Odd behaviour when declaring three states in .fep file but not on .inp file about qsource HOT 6 CLOSED

I will look at that as soon as I can get a working example. This seems to be something critical.

from qsource.

Nahh!
I don't think it's critical
If the user knows what is going on there are no issues.
The parallel implementation runs well when the three lambdas are declared, say:

[lambdas]
1.0 0.0 0.0

In "principle" it should be an easy fix.
Will send you an e-mail in a little while with a crashing example.

from qsource.

Can someone please check if this still exists in the current master version? Otherwise I will close this.

from qsource.

This is still an issue in the current master (test cases attached).
issue11.zip
Wrong inputs fail with major segmentation violation.
Proposed fix would be a check of the number of lambda states in .inp and .fep, if they mismatch kill the calculation.
Warning, this will not work with the qdyn5_r8 binary, because the setting of all precision values to default r8 breaks the reading in of restart files with mismatched precision, or old restarts. Build a debug binary to test it.

from qsource.

This issue is fixed in one of my experimental versions now, by testing the number of states defined in the input against the number in the FEP file, aborting the program if they don't match 08ebaad

from qsource.

Closed because of the previous fix

from qsource.