
 **Note:** the `setupEM` PyPI package itself requires:
 
-- gds2palace>=0.4.0
+- gds2palace>=0.4.3
 - PySide6
+- shiboken6
 - scipy
 - requests
 - scikit-rf
 - matplotlib
 - numpy
 - gdspy
+- meshio
 
 (Other Python modules mentioned above are only needed to run standalone helper scripts in this repository, not to use the installed package.)

{setupem-0.6.1 → setupem-0.6.4}/pyproject.toml

@@ -12,8 +12,9 @@ authors = [
     { name = "Volker Muehlhaus", email = "volker@muehlhaus.com" }
 ]
 dependencies = [
-    "gds2palace>=0.4.0",
+    "gds2palace>=0.4.3",
     "PySide6",
+    "shiboken6",
     "scipy",
     "requests",
     "scikit-rf",

{setupem-0.6.1 → setupem-0.6.4}/src/setupEM/__init__.py

@@ -24,4 +24,4 @@ A Python tool for EM setup using gds2palace.
 from .setupEM import main
 
 __all__ = ["main"]
-__version__ = "0.6.1"
+__version__ = "0.6.4"

{setupem-0.6.1 → setupem-0.6.4}/src/setupEM/palace_results.py

@@ -53,8 +53,10 @@ def find_paraview_files(run_path, model_basename):
     recursively. Palace's own field-dump feature (triggered by fdump/SaveStep) writes
     its category/excitation folder structure without gds2palace's control - confirmed
     (on two Palace versions) to only keep the last-solved excitation's Cycle files, so
-    no fixed filename or count can be assumed here. Falls back to loose .vtu if no .pvd
-    exists yet. Empty list if the output directory doesn't exist / has nothing yet.
+    no fixed filename or count can be assumed here. Prefers .pvd, then falls back to
+    .pvtu (the multi-partition index - ties per-partition pieces back into one dataset),
+    then loose .vtu if neither exists yet. Empty list if the output directory doesn't
+    exist / has nothing yet.
     """
     output_dir = find_output_dir(run_path, model_basename)
     if not os.path.isdir(output_dir):
@@ -62,6 +64,9 @@ def find_paraview_files(run_path, model_basename):
     pvd_files = sorted(glob.glob(os.path.join(output_dir, '**', '*.pvd'), recursive=True))
     if pvd_files:
         return pvd_files
+    pvtu_files = sorted(glob.glob(os.path.join(output_dir, '**', '*.pvtu'), recursive=True))
+    if pvtu_files:
+        return pvtu_files
     return sorted(glob.glob(os.path.join(output_dir, '**', '*.vtu'), recursive=True))
 
 

{setupem-0.6.1 → setupem-0.6.4}/src/setupEM/setupEM.py

@@ -165,10 +165,21 @@ class FrequenciesTab(QWidget):
         self.dump_layout = QHBoxLayout()
 
         self.fdump_layout = QVBoxLayout()
-        self.fdump_layout.addWidget(QLabel("fdump [GHz], values separated by comma"))
+        self.fdump_label = QLabel("fdump [GHz], values separated by comma")
+        self.fdump_layout.addWidget(self.fdump_label)
         self.fdump_edit = QLineEdit("")
         self.fdump_edit.setStyleSheet(EDIT_STYLE_OPTIONAL)
         self.fdump_layout.addWidget(self.fdump_edit)
+
+        # Elmer has no per-sample SaveStep like Palace - any fdump value turns on
+        # field-dump output at every solved frequency (sweep + fpoint together), so a
+        # per-frequency list is misleading there. Elmer mode shows this checkbox instead
+        # of fdump_edit; only one of the two is ever visible, toggled in setPalaceMode()/
+        # setElmerMode(). Both widgets always exist so mode switches are just a visibility
+        # flip, no rebuilding.
+        self.fdump_enabled_checkbox = QCheckBox("Enable field dump at all frequencies")
+        self.fdump_layout.addWidget(self.fdump_enabled_checkbox)
+
         self.dump_layout.addLayout(self.fdump_layout)
         self.dump_group.setLayout(self.dump_layout)
 
