Local NRT

Local NRT

Postby gregors » Fri Mar 01, 2013 4:16 am

Dear NBO team,

I am trying to perform a single-reference local NRT analysis of a system with 101 atoms. Since some phenyl rings are incorporated in the structure I wanted to use the new local NRT feature to focus on the area of interest and to speed up the calculation. The header of my *.47-File is the following:

>>>
$GENNBO NATOMS=101 NBAS=895 UPPER BODM $END
$NBO
NRT <101 100 96 94 2 5 95 1>
NRTLST=0.1
NRTCHG=2
NRTMEM=10
NRTDRAW
NRTMOL
$END
$COORD
[...]
<<<

With or without the keywords like NRTLST, NRTCHG, etc. I always get the error message:

>>>
User-requested resonance units:
1. <C1,C2,C5,N94,N95,O96,Si100,Cr101>

Warning: low valence density detected in VALDM; consider using NRTFDM option

NRTSTR: unexpected end-of-file
Maximum scratch memory used by GENNBO was 3213608 words
<<<

What does "unexpected end-of-file" mean? Is a local NRT analysis only possible when providing one NRT reference structure via $NRTSTR?
How do I get the correct NRTSTR for the leading NBO lewis formula most convenient. Due to the inclusion of 3CBONDs the CHOOSE-list is not suitable for this.

Thanks a lot for that beautiful new NBO6 version.

Best regards,

Gregor.
gregors
 
Posts: 1
Joined: Fri Mar 01, 2013 4:01 am

Re: Local NRT

Postby ericg » Fri Mar 01, 2013 7:56 am

Gregor,

You're not using $NRTSTR input, are you? It doesn't appear so from the bit of FILE47 that you've provided. Yet NRT is trying to read a $NRTSTR keylist.

Could you e-mail me (glendening@indstate.edu) a copy of your FILE47? I'll try this out myself. There's likely a glitch in the code somewhere.

Thanks for reporting this problem,

Eric
ericg
 
Posts: 277
Joined: Sat Dec 29, 2012 9:31 am

Re: Local NRT

Postby ericg » Sun Mar 03, 2013 10:07 pm

Gregor,

Indeed, there was an uninitialized variable in the NRT routines causing NBO to read a $NRTSTR keylist that didn't exist in your input file. I fixed the code and rebuilt the source and binary distributions. These were uploaded to Wisconsin and should become available for download about 2:00am CST. Go to the download site (https://software.wisc.edu/cgi-bin/chemi ... mistry.cgi) and re-enter your download code; you'll get an updated distribution.

I expect, however, that NRT will fail yet for the calculation that you refer to here. 101 atoms is a significant challenge for NRT. I've got a copy of your calculation running, and after twelve hours the code has generated nearly 1000 candidate resonance structures. NRT will exhaust its memory when the 1001st structure is encountered.

There are modifications that can be made to the algorithm which generates candidate structures. For small molecules, the current algorithm is fine. But for large systems like yours, the algorithm ends up considering far more structures than are necessary. I'd like to continue working with the FILE47 that you forwarded to me.

Also, I noticed that you had used the keyword NRTDRAW. We renamed this keyword NRTCML, and I see that the change never made it into the NBO6 manual. I'll fix the manual too.

Eric
ericg
 
Posts: 277
Joined: Sat Dec 29, 2012 9:31 am


Return to Bug Reports

Who is online

Users browsing this forum: No registered users and 1 guest

cron