Selected changes from the previous release (2024S)
New features:
* define custom math expressions via postProcList and use them in couplings (37b95f46)
* artificial diffusion and SUPG stabilization for the HeatPDE (09ffcc6e, 2b4fc593, 10e9452a)
* thinLayer interface condition in HeatPDE (000c90a1)
* GetRidOfZeros function to remove unnecessary zeros created by NCIs from the system matrix before solving (18c6118b)
* Jérémie Dumas C++ MMA implementations (f067b14f)
* Ginkgo solver package with CUDA support (430e8083, ba5c64ff, dfa3e0d5, 4fc9e5df, ece50f4e)
* acoustic topology optimization (1469652c, 45895d27)
* Lorentz Force Density in Frequency Domain (cc99aaac)
* Magnetic A-V,A formulation can handle moving material, including SUPG stabilization (2e821943, 21b4ea2d, aba69daf, 6d70738d, ccaa20dd)
* VolumeAcceleration in MechPDE (53f455c2)
* 2.5D (wave number) acoustic analysis type (66d3d5ee)
* Time Domain Equivalent Fluid model in acoustics (9661f7b4)
Breaking changes:
* Refactor Rayleigh damping (0abb75bd)
The input definition for _Rayleigh-type_ damping has changed to more clearly reflect the different damping types available in openCFS. We now have:
- 3 ways to define Rayleigh damping parameters in the material-XML file:
use `<coefficients>|<computeFromTanDeltaAtFrequency>|<computeFromTwoPoints>` in `<damping><rayleigh>`
- a new damping type `<adaptedLossTangensDelta>` in the simulation XML that represents Rayleigh damping with coefficients computed such that a specific loss factor is maintained for each frequency.
The material XML scheme has corresponding inputs that allow for constant and frequency-dependent loss factor definitions.
* non-conforming interface input changed (ff0d6d2f)
master->primary, slave->secondary
Bugfixes and improvements (selected):
* acoustic pressure output on surface elements (e2907e19)
* acoustic pressure as a nodal result for acouPotential formulation (d743e01f)
* averageRegionResults in AcouPDE (e421c2d6)
* additional postprocessing results for LinFlowPDE: surface velocity, impedance (4e439349)
* improved automatic layer generation (80d203fb)
* Correct Joule loss density in SpecialAV formulation (5a63d8d4)
* Corrected some postProcResults in the elecCurrentPDE and re-enabled transistor tests (83cec139, d39baca9)
* Improved externalSolver behaviour (cdd63365, 9a34e35f, d9deae7a)
* Orientation-dependent surface-volume integration point mapping (dcf77778)
* specialAV formulation with current forcing is now symmetric (4900e170)
* axi-symmetric LinFlowPDE works in the frequency domain now (8c2e3faa)
* solve the transpose and conjugate transpose problem with PARDISO (b45ae3a1)
* implement penalty method for multiple excitations (a5f82e8d)
* CoefFunctionGridElem to read element-based results as input (49ff0140)
* Fix various memory leaks, including GetRidOfZeros (be4bef44)
Developer notes:
* Updated CGNS to version 4.4.0 and switched CMake process to the new style (43e2cb58)
* Infrastructure improvements:
- smaller docker containers with shared MKL
- version labels
- shared cache for precompiled
- pipeline behaviour via mergre-request labels
* refactoring in AcousticPDE (26ee3c0a, 7deba41c)
* upgraded superLU (e840455f)
* use cdash.py to query the status of a test over time (4b5314e5)
* large refactoring in NCI handling (ff0d6d2f)
* new Intel-MKL finding based on cmake files provided in oneAPI (44756ced
* included the Testsuite repo (8f34e885)
* switch to c++17 (7b79f778)
* gidpost update (430e8083)
* zlib update (901e6253)
* cfsdeps now build in parallel (7db6735d,162abb21,3eab345c)