AICCM consistency matrix (experimental)¶
Generated by scripts/aiccm_consistency_matrix.py on the commit, native core
and date listed below. It covers tiny fixtures on a single machine only. It is
not a multinode qualification and not a scientific-accuracy qualification of
either AICCM line; both lines remain experimental. A REFUSED or NOT-RUN cell
is not agreement. Regenerate the page rather than editing it.
Commit:
fc210236e1b6b62e670486aab8da14ad2d9241ebNative core sha256:
1b77c75dba3284beef8dd35c8d539ce68815ba04a686e8600214181be982456bvibeqc.__version__:0.18.13Generated (UTC): 2026-11-12T20:14:10Z
Scope: tiny fixtures on a single machine; not a multinode test and not a scientific-accuracy qualification; AICCM is experimental.
Core budget: ranks x OpenMP threads <= 4; every cell is its own process; MPI cells run under
mpirun -n N, Gamma rows withVIBEQC_CCM_FOLD_MPI=1.
Tolerances¶
Comparison |
Tolerance |
|---|---|
SCF energy, separately converged (SCF to 1e-12 Ha / gradient 1e-09) |
1e-09 Ha/cell |
Correlation energy, separately converged references |
1e-08 Ha/cell |
Energy on the same converged reference reused |
1e-12 Ha/cell |
Analytic gradient, max abs component |
1e-07 Ha/bohr |
Reading the cells¶
Only PASS states agreement, and only within the stated tolerance. REFUSED is a by-design refusal of that configuration (the first line of its message is shown); it says nothing about agreement. NOT-RUN means the cell was not executed. DIFFERENT-OPERATOR compares two fitted operators that differ by construction (Libint against MANTA sources) and makes no pass/fail claim. FAIL is a same-operator deviation above tolerance or an exception that is not a recognised by-design refusal (NotImplementedError, an MPI CollectiveRefusal, or a ValueError message named in advance for that cell). dE is the energy (SCF energy per cell; for correlated rows the reference energy), dEc the correlation energy per cell, dG the largest gradient component difference. The rank spread is the largest difference between ranks of one MPI cell.
Cell counts: BASELINE 13, DIFFERENT-OPERATOR 3, NOT-RUN 12, PASS 55, REFUSED 21
Real-Gamma (variant="real-gamma")¶
RHF energy¶
Fixture: H2 in a 6 bohr cube (z 2.3/3.7), STO-3G, (2,1,1) torus, front door, HCORE guess, SCF 1e-12/1e-9.
Configuration |
Outcome |
Executed (rank 0 record) |
Ranks x threads |
Wall (s) |
|---|---|---|---|---|
baseline |
BASELINE E=-1.1945309126 |
backend libint; storage dense; fold serial x1 |
1 x 1 |
30.1 |
MANTA sources |
DIFFERENT-OPERATOR dE=2.5e-07 (no pass/fail claim) |
backend manta; storage dense; fold serial x1 |
1 x 1 |
67.1 |
block storage |
PASS dE=8.9e-16 (tol 1.0e-09) |
backend libint; storage block; fold serial x1 |
1 x 1 |
21.4 |
block-star storage |
PASS dE=8.9e-16 (tol 1.0e-09) |
backend libint; storage block-star; fold serial x1 |
1 x 1 |
20.8 |
2 OpenMP threads |
PASS dE=0 (tol 1.0e-09) |
backend libint; storage dense; fold serial x1 |
1 x 2 |
19.8 |
2 MPI ranks |
PASS dE=0 (tol 1.0e-09); rank spread 0 |
backend libint; storage dense; fold mpi-owned-q-channels x2 |
2 x 1 |
27.9 |
3 MPI ranks |
PASS dE=0 (tol 1.0e-09); rank spread 0 |
backend libint; storage dense; fold mpi-owned-q-channels x3 |
3 x 1 |
30.7 |
control_start=direct |
NOT-RUN: control_start applies to the Gamma SCF gradient only |
- |
1 x 1 |
0.0 |
PBE energy¶
Fixture: H2 in a 6 bohr cube (z 2.3/3.7), STO-3G, (2,1,1) torus, front door, HCORE guess, SCF 1e-12/1e-9.
