Jaguar

Jaguar is the main workhorse for non-periodic quantum work. Jaguar is licensed from Schrodinger, and is particularly well suited for calculations of large and/or metal-containing systems.


Helpful Jaguar information:

The default version on the MSC computers is Jaguar 5.5.

The manual can be found here.


Common Jaguar questions and possible solutions:

How do I kill a Jaguar job?
How do I find the jobid?
When I try to kill jaguar I get an error message, claiming the job is unreachable. Now what?
My wavefunction won’t converge.
My SCF crashes immediately after start.
My SCF crashes after a huge energy jump.
The geometry optimization appears to be stuck.
The geometry doesn't look at all like I thought it would.
My optimized geometry has imaginary frequencies.



How do I kill a Jaguar job?
% jaguar kill jobid

How do I find the jobid?
% jaguar jobs

When I try to kill jaguar I get an error message, claiming the job is unreachable. Now what?
The job has terminated but is still listed as active. Manually delete the file .schrodinger/.jobdb/jobid

My wavefunction won’t converge.
Program ends normally with a message such as “SCF ran out of iterations”
Check whether the energy of each iteration is descending or oscillating. If descending, increase the number of maximum iteration. If oscillating, try a different convergence algorithm (check manual)

SCF crashes immediately after start
Most likely a bad wavefunction guess.
Check whether there is something wrong with the geometry and/or electronic makeup of the molecule. If not, there is most likely a problem with the method – the initial guess wavefunction is so poor that the program doesn’t know how to implement it. Try obtaining a wavefunction with a different method – HF, B3LYP, MP2, BLYP - and then use the resulting wavefunction as the starting wavefunction for your desired job. Alternatively, turn off pseudo-spectrals (keyword nops=1) and run a single point.

My SCF crashes after a huge energy jump
Most likely a bug in the code – a combination of factors have lined up just right to cause the crash.
Perturb the geometry of the molecule somewhat (even increasing a bond 0.01 Angstrom might be enough) and try again. If this doesn't work, try optimizing with a different method (HF, MP2 etc) then reoptimizing with the method you used originally. Optimizing without pseudo-spectrals (keyword nops=1) also possible.

The geometry optimization appears to be stuck
The energy oscillates without reaching a minimum.
Decrease the step size (keyword trust=nnn.xxx, where nnn.xxx is less than the default 0.3), repotimize. Alternatively, Perturb the geometry of the molecule and try again.

The geometry doesn't look at all like I thought it would
Check whether the job actually converged (and not just crashed in the middle).
Check your starting guess geometry – did you set it up correctly?
Animate the optimization – did it take any weird paths? Are they reasonable?
Very rarely: Accept that you’ve found something new and unusual. Investigate. Publish.


My optimized geometry has imaginary frequencies
Animate the frequencies in question. Are you able to manually remove them? (for example, if the frequency corresponds to rotation of a methyl group, can you rotate the methyl group and reoptimize?). If not, try reoptimization from scratch, with a different starting geometry