Reks localization failed during reks DFT for pyrazine

[xiongyan21]

I am trying reks DFT for pyrazine using the same input as that reks-H2O-bhhlyp-grad-b, except using Huckel guess and with cc-pvdz, and get

REKS localization failed: before 0.00459008, after 0.00459008, angle 0.000000 rad. Using old vectors 1 0 0.000000000 -157.6301193138 1.028095711 -157.6301193138 175.884511699 0.771071783 0.300000000 2 1 0.999984755 -240.5556299788 0.852118459 -82.9255106650 1.009670121 0.590278164 0.300000000 REKS localization: before 0.00336861, after 0.00336580, angle -0.019322 rad. 3 2 0.899999976 -240.7353973301 0.745147561 -0.1797673513 0.898966834 0.790959082 0.300000000 ...

      DENSITY CONVERGED
      -----------------

...

2SI-2SA-REKS(2,2)-BHHLYP states: Ek C{PPS} C_{OSS}

SSR state 0 -264.171533705741 1.00000000 0.00000000 SSR state 1 -264.005178324861 0.00000000 1.00000000 Gap too high!

SSR: the first excited state is reported SSR: unrelaxed occupation numbers: FON( 21) = 1.000000 FON( 22) = 1.000000

SSR: final energy = -264.005178324861

SSR: transition dipole between the S1 and S0 states Dipole = ( -0.00000000, 0.00000000, -0.65643495); |D| = 0.65643495 Debye

The geometry is mp2 optimized with 6-311++G**.

Very Best Regards!

[xiongyan21]

The originat test also has this ... REKS localization failed: before 0.24725867, after 0.24725867, angle 0.000000 rad. Using old vectors

and when Huckel guess is employed it gets ...

2SI-2SA-REKS(2,2)-BHHLYP states: Ek C{PPS} C_{OSS}

SSR state 0 -76.302161073878 1.00000000 0.00000000 SSR state 1 -76.081128951002 0.00000000 1.00000000

...

The original test run gives

2SI-2SA-REKS(2,2)-BHHLYP states: Ek C{PPS} C_{OSS}

SSR state 0 -76.302161865410 1.00000000 0.00000000 SSR state 1 -76.081128159466 0.00000000 1.00000000

---

and a slightly different gradient.

IF ccd and lc-blyp are used in the same input, the first excitation energy of pyrazine is |2|nm deviating from 300 nm, but later things are wrong, i.e., CG iterations cannot converge ... CG: iteration 21, residue43008.949960261663 CG: iteration 22, residue52860.857593244262 CG: iteration 23, residue64497.674938057324 CG: iteration 24, residue78163.604834420432 CG: iteration 25, residue94126.205370176511 CG: iteration 26, residue** CG: iteration 27, residue** ...

Bhhlyp also has failed localization for pyrazine and unconverged CG.

Test reks-H2O-lc-blyp-grad-011 uses no DIIs and no CG, Huckel guess and the original get a slightly different gradient. When that is applied to pyrazine with ccd, the following error appears 109 H 10 Y -0.463075 0.000000 -0.532547 0.334256 110 H 10 Z -0.000000 -0.015503 -0.000000 -0.000000 ...... END OF REKS SCF CALCULATION ...... ... ddikick.x: application process 0 quit unexpectedly. ddikick.x: Fatal error detected. The error is most likely to be in the application, so check for input errors, disk space, memory needs, application bugs, etc. ddikick.x will now clean up all processes, and exit... DDI Process 1: terminated upon request. DDI Process 2: terminated upon request.

When 6-31G** is used, it seems it will take much long time to finish and I don't know whether it gets stuck.

Very Best Regards!