Issue with Negative Frequencies in Phonon Dispersion using TDEP

[Harish-IITB]

I am currently using the TDEP package to extract temperature-dependent second-order force constants and compute the phonon dispersion curves for Gd₂Zr₂O₇.

However, I am encountering negative frequencies in the dispersion curve. I would like to briefly outline the steps I am following:

Step 1: Generate canonical structures using initial (fake) force constants — 10 structures in total.

Step 2: Use the second-order force constants obtained from Step 1 to generate another set of 10 canonical     

structures. I then calculate the updated force constants and the phonon dispersion curve.

I am also attaching input files below

step_2_files.zip

[mjv500]

Hi, You will probably need to iterate a few more times. Rather than use too many configurations from the fake force constants, you can increase the number of canonical configurations generated at each step. Once the phonons converge as a function of number of configurations, you should have a stabilized band structure. The more complex and anharmonic your material, the longer you may need to iterate. If you have harmonic frequencies from abinit or qe they would be a better starting point. M. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Professor Matthieu J Verstraete Fellow, American Physical Society <https://www.aps.org> Chair, Steering Committee, European Theoretical Spectroscopy Facility <https://www.etsf.eu> Alumnus Fellow, Young Academy of Europe <https://www.yacadeuro.org> <http://yacadeuro.org/> Institute for Theoretical Physics (ITP) Department of Physics Buys Ballot Gebouw/Building, Princetonplein 5, office University of Utrecht, 3584 CC Utrecht Phone: +31 30 253 5926 Secretariat: +31 30 253 5929 and ***@***.*** Group web page (Liège): http://www.nanomat.ulg.ac.be/ Nanomat lab, Q-Mat center, Université de Liège Département de Physique, Bat. B5a, 4/50 Allée du 6 août, 19 B-4000 Sart Tilman, Liège Belgium Phone : +32 4 366 90 17 European Theoretical Spectroscopy Facility <https://www.etsf.eu> (ETSF) Mail : ***@***.*** ***@***.*** ***@***.***

…

On Sat, Jul 26, 2025, 12:50 Harish-IITB ***@***.***> wrote: *Harish-IITB* created an issue (tdep-developers/tdep#154) <#154> I am currently using the TDEP package to extract temperature-dependent second-order force constants and compute the phonon dispersion curves for Gd₂Zr₂O₇. However, I am encountering negative frequencies in the dispersion curve. I would like to briefly outline the steps I am following: Step 1: Generate canonical structures using initial (fake) force constants — 10 structures in total. Step 2: Use the second-order force constants obtained from Step 1 to generate another set of 10 canonical structures. I then calculate the updated force constants and the phonon dispersion curve. I am also attaching input files below step_2_files.zip <https://github.com/user-attachments/files/21443056/step_2_files.zip> — Reply to this email directly, view it on GitHub <#154>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABWOT32MHRKK4LYHRZQ5YWD3KNMQPAVCNFSM6AAAAACCNOBRJCVHI2DSMVQWIX3LMV43ASLTON2WKOZTGI3DKNBTGA2TIMQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[PhysXplorer]

Hi Matthieu,

This is a happy(?) coincidence. I face a similar error, quite surprisingly, while attempting to simulate quartz (SiO2). Despite multiple iterations with increasing configurations, a few negative frequencies do not seem to vanish.

I've confirmed my structure against literature (which usually run DFPT, so T=0K). The structure is also known to be stable in my temperature of interest – 300 K. Therefore, I would not expect it to have any instabilities.

I do not seem to have missed any computational step/instruction that would cause this. Your inputs and recommendations are most welcome.

Thanks in anticipation,
Sreerag.

PS: PFA a few files that I believe might be helpful.

Error-quartz.zip

[mjv500]

Dear Happy Xplorer, you can also try --relax for extractforceconstants, which refines the positions to zero out the 1st order part of the fit force constants. Spits out a new ucposcar. You have an elastic instability which is massive so your unit cell can not already be relaxed. Could be a xc functional issue, you might also try PBESOL Note that SiO2 can be very sensitive to the exact structure. This paper: https://journals.aps.org/prb/abstract/10.1103/PhysRevB.51.8610 already had stable DFPT phonons, so it must be possible. M.

