TDDFT with many q points run errors

[hongtang2023]

Dear Yambo team,

I am using Yambo.5.1.1.
I met a strange situation that I need your help.

I am doing with TDDFT ALDA kernel. I want to calculate the first 25 q points, i.e.
"
% QpntsRXd
1 | 25 | # [Xd] Transferred momenta
%
"

I did two testes.

Test 1. when do the nscf in QE, I use "
K_POINTS automatic
14 14 1 0 0 0
"
then, do the p2y and yambo initialization, then do the TDDFT calculation with " % QpntsRXd
1 | 25 | # [Xd] Transferred momenta
% ", then Yambo runs smoothly for the first 13 q points, then stop at the 14th q point. The reported errors "
srun: lua: Submitted job 383151
Assertion failed in file ../../src/mpid/ch4/src/intel/ch4_shm_coll.c at line 2266: comm->shm_numa_layout[my_numa_node].base_addr
Assertion failed in file ../../src/mpid/ch4/src/intel/ch4_shm_coll.c at line 2266: comm->shm_numa_layout[my_numa_node].base_addr
Assertion failed in file ../../src/mpid/ch4/src/intel/ch4_shm_coll.c at line 2266: comm->shm_numa_layout[my_numa_node].base_addr
Assertion failed in file ../../src/mpid/ch4/src/intel/ch4_shm_coll.c at line 2266: comm->shm_numa_layout[my_numa_node].base_addr
Assertion failed in file ../../src/mpid/ch4/src/intel/ch4_shm_coll.c at line 2266: comm->shm_numa_layout[my_numa_node].base_addr
Assertion failed in file ../../src/mpid/ch4/src/intel/ch4_shm_coll.c at line 2266: comm->shm_numa_layout[my_numa_node].base_addr
Assertion failed in file ../../src/mpid/ch4/src/intel/ch4_shm_coll.c at line 2266: comm->shm_numa_layout[my_numa_node].base_addr
Assertion failed in file ../../src/mpid/ch4/src/intel/ch4_shm_coll.c at line 2266: comm->shm_numa_layout[my_numa_node].base_addr
/home/packages/compilers/intel/mpi/2021.5.1//lib/release/libmpi.so.12(MPL_backtrace_show+0x1c) [0x14986e035bcc]
/home/packages/compilers/intel/mpi/2021.5.1//lib/release/libmpi.so.12(MPIR_Assert_fail+0x21) [0x14986da0fdf1]
/home/packages/compilers/intel/mpi/2021.5.1//lib/release/libmpi.so.12(+0x2b1eb9) [0x14986d6deeb9]
/home/packages/compilers/intel/mpi/2021.5.1//lib/release/libmpi.so.12(+0x1c104a) [0x14986d5ee04a]
/home/packages/compilers/intel/mpi/2021.5.1//lib/release/libmpi.so.12(+0x191ca5) [0x14986d5beca5]
/home/packages/compilers/intel/mpi/2021.5.1//lib/release/libmpi.so.12(+0x1717ec) [0x14986d59e7ec]
/home/packages/compilers/intel/mpi/2021.5.1//lib/release/libmpi.so.12(+0x2b4605) [0x14986d6e1605]
/home/packages/compilers/intel/mpi/2021.5.1//lib/release/libmpi.so.12(MPI_Barrier+0x254) [0x14986d578664]
/home/packages/compilers/intel/mpi/2021.5.1//lib/libmpifort.so.12(PMPI_BARRIER+0xc) [0x14986ed4027c]
/work/tang2017/Yambo511_Install/Installed_MCP07u_Yambo511/YAMBO_MCP07u_q/yambo-5.1.1_try_compile/yambo-5.1.1_mcp07_any_SOC/bin/yambo() [0x6fdac2]
/work/tang2017/Yambo511_Install/Installed_MCP07u_Yambo511/YAMBO_MCP07u_q/yambo-5.1.1_try_compile/yambo-5.1.1_mcp07_any_SOC/bin/yambo() [0x6b10b2]
/work/tang2017/Yambo511_Install/Installed_MCP07u_Yambo511/YAMBO_MCP07u_q/yambo-5.1.1_try_compile/yambo-5.1.1_mcp07_any_SOC/bin/yambo() [0x6accf8]
/work/tang2017/Yambo511_Install/Installed_MCP07u_Yambo511/YAMBO_MCP07u_q/yambo-5.1.1_try_compile/yambo-5.1.1_mcp07_any_SOC/bin/yambo() [0x4c3629]
/work/tang2017/Yambo511_Install/Installed_MCP07u_Yambo511/YAMBO_MCP07u_q/yambo-5.1.1_try_compile/yambo-5.1.1_mcp07_any_SOC/bin/yambo() [0x408ce8]
/work/tang2017/Yambo511_Install/Installed_MCP07u_Yambo511/YAMBO_MCP07u_q/yambo-5.1.1_try_compile/yambo-5.1.1_mcp07_any_SOC/bin/yambo() [0x77d2c5]
...
...
...