Configuration |
Outcome |
Executed (rank 0 record) |
Ranks x threads |
Wall (s) |
|---|---|---|---|---|
baseline |
BASELINE E=-1.1995479034 |
backend libint; storage dense; fold serial x1 |
1 x 1 |
67.5 |
MANTA sources |
DIFFERENT-OPERATOR dE=1.4e-05 (no pass/fail claim) |
backend manta; storage dense; fold serial x1 |
1 x 1 |
106.1 |
block storage |
PASS dE=4.4e-16 (tol 1.0e-09) |
backend libint; storage block; fold serial x1 |
1 x 1 |
58.7 |
block-star storage |
PASS dE=4.4e-16 (tol 1.0e-09) |
backend libint; storage block-star; fold serial x1 |
1 x 1 |
48.3 |
2 OpenMP threads |
PASS dE=0 (tol 1.0e-09) |
backend libint; storage dense; fold serial x1 |
1 x 2 |
12.3 |
2 MPI ranks |
PASS dE=0 (tol 1.0e-09); rank spread 0 |
backend libint; storage dense; fold mpi-owned-q-channels x2 |
2 x 1 |
20.1 |
3 MPI ranks |
PASS dE=0 (tol 1.0e-09); rank spread 0 |
backend libint; storage dense; fold mpi-owned-q-channels x3 |
3 x 1 |
17.8 |
control_start=direct |
NOT-RUN: control_start applies to the Gamma SCF gradient only |
- |
1 x 1 |
0.0 |
RI-MP2 energy¶
Fixture: H2 in a 6 bohr cube (z 2.3/3.7), STO-3G, (2,1,1) torus, front door, HCORE guess, SCF 1e-12/1e-9.
Configuration |
Outcome |
Executed (rank 0 record) |
Ranks x threads |
Wall (s) |
|---|---|---|---|---|
baseline |
BASELINE E=-1.1945309126, Ec=-0.0097087185 |
backend libint; storage dense; fold serial x1 |
1 x 1 |
8.2 |
MANTA sources |
DIFFERENT-OPERATOR dE=2.5e-07, dEc=2.5e-09 (no pass/fail claim) |
backend manta; storage dense; fold serial x1 |
1 x 1 |
46.3 |
block storage |
PASS dE=8.9e-16 (tol 1.0e-09), dEc=1.2e-17 (tol 1.0e-08) |
backend libint; storage block; fold serial x1; MP2 reduction translation |
1 x 1 |
30.1 |
block-star storage |
PASS dE=8.9e-16 (tol 1.0e-09), dEc=1.2e-17 (tol 1.0e-08) |
backend libint; storage block-star; fold serial x1; MP2 reduction star |
1 x 1 |
36.9 |
2 OpenMP threads |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08) |
backend libint; storage dense; fold serial x1 |
1 x 2 |
37.0 |
2 MPI ranks |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08); rank spread 0 |
backend libint; storage dense; fold mpi-owned-q-channels x2 |
2 x 1 |
43.6 |
3 MPI ranks |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08); rank spread 0 |
backend libint; storage dense; fold mpi-owned-q-channels x3 |
3 x 1 |
40.4 |
control_start=direct |
NOT-RUN: control_start applies to the Gamma SCF gradient only |
- |
1 x 1 |
0.0 |
DLPNO-MP2 front door, default settings (canonical RI-MP2, no truncation)¶
Fixture: H2 in a 6 bohr cube (z 2.3/3.7), STO-3G, (2,1,1) torus, front door, HCORE guess, SCF 1e-12/1e-9.
Configuration |
Outcome |
Executed (rank 0 record) |
Ranks x threads |
Wall (s) |
|---|---|---|---|---|
baseline |
BASELINE E=-1.1945309126, Ec=-0.0097087185 |
backend libint; storage dense; fold serial x1 |
1 x 1 |
43.1 |
MANTA sources |
REFUSED: NotImplementedError: run_periodic_job(method=’aiccm’, integral_engine=…): correlation=’dlpno-mp2’ is not available on the MANTA torus sources. Only canonic… |
- |
1 x 1 |
20.8 |
block storage |
PASS dE=8.9e-16 (tol 1.0e-09), dEc=1.2e-14 (tol 1.0e-08) |
backend libint; storage block; fold serial x1 |
1 x 1 |
21.1 |
block-star storage |
REFUSED: NotImplementedError: run_periodic_job: correlation=’dlpno-mp2’ on a RHF reference has no reduced kernel for aiccm_cderi_storage=’block-star’; only correlatio… |
- |
1 x 1 |
7.0 |
2 OpenMP threads |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08) |
backend libint; storage dense; fold serial x1 |
1 x 2 |
10.9 |
2 MPI ranks |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08); rank spread 0 |
backend libint; storage dense; fold mpi-owned-q-channels x2 |
2 x 1 |
19.2 |
3 MPI ranks |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08); rank spread 0 |
backend libint; storage dense; fold mpi-owned-q-channels x3 |
3 x 1 |
19.1 |
control_start=direct |
NOT-RUN: control_start applies to the Gamma SCF gradient only |
- |
1 x 1 |
0.0 |
DLPNO-MP2 energy, Pipek-Mezey with truncation¶
Fixture: H2 + He in a 6 bohr cube, 6-31G, (1,1,1) torus, run_ccm_rhf_direct (1e-12/1e-9) + ccm_dlpno_mp2(localize=’pm’, tcut_pno=1e-5, tcut_mkn=5e-2); library call (the front door passes no truncation).
Configuration |
Outcome |
Executed (rank 0 record) |
Ranks x threads |
Wall (s) |
|---|---|---|---|---|
baseline |
BASELINE E=-4.1083979668, Ec=-0.0255657383 |
storage dense; fold serial x1 |
1 x 1 |
14.4 |
MANTA sources |
REFUSED: NotImplementedError: run_periodic_job(method=’aiccm’, integral_engine=…): correlation=’dlpno-mp2’ is not available on the MANTA torus sources. Only canonic… |
- |
1 x 1 |
8.3 |
block storage |
PASS dE=1.8e-15 (tol 1.0e-09), dEc=2.8e-17 (tol 1.0e-08) |
storage block; fold serial x1 |
1 x 1 |
16.0 |
block-star storage |
PASS dE=1.8e-15 (tol 1.0e-09), dEc=2.8e-17 (tol 1.0e-08) |
storage block-star; fold serial x1 |
1 x 1 |
15.4 |
2 OpenMP threads |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08) |
storage dense; fold serial x1 |
1 x 2 |
11.3 |
2 MPI ranks |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08); rank spread 0 |
storage dense; fold mpi-owned-q-channels x2 |
2 x 1 |
17.7 |
3 MPI ranks |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08); rank spread 0 |
storage dense; fold mpi-owned-q-channels x3 |
3 x 1 |
16.8 |
control_start=direct |
NOT-RUN: control_start applies to the Gamma SCF gradient only |
- |
1 x 1 |
0.0 |
RHF SCF gradient¶
Fixture: H2 in a 6 bohr cube (z 2.2/3.8), STO-3G, (2,1,1) torus, run_ccm_rhf_direct (1e-12/1e-9) + run_ccm_direct_gradient. The forces come from the multi-k GDF control the gradient runs on the same Hamiltonian, so the storage and fold axes change the direct SCF (energy, parity gate, seeded density) and not the force assembly. MANTA cell through the front door (optimize=True).
