Tinker9 issue

[petraskun]

I have successfully installed both tinker8 and tinker9. Using GPU in tinker9 really speeds up calculations, thank you! However, I run into a problem. When I tried to minimize a structure, tinker9 gives the results shown below in Snapshot 1. However, if after removing file with extension *xyz_2 (to ensure that starting point is the same), tinker 8 is run, results are different (see Snapshot 2 below),

Also if you stop tinker8 after several steps when function values drops below 10^10 (that's somewhat empirical value, do not take it as carved in the stone...) and continue from that point with tinker9, everything runs smoothly (see snapshot 3). The force field used in the calculations is charm22.prm (from your distribution) and tinker.key does not contain any additional info, only path to forcefield file.

It seems that if F value is large at first step (and it may happen if your model file has some clashes), the value overflow happens and somewhere '-' sign is erroneously assigned to the results of calculations. Could you fix that problem? Without that it is not possible to use tinker9 for large-scale calculations in automated regime.

Thank you.

Petras Kundrotas, PhD Associate Research Professor Center for Computational Biology The University of Kansas Lawrence, Kansas 66046 United States of America tel. +1 785 864 1963 email: pkundro@ku.edu LinkedIn: se.linkedin.com/pub/petras-kundrotas/69/b23/251/

----------------------------------------------------- Snapshot 1 ------------------------------------------------------------------------------

/usr/local/tinker/tinker9/bin/minimize9 T233_model06 0.1

 ######################################################################

##########################################################################

Tinker9 -- Software Tools for Molecular Design

Version 1.4.0 February 2023

Copyright (c) Zhi Wang & the Ponder Lab

All Rights Reserved

########################################################################## ######################################################################

Compiled at: 10:26:23 Dec 11 2023 Commit Date: Thu Nov 9 12:11:35 2023 -0800 Commit: 70bd0529

GPU Device : Setting Device ID to 0 from GPU utilization

Limited Memory BFGS Quasi-Newton Optimization :

QN Iter F Value G RMS F Move X Move Angle FG Call Comment

 0   -0.4484D+09 0.1631D+09                                   1

LBFGS -- Normal Termination due to SmallFct

Final Function Value : -448433300.8223 Final RMS Gradient : 163068839.3241 Final Gradient Norm : 19517916398.3256 ---------------------------------------- End Snapshot 1 --------------------------------------------------------------------- ---------------------------------------- Snapshot 2 ---------------------------------------------------------------------

/usr/local/tinker/tinker8/bin/minimize T233_model06 0.1

 ######################################################################

##########################################################################

Tinker --- Software Tools for Molecular Design

Version 8.10.6 October 2023

Copyright (c) Jay William Ponder 1990-2023

All Rights Reserved

########################################################################## ######################################################################

Limited Memory BFGS Quasi-Newton Optimization :

