Cholesky decomposition failed

[Dat Ho]

Dear all,

I am trying to perform ab initio MD simulations on a plain water box containing 512 water molecules using the PW6B95 functional. The same input works flawlessly for smaller boxes containing 64, 128 or 265 molecules, but I can't seem to get it to work for 512 molecules. I always get the error message 'Cholesky decomposition failed'. Can anyone spot an error in the input? Also, what does this error message mean in detail? 

Any help would be much appreciated!

[Johann Pototschnig]

Cholesky decomposition computes a smaller representation of a tensor, which allows more efficient computations. 

It assumes a positive definite matrix, which, off course, numerically can not be guaranteed. 

It can be switched off, but certain functionalities will not work:

https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/SCF.html#CP2K_INPUT.FORCE_EVAL.DFT.SCF.CHOLESKY

https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/SCF/OT.html#CP2K_INPUT.FORCE_EVAL.DFT.SCF.OT.CHOLESKY

Most OT preconditioners use it. 

[Johann Pototschnig]

This problem is often caused by diffuse basis functions, so it might help to change:

BASIS_SET AUX_FIT cFIT3