Configuration |
Outcome |
Executed (rank 0 record) |
Ranks x threads |
Wall (s) |
|---|---|---|---|---|
baseline |
BASELINE E=-1.1931824118 |
storage dense; fold serial x1; control independent (3 cycles) |
1 x 1 |
165.5 |
MANTA sources |
REFUSED: NotImplementedError: run_periodic_job(method=’aiccm’, integral_engine=…): gradients, relaxation and force constants are not available on the MANTA torus so… |
- |
1 x 1 |
7.2 |
block storage |
PASS dE=2.2e-16 (tol 1.0e-09), dG=0 (tol 1.0e-07) |
storage block; fold serial x1; control independent (3 cycles) |
1 x 1 |
144.6 |
block-star storage |
PASS dE=2.2e-16 (tol 1.0e-09), dG=0 (tol 1.0e-07) |
storage block-star; fold serial x1; control independent (3 cycles) |
1 x 1 |
201.9 |
2 OpenMP threads |
PASS dE=0 (tol 1.0e-09), dG=2.6e-14 (tol 1.0e-07) |
storage dense; fold serial x1; control independent (3 cycles) |
1 x 2 |
198.8 |
2 MPI ranks |
PASS dE=0 (tol 1.0e-09), dG=0 (tol 1.0e-07); rank spread 0 |
storage dense; fold mpi-owned-q-channels x2; control independent (3 cycles) |
2 x 1 |
373.6 |
3 MPI ranks |
PASS dE=0 (tol 1.0e-09), dG=0 (tol 1.0e-07); rank spread 0 |
storage dense; fold mpi-owned-q-channels x3; control independent (3 cycles) |
3 x 1 |
157.1 |
control_start=direct |
PASS dE=0 (tol 1.0e-12), dG=2.0e-12 (tol 1.0e-07) |
storage dense; fold serial x1; control direct (2 cycles) |
1 x 1 |
125.2 |
RI-MP2 gradient¶
Fixture: as the RHF SCF gradient row; run_ccm_ri_mp2_gradient on the supplied direct RHF (it always seeds its control from the direct state). MANTA cell through the front door (correlation=’mp2’, optimize=True).
Configuration |
Outcome |
Executed (rank 0 record) |
Ranks x threads |
Wall (s) |
|---|---|---|---|---|
baseline |
BASELINE E=-1.1931824118, Ec=-0.0111568502 |
storage dense; fold serial x1 |
1 x 1 |
674.4 |
MANTA sources |
REFUSED: NotImplementedError: run_periodic_job(method=’aiccm’, integral_engine=…): gradients, relaxation and force constants are not available on the MANTA torus so… |
- |
1 x 1 |
13.7 |
block storage |
PASS dE=2.2e-16 (tol 1.0e-09), dEc=2.1e-17 (tol 1.0e-08), dG=1.2e-12 (tol 1.0e-07) |
storage block; fold serial x1 |
1 x 1 |
240.0 |
block-star storage |
PASS dE=2.2e-16 (tol 1.0e-09), dEc=2.1e-17 (tol 1.0e-08), dG=1.2e-12 (tol 1.0e-07) |
storage block-star; fold serial x1 |
1 x 1 |
222.5 |
2 OpenMP threads |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08), dG=3.1e-14 (tol 1.0e-07) |
storage dense; fold serial x1 |
1 x 2 |
146.4 |
2 MPI ranks |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08), dG=0 (tol 1.0e-07); rank spread 0 |
storage dense; fold mpi-owned-q-channels x2 |
2 x 1 |
274.8 |
3 MPI ranks |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08), dG=0 (tol 1.0e-07); rank spread 0 |
storage dense; fold mpi-owned-q-channels x3 |
3 x 1 |
293.7 |
control_start=direct |
NOT-RUN: run_ccm_ri_mp2_gradient always seeds its control from the direct state |
- |
1 x 1 |
0.0 |
DLPNO-MP2 frozen-selection gradient (Pipek-Mezey, truncated)¶
Fixture: as the truncated DLPNO-MP2 energy row; run_ccm_dlpno_mp2_gradient on the supplied direct RHF (localize=’pm’, tcut_pno=1e-5, tcut_mkn=5e-2; selections frozen at the reference geometry). MANTA cell through the front door (correlation=’dlpno-mp2’, optimize=True).
