running error
Moderators: myrta gruning, andrea marini, Daniele Varsano, Conor Hogan
-
- Posts: 68
- Joined: Mon Dec 05, 2016 5:09 pm
running error
Dear YAMBO developers,
I am using yambo Version 4.5.1 Revision 165 to calculate GGA-RPA. However, the calculations always crashed with the error shown below when I change in the input file of yambo.in "RL" by "Ry":
_| _| _|_| _| _| _|_|_| _|_|
_| _| _| _| _|_| _|_| _| _| _| _|
_| _|_|_|_| _| _| _| _|_|_| _| _|
_| _| _| _| _| _| _| _| _|
_| _| _| _| _| _|_|_| _|_|
<---> [01] CPU structure, Files & I/O Directories
<---> [02] CORE Variables Setup
<---> [02.01] Unit cells
<---> [02.02] Symmetries
<---> [02.03] RL shells
<---> [02.04] K-grid lattice
<---> Grid dimensions : 101 100
<---> [02.05] Energies [ev] & Occupations
<---> [03] Transferred momenta grid
<---> [04] Dipoles
<---> [DIP] Checking dipoles header
<---> [WARNING] DIPOLES database not correct or not present
<---> [x,Vnl] computed using 15 projectors
<---> Dipoles: P, V and iR (T): | | [000 <05s> Dipoles: P, V and iR (T): |####### | [017 <10s> Dipoles: P, V and iR (T): |############## | [035 <15s> Dipoles: P, V and iR (T): |##################### | [054 <20s> Dipoles: P, V and iR (T): |############################ | [072 <25s> Dipoles: P, V and iR (T): |#################################### | [090 <27s> Dipoles: P, V and iR (T): |########################################| [100%] 27s(E) 27s(X)
<27s> [DIP] Writing dipoles header
<27s> [05] Optics
<27s> [LA] SERIAL linear algebra
[ERROR] STOP signal received while in :[05] Optics
[ERROR]Allocation of X_par%blc failed
I will appreciate it if anyone could give me any hints about the problem.
Best regards,
Taouil lamia
I am using yambo Version 4.5.1 Revision 165 to calculate GGA-RPA. However, the calculations always crashed with the error shown below when I change in the input file of yambo.in "RL" by "Ry":
_| _| _|_| _| _| _|_|_| _|_|
_| _| _| _| _|_| _|_| _| _| _| _|
_| _|_|_|_| _| _| _| _|_|_| _| _|
_| _| _| _| _| _| _| _| _|
_| _| _| _| _| _|_|_| _|_|
<---> [01] CPU structure, Files & I/O Directories
<---> [02] CORE Variables Setup
<---> [02.01] Unit cells
<---> [02.02] Symmetries
<---> [02.03] RL shells
<---> [02.04] K-grid lattice
<---> Grid dimensions : 101 100
<---> [02.05] Energies [ev] & Occupations
<---> [03] Transferred momenta grid
<---> [04] Dipoles
<---> [DIP] Checking dipoles header
<---> [WARNING] DIPOLES database not correct or not present
<---> [x,Vnl] computed using 15 projectors
<---> Dipoles: P, V and iR (T): | | [000 <05s> Dipoles: P, V and iR (T): |####### | [017 <10s> Dipoles: P, V and iR (T): |############## | [035 <15s> Dipoles: P, V and iR (T): |##################### | [054 <20s> Dipoles: P, V and iR (T): |############################ | [072 <25s> Dipoles: P, V and iR (T): |#################################### | [090 <27s> Dipoles: P, V and iR (T): |########################################| [100%] 27s(E) 27s(X)
<27s> [DIP] Writing dipoles header
<27s> [05] Optics
<27s> [LA] SERIAL linear algebra
[ERROR] STOP signal received while in :[05] Optics
[ERROR]Allocation of X_par%blc failed
I will appreciate it if anyone could give me any hints about the problem.
Best regards,
Taouil lamia
Taouil Lamia
University of Mohammed V
Faculty of science - Rabat
Morocco
University of Mohammed V
Faculty of science - Rabat
Morocco
-
- Posts: 206
- Joined: Fri Jan 31, 2014 11:13 am
Re: running error
Dear Taouil,
this is typically related to memory usage. X_par%blc may be a large matrix with dimensions (G,G',freq), and you need to pay a lot
of attention to the variables governing its size. Surely the number of frequencies, but also the number of reciprocal lattice vectors (RL).
The fact that the error shows up when passing one of these variables from RL to Ry (ie passing from giving the actual number of lattice vectors to giving en energy cutoff in Ry to determine such number), further confirms the suspect.
Start with a lower kinetic energy cutoff in the calculation of the response function until your calculation fits the resources you want to use, then increase the cutoff to convergence while (probably) also increasing the processor/nodes you have.
Also note that not all parallelism schemes are equally efficient in distributing memory.
In order to help further you may probably need to post your input file and tell something more about the system you are studying and the computational resources you have at hand.
take care
Andrea
this is typically related to memory usage. X_par%blc may be a large matrix with dimensions (G,G',freq), and you need to pay a lot
of attention to the variables governing its size. Surely the number of frequencies, but also the number of reciprocal lattice vectors (RL).
