I would like to obtain the Cartesian gradient from a Gaussian formatted checkpoint file with more digits than 8 (specified in scientific notation). Is there any way to do this?

  • 3
    $\begingroup$ I doubt if this is necessary, since the numerical error of a gradient is typically on the order of 10^(-5), or at best 10^(-6) $\endgroup$
    – wzkchem5
    Aug 10, 2021 at 15:00
  • 1
    $\begingroup$ +1 but I agree with wzkchem5, and additionally, the answer to your question would be to change the source code and recompile, so perhaps don't worry about it! $\endgroup$ Aug 10, 2021 at 15:04
  • 1
    $\begingroup$ I'm afraid that if you only have the formatted checkpoint file, you won't get more out of it than what you see. You will have to get it from the routines itself. $\endgroup$ Aug 10, 2021 at 16:35
  • 1
    $\begingroup$ I agree with Tyberius. If you have the checkpoint file, you can usually use that as a starting point for further calculations. You can't extract something that simply isn't there. I think there is an iop to adjust the convergence criteria, there might be a switch to control what is printed. But I don't see myself as an expert for that anymore. So the best way to go about is to modify the question. $\endgroup$ Aug 10, 2021 at 21:35
  • 2
    $\begingroup$ I recommend asking the Gaussian support, [email protected]. This is probably the most straightforward way to get a definite answer. This is exactly the kind of technical question that they can answer with certainty. $\endgroup$
    – Hans Wurst
    Aug 12, 2021 at 7:53


Browse other questions tagged .