Configuration |
Outcome |
Executed (rank 0 record) |
Ranks x threads |
Wall (s) |
|---|---|---|---|---|
baseline |
BASELINE E=-4.1083979668, Ec=-0.0255657383 |
storage dense; fold serial x1 |
1 x 1 |
200.6 |
MANTA sources |
REFUSED: NotImplementedError: run_periodic_job(method=’aiccm’, integral_engine=…): correlation=’dlpno-mp2’ is not available on the MANTA torus sources. Only canonic… |
- |
1 x 1 |
2.4 |
block storage |
PASS dE=1.8e-15 (tol 1.0e-09), dEc=3.5e-17 (tol 1.0e-08), dG=7.3e-14 (tol 1.0e-07) |
storage block; fold serial x1 |
1 x 1 |
255.3 |
block-star storage |
PASS dE=1.8e-15 (tol 1.0e-09), dEc=3.5e-17 (tol 1.0e-08), dG=7.3e-14 (tol 1.0e-07) |
storage block-star; fold serial x1 |
1 x 1 |
258.4 |
2 OpenMP threads |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08), dG=1.4e-14 (tol 1.0e-07) |
storage dense; fold serial x1 |
1 x 2 |
132.1 |
2 MPI ranks |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08), dG=0 (tol 1.0e-07); rank spread 0 |
storage dense; fold mpi-owned-q-channels x2 |
2 x 1 |
215.1 |
3 MPI ranks |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08), dG=0 (tol 1.0e-07); rank spread 0 |
storage dense; fold mpi-owned-q-channels x3 |
3 x 1 |
260.4 |
control_start=direct |
NOT-RUN: run_ccm_dlpno_mp2_gradient always seeds its control from the direct state |
- |
1 x 1 |
0.0 |
chi (variant="chi")¶
RHF energy¶
Fixture: H2 in an 8 bohr cube, STO-3G, (2,1,1) mesh, ri backend, explicit classic ChiCoulombControls(6.2, 6.2, 18, 8, 5, 0.6), front door, SCF 1e-12/1e-9.
Configuration |
Outcome |
Executed (rank 0 record) |
Ranks x threads |
Wall (s) |
|---|---|---|---|---|
baseline |
BASELINE E=-1.1282073149 |
residue transform cyclic x1 tasks [2] |
1 x 1 |
7.3 |
MANTA sources |
REFUSED: NotImplementedError: run_periodic_job(method=’aiccm’, integral_engine=…): integral_engine= is wired for variant=’real-gamma’ only; variant=’chi’ has no MAN… |
- |
1 x 1 |
7.2 |
block storage |
REFUSED: ValueError: run_periodic_job: aiccm_cderi_storage= selects how variant=’real-gamma’ holds its fitted integrals; got variant=’chi’, which has no such storage…. |
- |
1 x 1 |
13.0 |
block-star storage |
REFUSED: ValueError: run_periodic_job: aiccm_cderi_storage= selects how variant=’real-gamma’ holds its fitted integrals; got variant=’chi’, which has no such storage…. |
- |
1 x 1 |
6.3 |
2 OpenMP threads |
PASS dE=0 (tol 1.0e-09) |
residue transform cyclic x1 tasks [2] |
1 x 2 |
4.0 |
2 MPI ranks |
PASS dE=0 (tol 1.0e-09); rank spread 0 |
residue transform cyclic x2 tasks [1, 1] |
2 x 1 |
9.2 |
3 MPI ranks |
PASS dE=0 (tol 1.0e-09); rank spread 0 |
residue transform cyclic x3 tasks [1, 1, 0] |
3 x 1 |
12.2 |
control_start=direct |
NOT-RUN: control_start applies to the Gamma SCF gradient only |
- |
1 x 1 |
0.0 |
RHF energy, four-centre backend¶
Fixture: as chi RHF energy with aiccm_backend=’four_center’ (direct ERIs; MPI output-cell farming under mpirun).
