Difference between revisions of "CamCASP/Bugs/8"
import>Am592 |
import>Am592 |
||
| Line 1: | Line 1: | ||
| − | [[ CamCASP ]] => [[ Bugs ]] => Energy Scan |
+ | [[ CamCASP ]] => [[ CamCASP/Bugs | Bugs ]] => [[ CamCASP/Bugs/8 | Energy Scan ]] |
=Error in Energy Scan= |
=Error in Energy Scan= |
||
Revision as of 14:47, 30 June 2011
CamCASP => Bugs => Energy Scan
Error in Energy Scan
Spotted by Anthony with the adenine-thymine system. Data in
/home/am592/DistProp/systems/adenine-thymine/scan
Anthony had calculated all energy components *before* calling ENERGY-SCAN. One geometry in ENERGY-SCAN was the original geometry, and for this case, the energy components seem correct. That is, they agree with those calculated before ENERGY-SCAN was called.
What this seems to suggest is that not all integrals are being correctly re-calculated when geometries are changed.
Now, tests done on He2 and (H2O)2 do not show these problems. These calculations are in
/home/am592/CamCASP/tests/energy_scan/
But these are small systems. So perhaps this is a large-system problem. I've tried reducing the memory given to the (H2O)2 job to try to force this error, but it is still correct. But this test is not conclusive as the way objects are released from memory is hard to control.
Pyridine also shows this problem. Here, <math>E^{(1)}_{\rm exch}</math> comes out correct, but not <math>E^{(1)}_{\rm elst}</math>. Data in
/home/am592/DistProp/systems/pyridine/scans/overlap/JK-tzvpp/scan3-3000pts
For the larger adenine-thymine system, not even <math>E^{(1)}_{\rm exch}</math> is correct.