- Nektar-users - imperial.ac.uk

Question on Restarting APE Solver from Checkpoint Files
by dxz324＠student.bham.ac.uk 30 Jul '25

30 Jul '25

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Dear Nektar++ Development Team, I hope you are doing well. I am currently running the APE solver of Nektar++ on the ARCHER2 supercomputing platform. I have already enabled checkpointing and successfully generated .chk files during my simulation runs. However, I could not find detailed instructions in the user guide on how to restart the solver from these checkpoint files automatically, once the allocated job time expires. Could you please advise on the correct way to configure the solver (e.g., input files or command-line options) so that it restarts from a given checkpoint file rather than starting from the initial conditions? Thank you for your time and support. Best regards, Dao

2 1

CWIPI Version Mismatch and Forcing Newer CWIPI in Nektar++
by dxz324＠student.bham.ac.uk 29 Jul '25

29 Jul '25

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Dear Nektar++ Team, I have two questions about using CWIPI coupling in Nektar++ 5.2.0 on ARCHER2: Can a mismatch between the CWIPI library versions (the coupling application using CWIPI 1.3.0 vs. Nektar++ loading its bundled CWIPI 0.11.1) cause the MPI data exchange to deadlock at initialization? If so, what’s the recommended way to override Nektar++’s default CWIPI 0.11.1 at build/install time so it links against an external CWIPI 1.3.0 (or later) instead? For context, I’ve observed that on startup Nektar++ prints many: cwipi 0.11.1 initializing lines, while the external coupling application prints: cwipi 1.3.0 initializing and neither ever progresses to the “Coupling Config:” stage. I tried pointing -DCMAKE_PREFIX_PATH at my CWIPI 1.3.0 install, but ldd AcousticSolver still shows libcwp.so.0.11.1. I also attempted manually replacing libcwp.so.0.11.1 with a symlink to libcwipi.so.1, without success. Could you please confirm: Whether version mismatch alone can block the CWIPI handshake? The exact CMake or install steps needed to force Nektar++ to use an external CWIPI 1.3.0 instead of its internal copy? Thank you for your guidance. Best wishes, Dao

2 1

Merging two or more solvers: Electrohydrodynamics
by Ananth Narayan S am24d038 23 Jul '25

23 Jul '25

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Dear Nektar users, I am new to Nektar and planning to incorporate the IncNavierStokesSolver with ADRSolver- Poisson. I am currently working on Electrohydrodynamics, where Navier Stokes, Poisson equation (for potential difference) and species transport (for positive and negative species with custom forcing term). [image: image.png] As a first step, I am trying to solve Navier Stokes with Poisson equation (not coupled), further solve the species equation then couple the force (\rho_e\vec{E}) in the momentum equation. I have already written this solver in OpenFOAM, so I can grasp the concepts at earliest. Since the coding style is completely different, I need support on implementing this solver. I have seen the documentation on implementing the scalar transport along with IncNavierStokes, but found no examples or other documentation on implementation. Regarding the coding part, I am not sure where to add Poisson solver as soon as Navier stokes is solved for a time step. Thank you Ananth Narayan PhD, Fluid Mechanics IIT Madras

1 0

2D stability analysis question
by Ben 15 Jul '25

15 Jul '25

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Hello everyone, I have encountered some problems while calculating 2D linear stability analysis. Anyone can provide suggestions. 1. The extension does not take effect in direct and adjoint, but only in the calculation of the baseflow. The meaning is that the dat grids of direct and adjoint are not encrypted according to the settings in the extension.Set as follows <EXPANSIONS>         <E COMPOSITE="C[37]" NUMMODES="3, 3" BASISTYPE="Modified_A,Modified_A" POINTSTYPE="GaussLobattoLegendre,GaussLobattoLegendre"   NUMPOINTS="4,4" FIELDS="u,v" />         <E COMPOSITE="C[37]" NUMMODES="2, 2" BASISTYPE="Modified_A,Modified_A" POINTSTYPE="GaussLobattoLegendre,GaussLobattoLegendre"   NUMPOINTS="4,4" FIELDS="p" />     </EXPANSIONS> 2. The vorticity and modal plots in tecplot are very blurry. Do I need to perform additional post-processing on the results (such as C0Projection) or increase the grid order for recalculation? Ben

1 0

Basic questions related to Nektar++ Steady State Solver
by Gaurav Gupta 10 Jul '25

