Nicolas, i am a new follower. İ find your videos useful. İ will watch them again and again, especially those related with esr studies. But it will be my demand from you. Some calculations take long time when a molecule has a lot of atoms. For now, i am running on the 4-core pc :) For large molecules, i need to send job files (slurms) from my pc to high performance computers (hpc) in remote. İ Wonder if you prepare such video on going from zero (by starting installing orca on hpc ) to different slurms for different job files of orca. Sedat.
@@sedatgumus414 Hi Sedat, I will make a video about this in some time, but the last months I have been very busy and haven't been able to upload many videos. Each HPC cluster is different, and maybe the people in charge of the cluster have already installed orca and you only need to ask for access. If not, you need a file-exchange software like WinSCP or Filezilla to move files and folders from your computer to the cluster and back. And the terminal in Linux or something like Putty in windows to access the cluster and launch jobs and other stuff. Then there are several steps of installation of ORCA, the proper version of OpenMPI, adding the folders to the path and writing the correct .sh or .slurm files. That depends on each cluster. It will take me some time to make a video on that. But maybe the administrators of the cluster can help you much better. I'm not an expert in HPC clusters, and I use what other people have set up for me.
@@niconeuman thank you for the message. You know that ORCA team constantly updates the ORCA software so I am not sure if HPC administer loads the last version. I am very unexperienced with HPC . thanks again for your kind message. I will follow the whole videos related with ORCA. I know you are expert on esr. please let's keep in touch.
Dear Nicolas, is it possible for you to send me the orca esr input file and the corresponding slurm file? Nowadays, i am giving many efforts to prepare slurm & to overcome some mistakes. But i dont know your email adress.
Hi Nicolas, if I understand you right, you did the relaxed surface scan with lower tolerances for an accelerated calculation. Thats a great idea. However, when you subsequently obtain more reliable energies as shown in this video, you are using the non perfectly geometry optimized structures. This can cause slight errors in their energies right?
yes, there can be errors. all approaches do and you have to balance your time and objectives. but the tolerances are lower than significant chemical energies. you could do constrained optimisations on several steps of the scan with tighter bounds, but I'm not 100% sure this can be done using multiple xyz feature alone. if your scan has enough points, and the high level energy maximum does not match the low level energy maximum, it would tell you that you should refine a bit more your scan.
Hi! You said at the end that the accuracy of the more expensive calculation could not be accurate because the used geometry was obtained by a cheaper method. What's the right methodology in this case? Should you do a TS search with the more expensive method on the "TS" found by the cheaper method?
Hi maybe I explained myself wrong. What I meant is that the accuracy should be checked, because when combining geometries from one method with energies from another sometimes the geometry conditions the quality of the final results. I do believe this particular approach to be good, but I warned that people should do their own tests and draw their own conclusions.
What we found is a local maximum in the scan that could be a good guess for a TS We dont know if what we found is a TS because we didn't run Freq calculation. But it can be a starting point. The TS calculations are tricky and have their own requirements. For that I would recommend using tight SCF and geom convergence criteria, and if OptTS doesn't work I would use the NEB method
Hi, it would be very useful if you post your videos in your native language, it is difficult to follow your explanation in English. It is plenty full of videos in English, try in your native language, it may be more constructive.
![Lp. Сердце Вселенной #60 РОЖДЕНИЕ ЛОЛОЛОШКИ [Финал] • Майнкрафт](http://i.ytimg.com/vi/YoR0pAV9FVQ/mqdefault.jpg)
Leave your suggestions for other videos and questions in the comments!
Nicolas, i am a new follower. İ find your videos useful. İ will watch them again and again, especially those related with esr studies. But it will be my demand from you. Some calculations take long time when a molecule has a lot of atoms. For now, i am running on the 4-core pc :) For large molecules, i need to send job files (slurms) from my pc to high performance computers (hpc) in remote. İ Wonder if you prepare such video on going from zero (by starting installing orca on hpc ) to different slurms for different job files of orca. Sedat.
@@sedatgumus414 Hi Sedat, I will make a video about this in some time, but the last months I have been very busy and haven't been able to upload many videos. Each HPC cluster is different, and maybe the people in charge of the cluster have already installed orca and you only need to ask for access. If not, you need a file-exchange software like WinSCP or Filezilla to move files and folders from your computer to the cluster and back. And the terminal in Linux or something like Putty in windows to access the cluster and launch jobs and other stuff.
Then there are several steps of installation of ORCA, the proper version of OpenMPI, adding the folders to the path and writing the correct .sh or .slurm files. That depends on each cluster. It will take me some time to make a video on that. But maybe the administrators of the cluster can help you much better. I'm not an expert in HPC clusters, and I use what other people have set up for me.
@@niconeuman thank you for the message. You know that ORCA team constantly updates the ORCA software so I am not sure if HPC administer loads the last version. I am very unexperienced with HPC . thanks again for your kind message. I will follow the whole videos related with ORCA. I know you are expert on esr. please let's keep in touch.
Dear Nicolas, is it possible for you to send me the orca esr input file and the corresponding slurm file? Nowadays, i am giving many efforts to prepare slurm & to overcome some mistakes. But i dont know your email adress.
Hi Nicolas, if I understand you right, you did the relaxed surface scan with lower tolerances for an accelerated calculation. Thats a great idea. However, when you subsequently obtain more reliable energies as shown in this video, you are using the non perfectly geometry optimized structures. This can cause slight errors in their energies right?
yes, there can be errors. all approaches do and you have to balance your time and objectives. but the tolerances are lower than significant chemical energies. you could do constrained optimisations on several steps of the scan with tighter bounds, but I'm not 100% sure this can be done using multiple xyz feature alone. if your scan has enough points, and the high level energy maximum does not match the low level energy maximum, it would tell you that you should refine a bit more your scan.
Hi! You said at the end that the accuracy of the more expensive calculation could not be accurate because the used geometry was obtained by a cheaper method. What's the right methodology in this case? Should you do a TS search with the more expensive method on the "TS" found by the cheaper method?
Hi maybe I explained myself wrong. What I meant is that the accuracy should be checked, because when combining geometries from one method with energies from another sometimes the geometry conditions the quality of the final results. I do believe this particular approach to be good, but I warned that people should do their own tests and draw their own conclusions.
What we found is a local maximum in the scan that could be a good guess for a TS
We dont know if what we found is a TS because we didn't run Freq calculation. But it can be a starting point. The TS calculations are tricky and have their own requirements. For that I would recommend using tight SCF and geom convergence criteria, and if OptTS doesn't work I would use the NEB method
Hi, it would be very useful if you post your videos in your native language, it is difficult to follow your explanation in English. It is plenty full of videos in English, try in your native language, it may be more constructive.
Thanks for the feedback. I will eventually also do videos in Spanish. At this time I don't have time to record in both languages.