Re: [petsc-users] Global Numbering

2018-03-21 Thread Mohammad Hassan Baghaei 
Yeah! I think! You are completely right! Thanks for your suggestion. I would 
definitely look at them! 

Thanks

Amir

 

From: Matthew Knepley [mailto:knep...@gmail.com] 
Sent: Thursday, March 22, 2018 2:25 AM
To: Mohammad Hassan Baghaei 
Cc: PETSc 
Subject: Re: [petsc-users] Global Numbering

 

On Wed, Mar 21, 2018 at 2:18 PM, Mohammad Hassan Baghaei 
mailto:mhbagh...@mail.sjtu.edu.cn> > wrote:

Sorry ! In fact, I am trying to construct the radius and theta coordinate for 
the circular domain I am having before going to the main local functions. I 
needed that for my local functions. 

 

It sounds like you are having trouble with parallel programming. Maybe get your 
stuff working in serial, and then

look at a book like Using MPI by Gropp, et.al <http://et.al> . or Scientific 
Computing by Eijkhout to understand more about parallelism.

 

  Thanks,

 

Matt

 

From: Matthew Knepley [mailto:knep...@gmail.com <mailto:knep...@gmail.com> ] 
Sent: Thursday, March 22, 2018 1:45 AM
To: Mohammad Hassan Baghaei mailto:mhbagh...@mail.sjtu.edu.cn> >
Cc: PETSc mailto:petsc-users@mcs.anl.gov> >
Subject: Re: [petsc-users] Global Numbering

 

On Wed, Mar 21, 2018 at 1:28 PM, Mohammad Hassan Baghaei 
mailto:mhbagh...@mail.sjtu.edu.cn> > wrote:

I am trying to fill a vector based on global coordinate.

 

You can see how this might be too vague for me to follow.

 

As I am trying to use VecGetValues() on the global coordinate,

 

Okay, this does not work for any parallel vector.

 

I find it that argument goes out of range, since I am using global numbering 
rather than local numbering.

 

It is not the numbering that is a problem. It is that parallel Vectors in PETSc 
are distributed. You cannot directly access

data you do not own. You can send it to one process (for example using 
VecScatterCreateToZero()), but this is of course

not scalable.

 

  Thanks,

 

Matt

 

From: Matthew Knepley [mailto:knep...@gmail.com <mailto:knep...@gmail.com> ] 
Sent: Wednesday, March 21, 2018 10:38 PM
To: Mohammad Hassan Baghaei mailto:mhbagh...@mail.sjtu.edu.cn> >
Cc: PETSc mailto:petsc-users@mcs.anl.gov> >
Subject: Re: [petsc-users] Global Numbering

 

On Wed, Mar 21, 2018 at 8:18 AM, Mohammad Hassan Baghaei 
mailto:mhbagh...@mail.sjtu.edu.cn> > wrote:

Hello

I am trying to create a global vector of coordinates based on the parallelized 
dm. I find it really hard to work with the local numbering, how can I manage to 
create a kind of numbering on global level so that I could easily access to the 
nodes based on the new numbering. Thanks for your time.

 

DMGetCoordinates() gives you a global vector.

DMGetCoordinatesLocal() gives you a local vector.

 

What are you trying to do?

 

  Thanks,

 

 Matt

 

Amir





 

-- 

What most experimenters take for granted before they begin their experiments is 
infinitely more interesting than any results to which their experiments lead.
-- Norbert Wiener

 

https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/> 





 

-- 

What most experimenters take for granted before they begin their experiments is 
infinitely more interesting than any results to which their experiments lead.
-- Norbert Wiener

 

https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/> 





 

-- 

What most experimenters take for granted before they begin their experiments is 
infinitely more interesting than any results to which their experiments lead.
-- Norbert Wiener

 

https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/>

Re: [petsc-users] Global Numbering

2018-03-21 Thread Matthew Knepley 
On Wed, Mar 21, 2018 at 2:18 PM, Mohammad Hassan Baghaei <
mhbagh...@mail.sjtu.edu.cn> wrote:

> Sorry ! In fact, I am trying to construct the radius and theta coordinate
> for the circular domain I am having before going to the main local
> functions. I needed that for my local functions.
>

It sounds like you are having trouble with parallel programming. Maybe get
your stuff working in serial, and then
look at a book like Using MPI by Gropp, et.al. or Scientific Computing by
Eijkhout to understand more about parallelism.