@@ -184,7 +195,9 @@ class FrequenciesTab(QWidget):
     def save_values(self):
         # fstart/fstop may be left empty only if fpoint or fdump supplies at least
         # one frequency instead (gds2palace treats fstart/fstop as fully optional)
-        discrete_freqs_given = (self.fpoint_edit.text() != "" or self.fdump_edit.text() != "")
+        fdump_given = (self.fdump_enabled_checkbox.isChecked() if self.MainWindow.ElmerMode
+                       else self.fdump_edit.text() != "")
+        discrete_freqs_given = (self.fpoint_edit.text() != "" or fdump_given)
 
         fstart = None
         if self.start_edit.text() == "":
@@ -239,18 +252,35 @@ class FrequenciesTab(QWidget):
         else:
             saved_values.pop("fpoint",None) # delete key
 
-        text = self.fdump_edit.text()
-        if text != "":
-            saved_values ["fdump"] = ast.literal_eval('['+text+']')
+        if self.MainWindow.ElmerMode:
+            saved_values["fdump_enabled"] = self.fdump_enabled_checkbox.isChecked()
+            # don't touch saved_values["fdump"] here - it's Palace-only data (a real
+            # frequency list) and must survive a round-trip through Elmer mode untouched
+            have_sweep = ("fstart" in saved_values and "fstop" in saved_values)
+            have_fpoint = bool(saved_values.get("fpoint"))
+            if self.fdump_enabled_checkbox.isChecked() and not have_sweep and not have_fpoint:
+                QMessageBox.warning(self, "Error",
+                    "Field dump needs at least one frequency: set fstart/fstop or fpoint first.")
+                self.fdump_enabled_checkbox.setChecked(False)
+                saved_values["fdump_enabled"] = False
+                return False
         else:
-            saved_values.pop("fdump",None)
+            text = self.fdump_edit.text()
+            if text != "":
+                saved_values ["fdump"] = ast.literal_eval('['+text+']')
+            else:
+                saved_values.pop("fdump",None)
+            # don't touch saved_values["fdump_enabled"] here either, same reasoning
         # "View fields in Paraview" is only relevant once fdump is set
         self.MainWindow.create_model_tab._update_paraview_button_visibility()
 
         # if fstart == fstop == fdump or fstart == fstop == fstep, then remove fstart, fstop
         if fstart is not None and fstop is not None and fstart == fstop:
             discrete_list1 = saved_values.get("fpoint", [])
-            discrete_list2 = saved_values.get("fdump", [])
+            # fdump no longer holds real per-frequency data in Elmer mode, and collapsing
+            # the sweep away here would remove the very fstop value the field-dump
+            # checkbox's generated code depends on (see ModelEditorTab.create_model_text)
+            discrete_list2 = [] if self.MainWindow.ElmerMode else saved_values.get("fdump", [])
             if fstart in discrete_list1 or fstart in discrete_list2:
                 saved_values.pop("fstart", None)
                 saved_values.pop("fstop", None)
@@ -265,7 +295,9 @@ class FrequenciesTab(QWidget):
     def load_values(self):
         # if fstart/fstop were intentionally omitted in favor of fpoint/fdump, keep the
         # fields blank on reload instead of repopulating the "0"/"50" sweep defaults
-        discrete_freqs_given = ("fpoint" in saved_values) or ("fdump" in saved_values)
+        fdump_given = saved_values.get("fdump_enabled", False) if self.MainWindow.ElmerMode \
+                      else ("fdump" in saved_values)
+        discrete_freqs_given = ("fpoint" in saved_values) or fdump_given
         fstart_default = "" if discrete_freqs_given else "0"
         fstop_default  = "" if discrete_freqs_given else "50"
         self.start_edit.setText(str(saved_values.get("fstart",fstart_default)))
@@ -277,6 +309,7 @@ class FrequenciesTab(QWidget):
 