10 Jul '25

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Hello, I am a new user of Nektar++ and will be using the Linear Stability Analysis solver. I have been able to run simulations successfully and I have a few basic questions related to post-processing in Nektar++. I would be grateful if someone could help me with the following questions: 1. What are the headers of the ConvergenceHistory.txt file ? There are 5 columns where I understand the first three are iteration number, simulation time, computation time. Are the other two Linf and L2 ? 2. I had run the solver in parallel which generates folders with n chk/fld files for n processes. Is there any way I can combine these n files into a single vtu file for post-processing ? I couldn't find answers to the above questions in user-guide as well as in the email archive. Hence, I am posting this email. Thanks! Sincerely, *Gaurav Gupta* Senior Undergrad Student Dept. of Aerospace Engineering Indian Institute of Space Science and Technology

2 1

Questions on implicit integration and file coupling in AcousticSolver (v5.2.0)
by dxz324＠student.bham.ac.uk 29 Jun '25

29 Jun '25

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Dear Nektar++ developers, I am a researcher at the University of Birmingham using AcousticSolver v5.2.0 on ARCHER2 to solve the acoustic perturbation equations. My work involves sparse source data for a mixing layer, and with only explicit schemes (for example RK4), I must use very small "TimeStep" to maintain stability(The smallest element size of the grid cannot be modified, as it is already the largest option for identify source). I have two questions: 1. Implicit integrator support Currently AcousticSolver does not include any implicit time�\integration methods. Could you provide a test or development version that includes an implicit scheme such as backward Euler or Crank�\Nicolson? If no such version exists, could you advise how I might modify the solver code to add an implicit integrator? 2. Errors when inputting files As shown in my code(figure below), I input the source data using the "COUPLING" element with TYPE="File" and provide a series of .pts files named in floating point format (%14.8E). I also set TIMEDEPENDENT="1" so that they are read in time order. Since the time interval between these files is much larger than my solver TimeStep, when I tried to adjust ReceiveSteps to read a new file every n steps, the solver still tried to load the file at each dt(TimeStep) and failed when the file at the interpolation time did not exist. Is there a way to instruct AcousticSolver to read coupling files strictly according to the ReceiveSteps schedule��i.e. to load only when a new file index is due��without triggering ��file not found�� errors at intermediate time steps? Any guidance or pointers to relevant code sections or example configurations would be greatly appreciated. Thank you very much for your time and assistance. Best regards, Dao University of Birmingham [cid:511f73a2-b6b1-4275-bc1a-e1a0ac686633]

2 1

Promote your Nektar++-related publications on the website!
by Chris Cantwell 28 Jun '25

28 Jun '25

Dear all, As announced by Dave at the recent Nektar++ Workshop, we have a publications page on the website which strives to list the many publications which have contributed to, or made use of, Nektar++ from across the community: https://www.nektar.info/community/publications/ *If you have such a paper, we encourage you to add it to this page!* As well as giving a little extra promotion for your publication, it also helps us to demonstrate to potential funders the impact of the software, which can support further development. We have recently sought to make it easier for people to contribute their publications, by generating the webpage dynamically from BibTeX files stored in a GitLab repository. You can therefore contribute to the list as follows: 1. Fork the https://gitlab.nektar.info/nektar/nektar-publications repository into your own namespace on GitLab. 2. Clone the repository and create a new branch. 3. Update the BibTeX files accordingly to add your publication(s), tagging them with a 'group' attribute as 'core','app' or 'thesis'. 4. Commit and push your changes to GitLab. 5. Create a merge request from the branch in your fork, to 'master' in the main nektar-publications repository. Some automated checks will be performed to ensure the BibTeX is valid. Once these pass, we will then merge the MR and it will appear on the webpage. For further details on the process, please see: https://gitlab.nektar.info/nektar/nektar-publications/-/blob/master/README.… Finally, thanks to Dave for getting all of the GitLab CI and webpage-generation magic to work! Kind regards, Chris -- Chris Cantwell Reader in Computational Engineering Director of Postgraduate Studies (PhD) Department of Aeronautics Imperial College London South Kensington Campus London SW7 2AZ Tel: +44 (0)20 759 45050 Email:c.cantwell@imperial.ac.uk www.imperial.ac.uk/people/c.cantwell

1 0

2D linear stability analysis
by Ben 26 Jun '25

26 Jun '25

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Dear all, I am studying 2D linear stability analysis. I have browsed through the emails of the nektar community and learned a lot about solving settings, but there are still some questions that I cannot understand. Everyone can offer their opinions. 1. Is the combination of fine mesh and low order unfolding more effective? Is it better for the order of pressure 'p' to be 1 order lower than the velocity 'u, v'? 2. When using Arpack for serial computing, will setting kdim to a larger value result in faster solving speed, such as 64 and 128? Because I encountered incorrect eigenvalues in my calculations, such as an unstable flow with a growth rate less than 0. Do I need to set the nvc to a large value? 3. The post mentioned that using smooth initial conditions can avoid calculating eigenvalues as 1. How should I set it up? Can you provide a specific example? 4. How should base flow be calculated? Should I average the extraction time of DNS results or use the SFD method? How to set the parameters in the adaptive SFD method to ensure both solution convergence and fast computation speed? The direct.xml and SFD.xml files are attached.Thank you for your suggestions. Qiancheng