  Thanks,

Matt


> *From:* Matthew Knepley [mailto:knep...@gmail.com]
> *Sent:* Thursday, March 22, 2018 1:45 AM
> *To:* Mohammad Hassan Baghaei 
> *Cc:* PETSc 
> *Subject:* Re: [petsc-users] Global Numbering
>
>
>
> On Wed, Mar 21, 2018 at 1:28 PM, Mohammad Hassan Baghaei <
> mhbagh...@mail.sjtu.edu.cn> wrote:
>
> I am trying to fill a vector based on global coordinate.
>
>
>
> You can see how this might be too vague for me to follow.
>
>
>
> As I am trying to use VecGetValues() on the global coordinate,
>
>
>
> Okay, this does not work for any parallel vector.
>
>
>
> I find it that argument goes out of range, since I am using global
> numbering rather than local numbering.
>
>
>
> It is not the numbering that is a problem. It is that parallel Vectors in
> PETSc are distributed. You cannot directly access
>
> data you do not own. You can send it to one process (for example using
> VecScatterCreateToZero()), but this is of course
>
> not scalable.
>
>
>
>   Thanks,
>
>
>
> Matt
>
>
>
> *From:* Matthew Knepley [mailto:knep...@gmail.com]
> *Sent:* Wednesday, March 21, 2018 10:38 PM
> *To:* Mohammad Hassan Baghaei 
> *Cc:* PETSc 
> *Subject:* Re: [petsc-users] Global Numbering
>
>
>
> On Wed, Mar 21, 2018 at 8:18 AM, Mohammad Hassan Baghaei <
> mhbagh...@mail.sjtu.edu.cn> wrote:
>
> Hello
>
> I am trying to create a global vector of coordinates based on the
> parallelized dm. I find it really hard to work with the local numbering,
> how can I manage to create a kind of numbering on global level so that I
> could easily access to the nodes based on the new numbering. Thanks for
> your time.
>
>
>
> DMGetCoordinates() gives you a global vector.
>
> DMGetCoordinatesLocal() gives you a local vector.
>
>
>
> What are you trying to do?
>
>
>
>   Thanks,
>
>
>
>  Matt
>
>
>
> Amir
>
>
>
>
>
> --
>
> What most experimenters take for granted before they begin their
> experiments is infinitely more interesting than any results to which their
> experiments lead.
> -- Norbert Wiener
>
>
>
> https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/>
>
>
>
>
>
> --
>
> What most experimenters take for granted before they begin their
> experiments is infinitely more interesting than any results to which their
> experiments lead.
> -- Norbert Wiener
>
>
>
> https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/>
>



-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener

https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/>

Re: [petsc-users] Global Numbering

2018-03-21 Thread Mohammad Hassan Baghaei 
Sorry ! In fact, I am trying to construct the radius and theta coordinate for 
the circular domain I am having before going to the main local functions. I 
needed that for my local functions.

 

 

From: Matthew Knepley [mailto:knep...@gmail.com] 
Sent: Thursday, March 22, 2018 1:45 AM
To: Mohammad Hassan Baghaei 
Cc: PETSc 
Subject: Re: [petsc-users] Global Numbering

 

On Wed, Mar 21, 2018 at 1:28 PM, Mohammad Hassan Baghaei 
mailto:mhbagh...@mail.sjtu.edu.cn> > wrote:

I am trying to fill a vector based on global coordinate.

 

You can see how this might be too vague for me to follow.

 

As I am trying to use VecGetValues() on the global coordinate,

 

Okay, this does not work for any parallel vector.

 

I find it that argument goes out of range, since I am using global numbering 
rather than local numbering.

 

It is not the numbering that is a problem. It is that parallel Vectors in PETSc 
are distributed. You cannot directly access

data you do not own. You can send it to one process (for example using 
VecScatterCreateToZero()), but this is of course

not scalable.