         float_list  = saved_values.get("fdump","")
         self.fdump_edit.setText(','.join(map(str, float_list)))
+        self.fdump_enabled_checkbox.setChecked(bool(saved_values.get("fdump_enabled", False)))
 
 
 class PortsTab(QWidget):
@@ -1386,7 +1419,11 @@ class CreateModelTab(CreateModelTabBase):
         self.log_area.appendPlainText("\n" + summary + "\n")
 
     def _update_paraview_button_visibility(self):
-        self.paraview_btn.setVisible(bool(saved_values.get('fdump')))
+        if self.MainWindow.ElmerMode:
+            visible = bool(saved_values.get('fdump_enabled'))
+        else:
+            visible = bool(saved_values.get('fdump'))
+        self.paraview_btn.setVisible(visible)
 
     def load_values(self):
         super().load_values()
@@ -1409,7 +1446,7 @@ class CreateModelTab(CreateModelTabBase):
             # found for thermal_results.vtu). Check run_path too, defensively.
             search_dirs = [os.path.join(run_path, "mesh"), run_path]
             file_paths = []
-            for pattern in ("fields*.pvd", "fields*.vtu"):
+            for pattern in ("fields*.pvd", "fields*.pvtu", "fields*.vtu"):
                 for d in search_dirs:
                     file_paths = sorted(glob.glob(os.path.join(d, pattern)))
                     if file_paths:
@@ -1418,7 +1455,7 @@ class CreateModelTab(CreateModelTabBase):
                     break
             not_found = (
                 f"⚠️ No Elmer field-dump output found under {run_path}\n"
-                "(set fdump to specific frequencies before running the simulation)\n"
+                "(enable field dump before running the simulation)\n"
             )
         self._open_in_paraview(file_paths, not_found)
 
@@ -1731,13 +1768,22 @@ class ModelEditorTab(QWidget):
         special_keylist = ['simulation_ports','materials_list','dielectrics_list','metals_list',
                            'layernumbers','allpolygons']
         # List of keys that we don't write to Python model code editor
-        ignore_list     = ['model_basename','sim_path']
+        # fdump_enabled is a GUI-only toggle (Elmer mode), never a real gds2palace setting
+        ignore_list     = ['model_basename','sim_path','fdump_enabled']
 
 
         # Keywords that are excluded in Palace mode
         if self.MainWindow.PalaceMode:
             ignore_list.append('iterative')
 
+        if self.MainWindow.ElmerMode:
+            # Elmer mode shows a plain on/off checkbox (fdump_enabled) instead of Palace's
+            # per-frequency fdump list - suppress the generic per-key emission of fdump
+            # here so a stale Palace-mode list saved earlier in the same session doesn't
+            # leak into an Elmer-mode script; the synthesized settings['fdump'] line
+            # (below, after special_keylist) is emitted instead when the checkbox is on.
+            ignore_list.append('fdump')
+
         if forExport:
             # these commands are only used within this GUI application to control gmsh
             ignore_list.extend(['preview_only','no_preview'])
@@ -1747,6 +1793,23 @@ class ModelEditorTab(QWidget):
                 if not key in ignore_list:
                     add_key(key)
 