1 0

Filter_Integral_for 3DH1D_segmentation fault
by Wang Boning 24 Jun '25

24 Jun '25

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Dear Nektar++ team, I hope this email finds you well! I'm a student from TUM in Munich, Germany. And I am currently running a 3DH1D simulation using the IncSolver module in Nektar++ version 5.8.0. I encountered an issue when trying to use the Integral filter. The simulation terminates with an error message (included at the end of this email). Moreover, when I comment out the <FILTER TYPE="Integral"> block, the simulation runs normally. I found an early issue, 4yr ago, on GitLab (#227), which reveals that the Integral filter should be supported in homogenous simulations. I also tried using same modals for all composites/variables, still same error message. My question is, dose ' integral' support 3DH1D since composites are 2D(Quadrilateral), which is not mentioned specifically in the doc? And if yes, what can I do to fix my issue for 3DH1D simulation? I really tested a lot before I send this email. Your sincerely, Boning P.S. Partial set-up are listed here, and full session.xml, mesh.xml and log (verbose) file are attached. <EXPANSIONS> <E COMPOSITE="C[1,2,3,5]" NUMMODES="6" TYPE="MODIFIED" FIELDS="u,v,w,p" /> <E COMPOSITE="C[4,6]" NUMMODES="8" TYPE="MODIFIED" FIELDS="u,v,w" /> <E COMPOSITE="C[4,6]" NUMMODES="7" TYPE="MODIFIEDQUADPLUS1" FIELDS="p" /> </EXPANSIONS> <FILTER TYPE="Integral"> <PARAM NAME="OutputFile">Integral</PARAM> <PARAM NAME="OutputFrequency">5</PARAM> <PARAM NAME="OutputPrecision"> 12 </PARAM> <PARAM NAME="Composites">C[4,5], C[4,6], C[1-4]</PARAM> </FILTER> [hyd39:90887] *** Process received signal *** [hyd39:90887] Signal: Segmentation fault (11) [hyd39:90887] Signal code: Address not mapped (1) [hyd39:90887] Failing at address: 0xb0 [hyd39:90887] [ 0] /lib64/libc.so.6(+0x57900)[0x7fb616e57900] [hyd39:90887] [ 1] /opt/nektar++/5.8.0/lib64/libSolverUtils.so.5.8.0(_ZN6Nektar11SolverUtils14FilterIntegral12v_InitialiseERKNS_5ArrayINS_4OneDEKSt10shared_ptrINS_12MultiRegions7ExpListEEEERKd+0x23b)[0x7fb61ce9f2cb] [hyd39:90887] [ 2] /opt/nektar++/5.8.0/lib64/libSolverUtils.so.5.8.0(_ZN6Nektar11SolverUtils14UnsteadySystem9v_DoSolveEv+0x360)[0x7fb61cee3240] [hyd39:90887] [ 3] /opt/nektar++/5.8.0/lib64/libSolverUtils.so.5.8.0(_ZN6Nektar11SolverUtils14DriverStandard9v_ExecuteERSo+0x30a)[0x7fb61ce2ceba] [hyd39:90887] [ 4] IncNavierStokesSolver[0x44928a] [hyd39:90887] [ 5] /lib64/libc.so.6(+0x40e6c)[0x7fb616e40e6c] [hyd39:90887] [ 6] /lib64/libc.so.6(__libc_start_main+0x87)[0x7fb616e40f35] [hyd39:90887] [ 7] IncNavierStokesSolver[0x4782f1] [hyd39:90887] *** End of error message ***

1 0

JFM Paper on Instability Suppression Published
by Abhishek Kumar 23 Jun '25

23 Jun '25

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Dear all, I hope you're all doing well. I’m pleased to share that our paper, *“Suppressing instabilities in mixed baroclinic flow using an actuation based on receptivity,”* has now been published in the *Journal of Fluid Mechanics*: https://doi.org/10.1017/jfm.2025.10241 The study presents a receptivity-based strategy to suppress instabilities in a mixed convective flow, with actuation designed from adjoint eigenmodes. The work builds on our previous stability analysis and was carried out entirely within the Nektar++ framework. Some of the earlier results were presented at a previous Nektar++ meeting. I’ve attached the published version in case it’s of interest. Unfortunately, I won’t be able to attend this year’s Nektar++ meeting due to personal circumstances. I wish you all a successful and enjoyable meeting! Best regards, Abhishek

1 0