Page 1 of 1

Manual says .DFTTHR takes 2 values, but needs 3

Posted: 05 Nov 2020, 10:56
by aba
Hi,
The Dalton 2020.0 Manual states that the keyword .DFTTHRin the *DFT INPUT section takes two numbers:
The first number determines the density threshold (contributions to a property from places where the density is below the threshold will be skipped) and the second one ā€“ orbital threshold (orbitals are assumed to be exactly 0 if they are below the threshold). The default value for DFTHR0 is 1.0Dāˆ’9 and for DFTRHI is 1.0Dāˆ’13
However, when performing a calculation with these two values set to 0.0, the program crashes (see output
pbe_water_monA.out
(19.11 KiB) Downloaded 3 times
). The screen output contains the message
Fortran runtime error: Bad real number in item 3 of list input
Suggesting that it looks for a third input value. Furthermore when looking at the tests, the file $DALTON_SOURCE/dft/test/linear-high.sh (where $DALTON_SOURCE denotes the source code directory) contains the lines
.DFTTHRESHOLD
0 0 0
again suggesting that it requires three values. With all three values set to 0.0, the program runs successfully. Have I misunderstood the description in the manual or does it indeed require three values? If so, what does the third value denote?

Thanks in advance for your answers!

Re: Manual says .DFTTHR takes 2 values, but needs 3

Posted: 21 Nov 2020, 08:59
by hjaaj
I confirm this error. The third threshold is not used any more in Dalton. I will correct it in the Dalton master branch on gitlab.