Configuration |
Outcome |
Executed (rank 0 record) |
Ranks x threads |
Wall (s) |
|---|---|---|---|---|
baseline |
BASELINE E=-1.1345221230 |
output-cell farming cyclic x1 tasks [699]; residue transform cyclic x1 tasks [2] |
1 x 1 |
9.5 |
MANTA sources |
REFUSED: NotImplementedError: run_periodic_job(method=’aiccm’, integral_engine=…): integral_engine= is wired for variant=’real-gamma’ only; variant=’chi’ has no MAN… |
- |
1 x 1 |
7.1 |
block storage |
REFUSED: ValueError: run_periodic_job: aiccm_cderi_storage= selects how variant=’real-gamma’ holds its fitted integrals; got variant=’chi’, which has no such storage…. |
- |
1 x 1 |
6.3 |
block-star storage |
REFUSED: ValueError: run_periodic_job: aiccm_cderi_storage= selects how variant=’real-gamma’ holds its fitted integrals; got variant=’chi’, which has no such storage…. |
- |
1 x 1 |
4.6 |
2 OpenMP threads |
PASS dE=0 (tol 1.0e-09) |
output-cell farming cyclic x1 tasks [699]; residue transform cyclic x1 tasks [2] |
1 x 2 |
6.0 |
2 MPI ranks |
PASS dE=4.4e-16 (tol 1.0e-09); rank spread 0 |
output-cell farming cyclic x2 tasks [350, 349]; residue transform cyclic x2 tasks [1, 1] |
2 x 1 |
10.8 |
3 MPI ranks |
PASS dE=4.4e-16 (tol 1.0e-09); rank spread 0 |
output-cell farming cyclic x3 tasks [233, 233, 233]; residue transform cyclic x3 tasks [1, 1, 0] |
3 x 1 |
14.6 |
control_start=direct |
NOT-RUN: control_start applies to the Gamma SCF gradient only |
- |
1 x 1 |
0.0 |
RI-MP2 energy¶
Fixture: as chi RHF energy, correlation=’mp2’.
Configuration |
Outcome |
Executed (rank 0 record) |
Ranks x threads |
Wall (s) |
|---|---|---|---|---|
baseline |
BASELINE E=-1.1282073149, Ec=-0.0118223726 |
residue transform cyclic x1 tasks [2] |
1 x 1 |
11.8 |
MANTA sources |
REFUSED: NotImplementedError: run_periodic_job(method=’aiccm’, integral_engine=…): integral_engine= is wired for variant=’real-gamma’ only; variant=’chi’ has no MAN… |
- |
1 x 1 |
6.2 |
block storage |
REFUSED: ValueError: run_periodic_job: aiccm_cderi_storage= selects how variant=’real-gamma’ holds its fitted integrals; got variant=’chi’, which has no such storage…. |
- |
1 x 1 |
5.4 |
block-star storage |
REFUSED: ValueError: run_periodic_job: aiccm_cderi_storage= selects how variant=’real-gamma’ holds its fitted integrals; got variant=’chi’, which has no such storage…. |
- |
1 x 1 |
5.4 |
2 OpenMP threads |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08) |
residue transform cyclic x1 tasks [2] |
1 x 2 |
10.5 |
2 MPI ranks |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08); rank spread 0 |
residue transform cyclic x2 tasks [1, 1] |
2 x 1 |
13.7 |
3 MPI ranks |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08); rank spread 0 |
residue transform cyclic x3 tasks [1, 1, 0] |
3 x 1 |
10.3 |
control_start=direct |
NOT-RUN: control_start applies to the Gamma SCF gradient only |
- |
1 x 1 |
0.0 |
RI-MP2 gradient (explicit controls, selections frozen)¶
Fixture: tilted H2 in a 6 bohr cube, STO-3G, (2,1,1) mesh, ChiCoulombControls(7, 7, 18, 7.8, 5, 0.6), def2-svp-jk, prepare_chi_hf_reference + run_chi_mp2 + compute_aiccm2026dev_b_mp2_gradient; MANTA and storage cells through the front door (correlation=’mp2’, optimize=True).
