[gmx-users] segmentation fault in vanilla TI calculation, death without further noitice

[michael.b]

figured it out ... the new soft core thingy (48-1) does not
seem to be very stable at single precision ...

if i use
sc-alpha                 = 0.5
sc-r-power               = 6

instead of

sc-alpha                 = 0.002
sc-r-power               = 48

it seems to work

this has in fact been observed earlier ... see the comments in the
ethanol solvation free energy example at alchemistry.org

mic



--
View this message in context: http://gromacs.5086.x6.nabble.com/segmentation-fault-in-vanilla-TI-calculation-death-without-further-noitice-tp5015242p5015247.html
Sent from the GROMACS Users Forum mailing list archive at Nabble.com.