+        if self.MainWindow.ElmerMode and bool(saved_values.get('fdump_enabled')):
+            # Elmer has no per-frequency SaveStep like Palace - it dumps fields at every
+            # solved frequency once any fdump value is set, so reuse an already-solved
+            # frequency here rather than asking the user to pick one. This is harmless
+            # (adds no extra solve) as long as util_elmer.write_elmer_frequencies()
+            # dedupes the combined frequency list, which it does.
+            # Known limitation: this line's RHS references "settings", so
+            # parse_assignments() (setup_common.py) skips it entirely on Import -
+            # re-importing an exported Elmer+fdump script loses fdump_enabled (loads
+            # unchecked). Not fixed here since it's a silent no-op, not a crash.
+            add_text("\n# 'Enable field dump' checkbox: Elmer dumps fields at every solved")
+            add_text("# frequency once fdump is non-empty, so reuse an already-solved one")
+            if 'fstop' in saved_values:
+                add_text("settings['fdump'] = [settings['fstop']]")
+            elif saved_values.get('fpoint'):
+                add_text("settings['fdump'] = [settings['fpoint'][0]]")
+
 
         add_text("\n# ===================== port definitions =======================")
         add_text("simulation_ports = simulation_setup.all_simulation_ports()")
@@ -1951,6 +2014,10 @@ class MainWindow(MainWindowBase):
         self.ElmerMode  = False
         self.setWindowTitle(APP_NAME + ' Palace')
         self.frequencies_tab.dump_group.setVisible(True)
+        self.frequencies_tab.dump_group.setTitle("Optional list of fixed frequencies creating field dump data for visualization (Paraview files))")
+        self.frequencies_tab.fdump_label.setVisible(True)
+        self.frequencies_tab.fdump_edit.setVisible(True)
+        self.frequencies_tab.fdump_enabled_checkbox.setVisible(False)
         self.mesh_tab.AMR_group.setVisible(True)
         self.mesh_tab.Elmer_group.setVisible(False)
 
@@ -1963,9 +2030,14 @@ class MainWindow(MainWindowBase):
         self.PalaceMode = False
         self.ElmerMode  = True
         self.setWindowTitle(APP_NAME + ' Elmer')
-        # fdump now also enables Elmer EM field dumps (Exec Solver = Always), not just
-        # Palace's SaveStep - so this group must stay visible/usable in Elmer mode too.
+        # Elmer has no per-sample SaveStep like Palace - any fdump value turns on
+        # field-dump output at every solved frequency, so it shows a plain on/off
+        # checkbox here instead of Palace's per-frequency fdump list.
         self.frequencies_tab.dump_group.setVisible(True)
+        self.frequencies_tab.dump_group.setTitle("Create field dump data for visualization (Paraview files)")
+        self.frequencies_tab.fdump_label.setVisible(False)
+        self.frequencies_tab.fdump_edit.setVisible(False)
+        self.frequencies_tab.fdump_enabled_checkbox.setVisible(True)
         self.mesh_tab.AMR_group.setVisible(False)
         self.mesh_tab.Elmer_group.setVisible(True)
 
@@ -1995,8 +2067,8 @@ class MainWindow(MainWindowBase):
 
 
     def show_version(self):
-        setupEM_version = importlib.metadata.version("setupEM")
-        gds2palace_version = importlib.metadata.version("gds2palace")
+        setupEM_version = self.get_setupEM_version()
+        gds2palace_version = self.get_gds2palace_version()
         # snp2le (used by Model Fit) is an optional dependency, not installed by default
         try:
             snp2le_version = importlib.metadata.version("snp2le")

{setupem-0.6.1 → setupem-0.6.4}/src/setupEM/setupThermal.py

@@ -53,14 +53,14 @@ if __package__ in (None, ""):
         FileDropLineEdit, FileInputTab, PythonHighlighter, CodeEditor,
         VectorWidget, PopUpWindow, CreateModelTabBase, MainWindowBase,
     )
-    from thermal_results import build_thermal_summary, format_source_table, find_thermal_vtu
+    from thermal_results import build_thermal_summary, format_source_table, find_thermal_paraview_file
 else:
     from .setup_common import (
         EDIT_STYLE_OPTIONAL, EDIT_STYLE_REQUIRED, COMBO_STYLE_REQUIRED, COMBO_STYLE_OPTIONAL,
         FileDropLineEdit, FileInputTab, PythonHighlighter, CodeEditor,
         VectorWidget, PopUpWindow, CreateModelTabBase, MainWindowBase,
     )
-    from .thermal_results import build_thermal_summary, format_source_table, find_thermal_vtu
+    from .thermal_results import build_thermal_summary, format_source_table, find_thermal_paraview_file
 