…

On Mon, Jul 28, 2025 at 6:51 AM PhysXplorer ***@***.***> wrote: *PhysXplorer* left a comment (tdep-developers/tdep#154) <#154 (comment)> Hi Matthieu, This is a happy(?) coincidence. I face a similar error, quite surprisingly, while attempting to simulate quartz (SiO2). Despite multiple iterations with increasing configurations, a few negative frequencies do not seem to vanish. I've confirmed my structure against literature (which usually run DFPT, so T=0K). The structure is also known to be stable in my temperature of interest – 300 K. Therefore, I would not expect it to have any instabilities. I do not seem to have missed any computational step/instruction that would cause this. Your inputs and recommendations are most welcome. Thanks in anticipation, Sreerag. PS: PFA a few files that I believe might be helpful. Error-quartz.zip <https://github.com/user-attachments/files/21460924/Error-quartz.zip> — Reply to this email directly, view it on GitHub <#154 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABWOT33MBTN3HCHUEFDRZUL3KWT4JAVCNFSM6AAAAACCNOBRJCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTCMRVGQ3DENJRGE> . You are receiving this because you commented.Message ID: ***@***.***>

-- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Professor Matthieu J Verstraete Fellow, American Physical Society Chair, Steering Committee, European Theoretical Spectroscopy Facility www.etsf.eu Alumnus Fellow, Young Academy of Europe yacadeuro.org Institute for Theoretical Physics (ITP) Department of Physics Buys Ballot Gebouw/Building, Princetonplein 5, office University of Utrecht, 3584 CC Utrecht ITP Secretariat: +31 30 253 5928 E-mail: ***@***.*** Group web page (Liège): http://www.nanomat.ulg.ac.be/ Nanomat lab, Q-Mat center, Université de Liège Département de Physique, Bat. B5a, 4/50 Allée du 6 août, 19 B-4000 Sart Tilman, Liège Belgium Phone : +32 4 366 90 17 European Theoretical Spectroscopy Facility (ETSF) Mail : ***@***.*** ***@***.*** ***@***.***

[PhysXplorer]

Oh that's an interesting tag, I'll try that out. Thank you for sharing that and your thoughts.

[PhysXplorer]

Dear Prof. Matthieu,

Back after attempting a few variations along with suggested --relax tag and would like to share my observations/queries. I request your insights on the same.

1. Changing the unitcell configuration or the xc-functional to PBEsol didn't seem to help either. The negative frequencies do not seem to resolve out of the system. This is very weird for me, especially with a such well-studied system. I'm stumped with not being able to find an explanation for this.

2. I used the --relax tag with extract_forceconstants, I appreciate your recommendation. The new_ucposcar thrown out is used in generation of configurations for the next iteration and subsequent calculations.

I'm attaching a few files with results from both attempts. I hope this helps.

Thanks,
S.

Error-update.zip

[mjv500]

Hello, I don't have much to add. I agree it should be easy. Perhaps revert to a phonopy calculation of the phonon bands, to check. I note that your Born charges are not diagonal (low symmetry). In some cases this has created problems with imposing the Hermiticity of the dynamical matrices, though in the latest tdep this should be fixed. You might artificially remove the off diagonal components. Basically, many other calculations in finite diff, DFPT etc have produced valid phonons for SiO2, so if things are relaxed etc everything should run. And your instabilities are elastic at q=0, so something kind of central is wrong, though I can not see what off the face of it. Perhaps the rc2 is too low, and there are longer range interactions? M.

…

On Sat, Aug 23, 2025 at 9:10 AM PhysXplorer ***@***.***> wrote: *PhysXplorer* left a comment (tdep-developers/tdep#154) <#154 (comment)> Dear Prof. Matthieu, Back after attempting a few variations along with suggested --relax tag and would like to share my observations/queries. I request your insights on the same. 1. Changing the unitcell configuration or the xc-functional to PBEsol didn't seem to help either. The negative frequencies do not seem to resolve out of the system. This is very weird for me, especially with a such well-studied system. I'm stumped with not being able to find an explanation for this. 2. I used the --relax tag with extract_forceconstants, I appreciate your recommendation. The new_ucposcar thrown out is used in generation of configurations for the next iteration and subsequent calculations. I'm attaching a few files with results from both attempts. I hope this helps. Thanks, S. Error-update.zip <https://github.com/user-attachments/files/21947798/Error-update.zip> — Reply to this email directly, view it on GitHub <#154 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABWOT346WN3DVAA6AQ63BCD3PAHWPAVCNFSM6AAAAACCNOBRJCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTEMJWGUYDMMRQGI> . You are receiving this because you commented.Message ID: ***@***.***>

-- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Professor Matthieu J Verstraete Fellow, American Physical Society Chair, Steering Committee, European Theoretical Spectroscopy Facility www.etsf.eu Alumnus Fellow, Young Academy of Europe yacadeuro.org Institute for Theoretical Physics (ITP) Department of Physics Buys Ballot Gebouw/Building, Princetonplein 5, office University of Utrecht, 3584 CC Utrecht ITP Secretariat: +31 30 253 5928 E-mail: ***@***.*** Group web page (Liège): http://www.nanomat.ulg.ac.be/ Nanomat lab, Q-Mat center, Université de Liège Département de Physique, Bat. B5a, 4/50 Allée du 6 août, 19 B-4000 Sart Tilman, Liège Belgium Phone : +32 4 366 90 17 European Theoretical Spectroscopy Facility (ETSF) Mail : ***@***.*** ***@***.*** ***@***.***