Configuration |
Outcome |
Executed (rank 0 record) |
Ranks x threads |
Wall (s) |
|---|---|---|---|---|
baseline |
BASELINE E=-1.1898031540, Ec=-0.0098782494 |
- |
1 x 1 |
8.6 |
MANTA sources |
REFUSED: NotImplementedError: run_periodic_job(method=’aiccm’, integral_engine=…): integral_engine= is wired for variant=’real-gamma’ only; variant=’chi’ has no MAN… |
- |
1 x 1 |
8.8 |
block storage |
REFUSED: ValueError: run_periodic_job: aiccm_cderi_storage= selects how variant=’real-gamma’ holds its fitted integrals; got variant=’chi’, which has no such storage…. |
- |
1 x 1 |
5.1 |
block-star storage |
REFUSED: ValueError: run_periodic_job: aiccm_cderi_storage= selects how variant=’real-gamma’ holds its fitted integrals; got variant=’chi’, which has no such storage…. |
- |
1 x 1 |
5.5 |
2 OpenMP threads |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08), dG=1.4e-16 (tol 1.0e-07) |
- |
1 x 2 |
8.8 |
2 MPI ranks |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08), dG=0 (tol 1.0e-07); rank spread 0 |
- |
2 x 1 |
8.8 |
3 MPI ranks |
PASS dE=0 (tol 1.0e-09), dEc=0 (tol 1.0e-08), dG=0 (tol 1.0e-07); rank spread 0 |
- |
3 x 1 |
9.8 |
control_start=direct |
NOT-RUN: control_start applies to the Gamma SCF gradient only |
- |
1 x 1 |
0.0 |
RI-RHF SCF gradient (explicit controls)¶
Fixture: tilted H2 in a 6 bohr cube, STO-3G, (3,1,1) mesh, ChiCoulombControls(7, 7, 18, 7.8, 5, 0.6), def2-svp-jk, run_aiccm2026dev_b_rhf + compute_aiccm2026dev_b_gradient; MANTA and storage cells through the front door (optimize=True).
Configuration |
Outcome |
Executed (rank 0 record) |
Ranks x threads |
Wall (s) |
|---|---|---|---|---|
baseline |
BASELINE E=-1.1829436234 |
- |
1 x 1 |
7.1 |
MANTA sources |
REFUSED: NotImplementedError: run_periodic_job(method=’aiccm’, integral_engine=…): integral_engine= is wired for variant=’real-gamma’ only; variant=’chi’ has no MAN… |
- |
1 x 1 |
6.4 |
block storage |
REFUSED: ValueError: run_periodic_job: aiccm_cderi_storage= selects how variant=’real-gamma’ holds its fitted integrals; got variant=’chi’, which has no such storage…. |
- |
1 x 1 |
4.6 |
block-star storage |
REFUSED: ValueError: run_periodic_job: aiccm_cderi_storage= selects how variant=’real-gamma’ holds its fitted integrals; got variant=’chi’, which has no such storage…. |
- |
1 x 1 |
2.7 |
2 OpenMP threads |
PASS dE=0 (tol 1.0e-09), dG=0 (tol 1.0e-07) |
- |
1 x 2 |
4.9 |
2 MPI ranks |
PASS dE=0 (tol 1.0e-09), dG=0 (tol 1.0e-07); rank spread 0 |
- |
2 x 1 |
5.0 |
3 MPI ranks |
PASS dE=0 (tol 1.0e-09), dG=0 (tol 1.0e-07); rank spread 0 |
- |
3 x 1 |
5.5 |
control_start=direct |
NOT-RUN: control_start applies to the Gamma SCF gradient only |
- |
1 x 1 |
0.0 |