 
 '''
@@ -501,6 +501,28 @@ class MeshTab(QWidget):
         self.main_layout.addWidget(self.mesh_group)
 
 
+        # ---------- ELMER SOLVER GROUP ----------
+        self.Elmer_group = QGroupBox("Elmer solver settings")
+        self.Elmer_layout = QVBoxLayout()
+
+        self.solver_layout = QHBoxLayout()
+        self.solverlabel = QLabel("Solver")
+        self.solverlabel.setFixedWidth(label_width)
+        self.solver_layout.addWidget(self.solverlabel)
+
+        self.solver_box = QComboBox()
+        self.solver_box.setFixedWidth(edit_width)
+        self.solver_box.setStyleSheet(COMBO_STYLE_OPTIONAL)
+        self.solver_box.addItems(["iterative","direct"])
+        self.solver_layout.addWidget(self.solver_box)
+        self.solver_box.setCurrentIndex(1)
+        self.solver_layout.addStretch()
+        self.Elmer_layout.addLayout(self.solver_layout)
+
+        self.Elmer_group.setLayout(self.Elmer_layout)
+        self.main_layout.addWidget(self.Elmer_group)
+
+
         # ---------- BOUNDARY GROUP ----------
         self.mesh_group = QGroupBox("Boundary settings")
         self.mesh_layout = QVBoxLayout()
@@ -551,6 +573,9 @@ class MeshTab(QWidget):
             return False
         saved_values ["margin"] = float(value)
 
+        # iterative or direct solver for Elmer
+        saved_values["iterative"] = "iterative" in self.solver_box.currentText()
+
         # all saved
         return True
 
@@ -560,6 +585,11 @@ class MeshTab(QWidget):
         self.cells_maxsize_edit.setText(str(saved_values.get("meshsize_max","100")))
         self.margins_edit.setText(str(saved_values.get("margin","100")))
 
+        if saved_values.get("iterative", False):
+            self.solver_box.setCurrentIndex(0)
+        else:
+            self.solver_box.setCurrentIndex(1)
+
 
 
 class CreateModelTab(CreateModelTabBase):
@@ -615,7 +645,7 @@ class CreateModelTab(CreateModelTabBase):
 
     def launch_paraview(self):
         run_path = saved_values['sim_path'] + "/elmer_model/" + saved_values['model_basename'] + "_data"
-        vtu_path = find_thermal_vtu(run_path)
+        vtu_path = find_thermal_paraview_file(run_path)
         self._open_in_paraview(
             [vtu_path] if vtu_path else [],
             f"⚠️ No thermal results .vtu found yet under {run_path}\n"
@@ -786,7 +816,7 @@ class ModelEditorTab(QWidget):
         special_keylist = ['thermal_objects','materials_list','dielectrics_list','metals_list',
                            'layernumbers','allpolygons']
         # List of keys that we don't write to Python model code editor
-        ignore_list     = ['model_basename','sim_path','iterative']
+        ignore_list     = ['model_basename','sim_path']
 
 
         if forExport:
@@ -935,8 +965,8 @@ class MainWindow(MainWindowBase):
     # ---------- Menu actions ----------
 
     def show_version(self):
-        setupEM_version = importlib.metadata.version("setupEM")
-        gds2palace_version = importlib.metadata.version("gds2palace")
+        setupEM_version = self.get_setupEM_version()
+        gds2palace_version = self.get_gds2palace_version()
         version_info = f"Installed:\nsetupEM {setupEM_version}\ngds2palace {gds2palace_version}"
 
         # get latest available version information