I have to do the TS optimization of a very large structure. Initially, I used the typical opt(ts,calcfc,noeigentest) at the route section, but during the optimization process (after the CalcFC part) the job was interrupted and terminated.

I want to continue the job by using the opt(restart) option and the chk file of the interrupted job. Below is what I found in the FAQ page of the Gaussian website:

If CalcFC or RCFC was among the original options, then you can omit it if you do not need to compute/retrieve force constants, e.g., if the optimization was beyond the first step. On the other hand, if you add CalcFC to the option list along with Restart, the force constant will be computed at the beginning/next point of the restarted optimization.

However I also found this from the Optimization page (example tab) of the Gaussian website:

Note that you must include CalcFC to compute the Hessian at the first point of the restarted job.

This made me confused, because (from my understanding) according to the first source the CalcFC option can be omitted but from the second source it can't be omitted.

I want to skip the CalcFC part because it took a long time to compute, so I omitted the CalcFC option, i.e. I just used

opt(restart,ts,noeigentest) geom(check) guess(read)

but it ended with an error saying that the CalcFC is needed.

Is there any way to restart/continue the interrupted TS optimization job without the CalcFC option?

  • 2
    $\begingroup$ Could you include the error message you get in the post? I believe you are misinterpreting the second quote. I think it could be rephrased as "If you want to compute the Hessian at the first point of a restart job, then you must include CalcFC". $\endgroup$
    – Tyberius
    Commented Aug 15, 2022 at 20:21
  • $\begingroup$ Thx for helping ! , below is the error message I got. >Opt=TS requires force constants or ModRedundant. >Error termination via Lnk1e in /usr/local/g16/g16/l1.exe at Tue Aug 16 06:16:32 2022. $\endgroup$
    – Duffoon
    Commented Aug 16, 2022 at 3:21

1 Answer 1


As your error message suggests, even after a restart, an optimization to a transition state requires either force constants or specifying to use ModRedundant coordinates. So if you don't want to eat the cost of running CalcFC again, you could try adding ModRedundant.

Alternatively, I have generally found opt=ts to be tricky, where you have to play with a lot of settings to get convergence. If you have reasonable guess structures for the reactant/ts/product, you may want to try a QST2 or QST3 calculation to search for the transition state. The same caveats apply of needing force constants or mod redundant coordinates, but having information about the various states of the reaction tends to help with convergence.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .