Unimplemented Python Features

Introduction

CSXCAD and openEMS began as C++ programs driven by users via Matlab/Octave, the Python bindings came relatively late in the project’s history. As a result, not all C++ and Matlab/Octave features are currently implemented in Python.

One can classify these missing features into two categories.

The C++ API was not linked to Python.

Solving this problem is relatively easy. One can add the missing C++ API binding without reimplementing any logic.

Both developers and ordinary openEMS users have contributed high-level pre-processing and post-processing features, but only specific to Matlab/Octave. Examples include numerical fitting code and model exporting code.

This problem is more difficult to solve, as the same logic must be reimplemented in Python.

The following list of missing Python features is meant for reference by developers and end users.

Known Problems

Some model importing and exporting functions are unimplemented

Type 2: Missing high-level feature.
Affected APIs: ImportPLY(), ImportSTL(), export_gerber(), export_excellon(), export_povray().
Workaround: To import an STL file, use AddPolyhedronReader(). To export models, use AppCSXCAD.

Mur ABC phase velocity parameter adjustment is unimplemented

Type 1: Missing C++ binding.
Affected APIs: SetBoundaryCond()’s optional argument MUR_PhaseVelocity is unimplemented. Mur’s ABC cannot be further optimized by tuning MUR_PhaseVelocity if the boundary doesn’t end at a vacuum.
Workaround: None.
Comment: The underlying Set_Mur_PhaseVel() is already declared in openEMS/python/openEMS/openEMS.pxd, it is only never called. The missing piece is an optional argument of SetBoundaryCond() that forwards it. Note that this is unrelated to the local absorbing boundary property CSPropAbsorbingBC, whose PhaseVelocity is available in Python.

Dispersive material fitting helpers are unimplemented

Type 2: Missing high-level feature.
Affected APIs: AddDjordjevicSarkarMaterial(), CalcDebyeMaterial(), CalcDjordjevicSarkarApprox(), CalcDrudeMaterial(), CalcLorentzMaterial().
Workaround: None. These Octave functions are helper functions that calculate the model’s output curves for the purpose of fitting parameters and preparing a simulation. No C++ APIs exist because they’re not actually used in the simulation.
Comment: The Debye and Lorentz materials themselves are available in Python as CSPropDebyeMaterial and CSPropLorentzMaterial, including their model parameters. Only the fitting helpers listed above, and the Djordjevic-Sarkar model built on top of them, are missing. See Dispersive Materials.

Delay fidelity post-processing for UWB systems is unimplemented

Type 2: Missing high-level feature.
Affected API: DelayFidelity()
Workaround: None.
Comment: This is a specialized post-processing function involved in the design of ultra-wideband radios and radars. In these applications it is important to know the delay and fidelity of RF pulses. The delay is the retardation of the signal from the source to the phase center of the antenna. It is composed out of linear delay, dispersion and minimum-phase delay. Dispersion due to waveguides or frequency-dependent permittivity and minimum-phase delay due to resonances will degrade the fidelity which is the normalized similarity between excitation and radiated signal.

CTB and scikit-rf Post-Processing

You may encounter Circuit Toolbox (CTB) in openEMS simulations, which is Matlab/Octave exclusive. However, this is not a “missing feature”.

CTB is developed by openEMS’s author Thorsten Liebig and is used with openEMS in some examples, but it’s an independent library outside openEMS’s codebase. It contains network parameter calculation functions to help analyzing simulation outputs, but they’re themselves not part of the simulator.

To analyze RF circuits in Python, use other Python RF engineering libraries, such as scikit-rf. This is a 3rd-party project not associated with openEMS.

Conversely, scikit-rf contains many sophisticated calibration, de-embedding and signal transform algorithms which represented multiple years of work. If you encounter openEMS examples with scikit-rf, Matlab/Octave alternatives would be even less straightforward to find.