The fact that the error shows up when passing one of these variables from RL to Ry (ie passing from giving the actual number of lattice vectors to giving en energy cutoff in Ry to determine such number), further confirms the suspect.
Start with a lower kinetic energy cutoff in the calculation of the response function until your calculation fits the resources you want to use, then increase the cutoff to convergence while (probably) also increasing the processor/nodes you have.
Also note that not all parallelism schemes are equally efficient in distributing memory.
In order to help further you may probably need to post your input file and tell something more about the system you are studying and the computational resources you have at hand.
take care
Andrea
Andrea Ferretti, PhD
CNR-NANO-S3 and MaX Centre
via Campi 213/A, 41125, Modena, Italy
Tel: +39 059 2055322; Skype: andrea_ferretti
URL: http://www.nano.cnr.it
CNR-NANO-S3 and MaX Centre
via Campi 213/A, 41125, Modena, Italy
Tel: +39 059 2055322; Skype: andrea_ferretti
URL: http://www.nano.cnr.it
-
- Posts: 68
- Joined: Mon Dec 05, 2016 5:09 pm
Re: running error
Dear Andrea,
Thank you for your quick reply. You will find the input file in the attachment.
Best,
Lamia
Thank you for your quick reply. You will find the input file in the attachment.
Best,
Lamia
You do not have the required permissions to view the files attached to this post.
Taouil Lamia
University of Mohammed V
Faculty of science - Rabat
Morocco
University of Mohammed V
Faculty of science - Rabat
Morocco
- Daniele Varsano
- Posts: 3816
- Joined: Tue Mar 17, 2009 2:23 pm
- Contact:
Re: running error
Dear Taouil Lamia,
the problem you are encountering it happens when running yambo,
in order to understand the problem please post the yambo input and report file that actually presents the problem.
Best,
Daniele
the problem you are encountering it happens when running yambo,
in order to understand the problem please post the yambo input and report file that actually presents the problem.
Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/
-
- Posts: 68
- Joined: Mon Dec 05, 2016 5:09 pm
Re: running error
Dear Daniele,
Thank you for your reply. yambo.in is in the attachment .
Best,
Lamia
Thank you for your reply. yambo.in is in the attachment .
Best,
Lamia
You do not have the required permissions to view the files attached to this post.
Taouil Lamia
University of Mohammed V
Faculty of science - Rabat
Morocco
University of Mohammed V
Faculty of science - Rabat
Morocco
- Daniele Varsano
- Posts: 3816
- Joined: Tue Mar 17, 2009 2:23 pm
- Contact:
Re: running error
Dear Taouil Lamia,
you are using 600Ry as a cutoff for the dimension of the dielectric function, which is 15 times larger of the cutoff you used for the wave functions.
In general, such a cutoff would correspond to an enormous number of lattice vectors and it would require an enormous amount of memory. Moreover being larger of the cutoff to represent the wfs these vectors are not imported from the QE calculations.
Usually, the calculation is converged in a range of 5-20 Ry depending on the non-homogeneity of the system.
Note that RL and Ry are two alternative ways to assign the dimension of the dielectric matrix, using RL you provide directly the dimension of the matrix, by using Ry you include a number of Gvector contained in the sphere |G|^2/2 < Ecut
Best,
Daniele
PS: you can use the attachments button here below to attach files
you are using 600Ry as a cutoff for the dimension of the dielectric function, which is 15 times larger of the cutoff you used for the wave functions.
In general, such a cutoff would correspond to an enormous number of lattice vectors and it would require an enormous amount of memory. Moreover being larger of the cutoff to represent the wfs these vectors are not imported from the QE calculations.
Usually, the calculation is converged in a range of 5-20 Ry depending on the non-homogeneity of the system.
Note that RL and Ry are two alternative ways to assign the dimension of the dielectric matrix, using RL you provide directly the dimension of the matrix, by using Ry you include a number of Gvector contained in the sphere |G|^2/2 < Ecut
Best,
Daniele
PS: you can use the attachments button here below to attach files
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/
-
- Posts: 68
- Joined: Mon Dec 05, 2016 5:09 pm
Re: running error
Dear Daniele,
Even I choose 20 Ry I get same error and when I run the calculation with RL it done very quickly (2 min). Is it normal?
Which value equal to 20 Ry if we talk RL ?
Best,
Lamia
Even I choose 20 Ry I get same error and when I run the calculation with RL it done very quickly (2 min). Is it normal?
Which value equal to 20 Ry if we talk RL ?
Best,
Lamia
Taouil Lamia
University of Mohammed V
Faculty of science - Rabat
Morocco
University of Mohammed V
Faculty of science - Rabat
Morocco
- Daniele Varsano
- Posts: 3816
- Joined: Tue Mar 17, 2009 2:23 pm
- Contact:
Re: running error
Dear Taouil Lamia,
In order to achieve convergence start with few G vectors and then increase until convergence is reached.
When using a cutoff in Rydberg, the number of corresponding RL is written in the report (X matrix size).
Best,
Daniele
Yes, it is, as the number of RL is totally different.Even I choose 20 Ry I get same error and when I run the calculation with RL it done very quickly (2 min). Is it normal?
In order to achieve convergence start with few G vectors and then increase until convergence is reached.
When using a cutoff in Rydberg, the number of corresponding RL is written in the report (X matrix size).
Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/