# Curse of the nano-Hartree

4 years 11 months ago #557 by Nike
Replied by Nike on topic Curse of the nano-Hartree
Greetings,
Before posting to the MOLPRO forum I triple-checked everything, and suspect that the problem might be that I made the GENBAS file using MOLPRO's output, but MOLPRO's output rounds off the last digit:

In the attached figure we can see that the contraction coefficient 0.0000058 became 0.000006 and -0.0000011 became -0.000001.

I am re-doing the calculation now.
With best wishes,
Nike
##### Attachments:

4 years 11 months ago - 4 years 11 months ago #558 by Nike
Replied by Nike on topic Curse of the nano-Hartree
Greetings!
The problem was indeed the fact that the 0.0000001 digit (1e-7) digit in the basis set numbers had been cut off in the MOLPRO output that I used to generate the GENBAS file. But MOLPRO had still been using that 1e-7 digit in the internal calculations, so MOLPRO was using a slightly more optimized basis set which gave a lower energy by 62 nano-Hartree.

With the fixed GENBAS, we have pico-Hartree agreement with MOLPRO;
MRCC (canonical)    -7.432 726 121 754
MRCC (semi-can)     -7.432 726 121 754
MOLPRO              -7.432 726 121 753

I am now proceeding to make another attempt at compiling with the new ovirt.f (and mrcc, xmrcc, goldstone, and combin) to see if the correlation problem with big basis sets goes away.

With best wishes!
Nike
Last edit: 4 years 11 months ago by Nike.

4 years 11 months ago #560 by rolik
Replied by rolik on topic Curse of the nano-Hartree
Dear Nike,

I have started to investigate the slow integral transformation issue. Could you upload the output file again? (Links above do not work.)

Best wishes, Zoltan Rolik

4 years 11 months ago #561 by Nike
Replied by Nike on topic Curse of the nano-Hartree
Dear Zoltan,
Thank you for looking into this!
I just clicked no every one of them and I am getting the output files displayed in my browser!
With best wishes!
Nike

4 years 11 months ago #562 by rolik
Replied by rolik on topic Curse of the nano-Hartree
Dear Nike,

this is the one: github.com/ndattani/AI_ENERGIES/blob/mas...OUR_notConverged.txt

Best wishes, Zoltan

4 years 11 months ago #563 by Nike
Replied by Nike on topic Curse of the nano-Hartree
Dear Zoltan,
That link never showed up in this thread, it must be from a different one.

I am finding this link to be working:

github.com/ndattani/AI_ENERGIES/blob/mas...OUR_notConverged.txt

github.com/ndattani/AI_ENERGIES/blob/mas...OUR_notConverged.txt

It appears that I had changed aug-cc-pCV6Z-NR-Koput to aug-cc-pCV6Z-NR because I am using Koput's basis sets as the "default" here since the Koput series goes up to 8Z and the Peterson series ends at 6Z and I want consistency for good extrapolations.

If I changed the name of a file, still all of my "ab initio energy" calculations can be found here though: github.com/ndattani/AI_ENERGIES/

It is a database of all ab initio energy calculations.

With best wishes!
Nike