the parameter setting of SSNEB

[YangHuang]

Dear experts,

I am using SSNEB to study the phase transition of AgI.
When setting LCLIMB = F, LNEBCELL = T
I will get the following error:
WARNING: Sub-Space-Matrix is not hermitian in DAV 13

However, when LCLIMB = T, LNEBCELL = T are set, the above error disappears.
So when using SSNEB, which of the above parameter settings is correct?

Thanks in advance.

Yang

[graeme]

I actually don't think that setting LCIMB is what it causing this warning/error. I realize that you made that one change and saw the DAV error, but in fact, this error shows up somewhat frequently in standard calculations. To solve them, we typically switch IALGO, or set ALGO to fast so that it switches from DAV to RMM. I'm sorry this is probably not a satisfying answer, but I just don't see any connection between the climbing image setting and the DAV error. [Note that the error is occurring in your final image, which is far from the saddle.]

I would alway set LCLIMB=T if you are interested in calculating a saddle. But again, this is because you want a saddle - I don't think it will change the likelihood of convergence problems of the electronic structure of the images.

[YangHuang]

Dear Professor,

Very grateful for your help.

Best regards,
Yang