QN Iter F Value G RMS F Move X Move Angle FG Call Comment

 0    0.1402D+18 0.1625D+18                                   1
 1    0.3335D+14 0.1466D+14  0.1402D+18   0.0035    0.00      2    Success
 2    0.2829D+14 0.1194D+14  0.5064D+13   0.0000    0.01      6    Success
 3    0.1996D+14 0.7912D+13  0.8331D+13   0.0001   10.60      7    Success
 4    0.1003D+14 0.3637D+13  0.9927D+13   0.0001   15.81      8    Success
 5    0.5671D+13 0.1929D+13  0.4360D+13   0.0001   17.05      9    Success
 6    0.3090D+13 0.9749D+12  0.2581D+13   0.0001   17.38     10    Success
 7    0.1738D+13 0.5033D+12  0.1353D+13   0.0001   17.51     11    Success
 8    0.9874D+12 0.2583D+12  0.7502D+12   0.0002   17.78     12    Success
 9    0.5704D+12 0.1331D+12  0.4170D+12   0.0002   18.43     13    Success
10    0.3306D+12 0.6842D+11  0.2399D+12   0.0002   19.50     14    Success
11    0.1910D+12 0.3506D+11  0.1395D+12   0.0002   20.59     15    Success
12    0.1105D+12 0.1792D+11  0.8054D+11   0.0002   21.33     16    Success
13    0.6481D+11 0.9187D+10  0.4567D+11   0.0003   21.78     17    Success
14    0.3872D+11 0.4738D+10  0.2609D+11   0.0003   22.14     18    Success
15    0.2328D+11 0.2439D+10  0.1544D+11   0.0003   22.54     19    Success
16    0.1406D+11 0.1252D+10  0.9220D+10   0.0004   22.76     20    Success
17    0.8589D+10 0.6443D+09  0.5470D+10   0.0005   22.75     21    Success
18    0.5320D+10 0.3319D+09  0.3269D+10   0.0005   22.69     22    Success
19    0.3355D+10 0.1722D+09  0.1965D+10   0.0006   22.68     23    Success
20    0.2134D+10 0.9091D+08  0.1221D+10   0.0007   23.03     24    Success
21    0.1344D+10 0.4937D+08  0.7894D+09   0.0009   24.44     25    Success
22    0.7973D+09 0.2437D+08  0.5470D+09   0.0011   19.90     26    Success

---------------------------------------- End snapshot 2 --------------------------------------------------------------------- --------------------------------------------- Snapshot 3 ---------------------------------------------------------------------

/usr/local/tinker/tinker8/bin/minimize T233_model06 0.1

 ######################################################################

##########################################################################

Tinker --- Software Tools for Molecular Design

Version 8.10.6 October 2023

Copyright (c) Jay William Ponder 1990-2023

All Rights Reserved

########################################################################## ######################################################################

Limited Memory BFGS Quasi-Newton Optimization :

QN Iter F Value G RMS F Move X Move Angle FG Call Comment

 0    0.1402D+18 0.1625D+18                                   1
 1    0.3335D+14 0.1466D+14  0.1402D+18   0.0035    0.00      2    Success
 2    0.2829D+14 0.1194D+14  0.5064D+13   0.0000    0.01      6    Success
 3    0.1996D+14 0.7912D+13  0.8331D+13   0.0001   10.60      7    Success
 4    0.1003D+14 0.3637D+13  0.9927D+13   0.0001   15.81      8    Success
 5    0.5671D+13 0.1929D+13  0.4360D+13   0.0001   17.05      9    Success
 6    0.3090D+13 0.9749D+12  0.2581D+13   0.0001   17.38     10    Success
 7    0.1738D+13 0.5033D+12  0.1353D+13   0.0001   17.51     11    Success
 8    0.9874D+12 0.2583D+12  0.7502D+12   0.0002   17.78     12    Success
 9    0.5704D+12 0.1331D+12  0.4170D+12   0.0002   18.43     13    Success
10    0.3306D+12 0.6842D+11  0.2399D+12   0.0002   19.50     14    Success
11    0.1910D+12 0.3506D+11  0.1395D+12   0.0002   20.59     15    Success
12    0.1105D+12 0.1792D+11  0.8054D+11   0.0002   21.33     16    Success
13    0.6481D+11 0.9187D+10  0.4567D+11   0.0003   21.78     17    Success
14    0.3872D+11 0.4738D+10  0.2609D+11   0.0003   22.14     18    Success
15    0.2328D+11 0.2439D+10  0.1544D+11   0.0003   22.54     19    Success
16    0.1406D+11 0.1252D+10  0.9220D+10   0.0004   22.76     20    Success
17    0.8589D+10 0.6443D+09  0.5470D+10   0.0005   22.75     21    Success

--------------------------------------- End snapshot 3 ---------------------------------------------------------------------

[zhi-wang]

Seemed like tinker9 couldn't handle your initial structure. I can't give you more definitive reasons only by speculation.