Crash for TDDFT with RKS_TRIPLETS and ADMM

[T deJ]

Dear CP2K developers,

Recently we have run into an bug where the TDDFPT subroutine crashes for RKS_TRIPLETS: to calculate triplet excited states from a closed shell singlet ground state. This occurs when you run hybrid functionals with ADMM.

The error occurs in the application of the second functional derivative (see attached error file). It looks like the code thinks that the system is unrestricted (LSD), as the alpha and beta rho are allocated.

It's likely a relatively recent bug, as we have done these type of calculations before.

Best regards,

Tjeerd de Jong

[Jürg Hutter]

Hi

please try

EXCH_CORRECTION_FUNC BECKE88X_LIBXC

Analytic 2nd derivatives are not implemented for (most) CP2K internal
functionals. Using LIBXC helps in those cases.

regards
JH

________________________________________
From: cp...@googlegroups.com <cp...@googlegroups.com> on behalf of T deJ <t.f....@gmail.com>
Sent: Thursday, November 6, 2025 1:01 PM
To: cp...@googlegroups.com
Subject: [CP2K:21951] Crash for TDDFT with RKS_TRIPLETS and ADMM

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com<mailto:cp2k+uns...@googlegroups.com>.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/CAAse_hiJRGXXpmUM4pba5-sK8C%3D7o-BcHGMfngFjwAQ74CoTEA%40mail.gmail.com<https://groups.google.com/d/msgid/cp2k/CAAse_hiJRGXXpmUM4pba5-sK8C%3D7o-BcHGMfngFjwAQ74CoTEA%40mail.gmail.com?utm_medium=email&utm_source=footer>.