Error: Cholesky decompose failed: the matrix is not positive definite or ill-conditioned.

[Yang Liu]

Dear CP2K users,

    I'm trying to run NVT using cp2k, but this error occurred. Could you please help me solve this problem? Thank you so much!

    attached with my input and xyz file.

[Yingchun Zhang]

You can also try to set EPS_DEFAULT to a smaller value, the default is 1e-10, you can try 1e-12

[Yang Liu]

Dear Travis,

    Thanks for your suggestion! I will try a larger box, see if it fixes the problem.

    Best regards,

    Yang

On Wednesday, November 6, 2019 at 2:00:35 AM UTC+9, Travis wrote:

Hi,

For gas phase simulations, the box size has to be large enough that the electron density is essentially zero at the box edges.

Refer to the message here, https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/MM/POISSON/MT.html

-T

[Yang Liu]

Dear Zhang,

    thanks for your suggestion. I tried EPS_SCF value 1e-12, but the same error occurred.

 

    All the best,

    Yang

[Yang Liu]

Dear Travis,

    Thanks for your help! it seems that using a larger box can solve my problem. What if I want to simulate NVT-MD in metal-salt solution, can I do it the same way?

    All the best,

    Yang

On Wednesday, November 6, 2019 at 2:00:35 AM UTC+9, Travis wrote:

Hi,

For gas phase simulations, the box size has to be large enough that the electron density is essentially zero at the box edges.

Refer to the message here, https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/MM/POISSON/MT.html

-T

On Tuesday, November 5, 2019 at 8:43:11 AM UTC-4, Yang Liu wrote:

[zhangyin...@126.com]

no,  it's eps_default.  eps_scf 1e-6 is also recommend. 

--
You received this message because you are subscribed to a topic in the Google Groups "cp2k" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/cp2k/5KdB95RnUMk/unsubscribe.
To unsubscribe from this group and all its topics, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/d64a9931-931b-4982-be54-21f329a7499b%40googlegroups.com.

[Yang Liu]

Dear Zhang,

    I tried EPS_DEFAULT 1e-12, 1e-14 and 1e-17, all failed. Until now, only if I used a larger box, this problem can be solved.

    Thanks a lot!

    Yang

On Wednesday, November 6, 2019 at 2:12:03 PM UTC+9, zhangyin...@126.com wrote:

no,  it's eps_default.  eps_scf 1e-6 is also recommend. 

-------- Original message --------
From: Yang Liu <3304...@qq.com>
Date: Tue, Nov 5, 2019, 23:43
To: cp2k <cp...@googlegroups.com>
Subject: [CP2K:12468] Re: Error: Cholesky decompose failed: the matrix is not
positive definite or ill-conditioned.

Dear Zhang,

    thanks for your suggestion. I tried EPS_SCF value 1e-12, but the same error occurred.

 

    All the best,

    Yang

On Wednesday, November 6, 2019 at 11:00:34 AM UTC+9, Yingchun Zhang wrote:

You can also try to set EPS_DEFAULT to a smaller value, the default is 1e-10, you can try 1e-12

--
You received this message because you are subscribed to a topic in the Google Groups "cp2k" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/cp2k/5KdB95RnUMk/unsubscribe.

To unsubscribe from this group and all its topics, send an email to cp...@googlegroups.com.