STATE_ANALYSIS in AIMD jobs

Questions
#1

Hi!

I wish to track the character of electronic transitions for my system along an excited AIMD trajectory. To do so, I include the STATE_ANALYSIS keyword in my input. The problem is that now in the output, for each root calculated, instead of having the normal list of relevant orbitals involved in the transition with their coefficients, I have ALL orbitals (Occupied orbitals # x Virtual orbitals #). This is done for each time step and for larger molecules/bases the output file grows very quickly. Is there a way around this? Without the STATE_ANALYSIS keyword, the orbitals list looks fine. Below is an example input for water.

$molecule
0 1
  O   -2.1904701    0.5202006    0.0237374
  H   -1.2011437    0.4799213    0.0090967
  H   -2.4798356   -0.3698583   -0.2997841
$end

$rem
   JOBTYPE                    aimd
   EXCHANGE                   wB97X-D
   BASIS                      6-311G*
   CIS_N_ROOTS                3
   SYMMETRY                   off
   SYM_IGNORE                 true
   CIS_SINGLETS               true
   CIS_TRIPLETS               false
   PROJ_TRANSROT              true
   AIMD_STEPS                 5  
   TIME_STEP                  40    
   AIMD_PRINT                 1
   AIMD_INIT_VELOC            thermal
   AIMD_TEMP                  500 
   AIMD_INTEGRATION           vverlet
   CIS_STATE_DERIV            1
   SCF_ALGORITHM              GDM
   SCF_MAX_CYCLES             200
   THRESH                     14
   SET_ITER                   50
   STATE_ANALYSIS = true
$end
#2

What’s happening is that CIS_AMPL_PRINT is being set to 0 for state analysis. You’ll notice that in the first time step, you get the much-reduced list of significant amplitudes only; only on subsequent time steps are all the amplitudes printed. That’s because CIS_AMPL_PRINT gets set to 0 and not switched back. (Clearly this was designed with single-point calculations in mind.)

This is an easy fix, please contact me privately and I can show you exactly what change to make. For anyone else who might be reading this, the workaround is to take snapshots from your AIMD trajectory and only run STATE_ANALYSIS on those.