"




Test 2. when doing nscf in QE, I use"
K_POINTS crystal
196
0.00000000 0.00000000 0.00000000 5.102041e-03
0.00000000 0.07142857 0.00000000 5.102041e-03
0.00000000 0.14285714 0.00000000 5.102041e-03
0.00000000 0.21428571 0.00000000 5.102041e-03
0.00000000 0.28571429 0.00000000 5.102041e-03
0.00000000 0.35714286 0.00000000 5.102041e-03
...
...
...
" , which means that I use the explicitly listed k points for a 14x14x1 mesh, which has totally 196 k points.
then I do p2y and yambo initialization. then do TDDFT calculation with "" % QpntsRXd
1 | 25 | # [Xd] Transferred momenta
% ", the yambo input for this run is the same as in Test 1. This time yambo runs only for the first q. then stops with no any errors. when I check the output file " o-alda_q100.eel_q1_inv_rpa_dyson ", it shows that "# | % QpntsRXd
# | 1 | 1 | # [Xd] Transferred momenta
# | % ", it is weird that yambo changed it from " 1 | 25 | " to " 1 | 1 | ". Why is that?


in the two tests, the only difference is that I do the nscf calculation in QE with different formats for the k grids.
But this results in completely different running outcome!


Please help me for this!

Appreciate your help very much!

Best,
Hong

____
my yambo input:

optics # [R] Linear Response optical properties
chi # [R][CHI] Dyson equation for Chi.
tddft # [R][K] Use TDDFT kernel
# ElecTemp= check r_setup file eV # Electronic Temperature
FFTGvecs= 2 Ry # [FFT] Plane-waves
X_Threads=2 # [OPENMP/X] Number of threads for response functions
DIP_Threads=2 # [OPENMP/X] Number of threads for dipoles
Chimod= "ALDA" # [X] IP/Hartree/ALDA/LRC/PF/BSfxc
# DipApproach= "Covariant" # [DIP] [G-space v/R-space x/Covariant/Shifted grids]
FxcGRLc= 1 Ry # [TDDFT] XC-kernel RL size
NGsBlkXd= 1 Ry # [Xd] Response block size
% QpntsRXd
1 | 25 | # [Xd] Transferred momenta
%
% BndsRnXd
1 | 70 | # [Xd] Polarization function bands
%
% EnRngeXd
0.00000 | 0.100000 | eV # [Xd] Energy range
%
% DmRngeXd
0.100000 | 0.100000 | eV # [Xd] Damping range
%
ETStpsXd= 3 # make a spce before = # [Xd] Total Energy steps, make a spce before =
% LongDrXd
1.000000 | 0.000000 | 0.000000 | # [Xd] [cc] Electric Field
%

[Daniele Varsano]

Dear Hong Tang,

in test 2 you did, most probably your k grid is not recognized as uniform, so calculations are performed only for q=0 and that's why the number of q point is changed.
You can check this in the r_setup file, where this is indicated (search X grid is uniform).

For the first test, the problem should be investigated, but before that it would be useful if you could update your Yambo version.
5.1.1. is rather old, please update to the 5.3. If the problem persists, please post here input/report and log files.

Best,

Daniele