 

  Thanks,

 

Matt

 

From: Matthew Knepley [mailto:knep...@gmail.com <mailto:knep...@gmail.com> ] 
Sent: Wednesday, March 21, 2018 10:38 PM
To: Mohammad Hassan Baghaei mailto:mhbagh...@mail.sjtu.edu.cn> >
Cc: PETSc mailto:petsc-users@mcs.anl.gov> >
Subject: Re: [petsc-users] Global Numbering

 

On Wed, Mar 21, 2018 at 8:18 AM, Mohammad Hassan Baghaei 
mailto:mhbagh...@mail.sjtu.edu.cn> > wrote:

Hello

I am trying to create a global vector of coordinates based on the parallelized 
dm. I find it really hard to work with the local numbering, how can I manage to 
create a kind of numbering on global level so that I could easily access to the 
nodes based on the new numbering. Thanks for your time.

 

DMGetCoordinates() gives you a global vector.

DMGetCoordinatesLocal() gives you a local vector.

 

What are you trying to do?

 

  Thanks,

 

 Matt

 

Amir





 

-- 

What most experimenters take for granted before they begin their experiments is 
infinitely more interesting than any results to which their experiments lead.
-- Norbert Wiener

 

https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/> 





 

-- 

What most experimenters take for granted before they begin their experiments is 
infinitely more interesting than any results to which their experiments lead.
-- Norbert Wiener

 

https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/>

Re: [petsc-users] Global Numbering

2018-03-21 Thread Matthew Knepley 
On Wed, Mar 21, 2018 at 1:28 PM, Mohammad Hassan Baghaei <
mhbagh...@mail.sjtu.edu.cn> wrote:

> I am trying to fill a vector based on global coordinate.
>

You can see how this might be too vague for me to follow.


> As I am trying to use VecGetValues() on the global coordinate,
>

Okay, this does not work for any parallel vector.


> I find it that argument goes out of range, since I am using global
> numbering rather than local numbering.
>

It is not the numbering that is a problem. It is that parallel Vectors in
PETSc are distributed. You cannot directly access
data you do not own. You can send it to one process (for example using
VecScatterCreateToZero()), but this is of course
not scalable.

  Thanks,

Matt


> *From:* Matthew Knepley [mailto:knep...@gmail.com]
> *Sent:* Wednesday, March 21, 2018 10:38 PM
> *To:* Mohammad Hassan Baghaei 
> *Cc:* PETSc 
> *Subject:* Re: [petsc-users] Global Numbering
>
>
>
> On Wed, Mar 21, 2018 at 8:18 AM, Mohammad Hassan Baghaei <
> mhbagh...@mail.sjtu.edu.cn> wrote:
>
> Hello
>
> I am trying to create a global vector of coordinates based on the
> parallelized dm. I find it really hard to work with the local numbering,
> how can I manage to create a kind of numbering on global level so that I
> could easily access to the nodes based on the new numbering. Thanks for
> your time.
>
>
>
> DMGetCoordinates() gives you a global vector.
>
> DMGetCoordinatesLocal() gives you a local vector.
>
>
>
> What are you trying to do?
>
>
>
>   Thanks,
>
>
>
>  Matt
>
>
>
> Amir
>
>
>
>
>
> --
>
> What most experimenters take for granted before they begin their
> experiments is infinitely more interesting than any results to which their
> experiments lead.
> -- Norbert Wiener
>
>
>
> https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/>
>



-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener

https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/>

Re: [petsc-users] Global Numbering

2018-03-21 Thread Mohammad Hassan Baghaei 
I am trying to fill a vector based on global coordinate. As I am trying to use 
VecGetValues() on the global coordinate, I find it that argument goes out of 
range, since I am using global numbering rather than local numbering.

 

From: Matthew Knepley [mailto:knep...@gmail.com] 
Sent: Wednesday, March 21, 2018 10:38 PM
To: Mohammad Hassan Baghaei 
Cc: PETSc 
Subject: Re: [petsc-users] Global Numbering

 

On Wed, Mar 21, 2018 at 8:18 AM, Mohammad Hassan Baghaei 
mailto:mhbagh...@mail.sjtu.edu.cn> > wrote:

Hello

I am trying to create a global vector of coordinates based on the parallelized 
dm. I find it really hard to work with the local numbering, how can I manage to 
create a kind of numbering on global level so that I could easily access to the 
nodes based on the new numbering. Thanks for your time.

 

DMGetCoordinates() gives you a global vector.

DMGetCoordinatesLocal() gives you a local vector.

 

What are you trying to do?

 

  Thanks,

 

 Matt

 

Amir





 

-- 

What most experimenters take for granted before they begin their experiments is 
infinitely more interesting than any results to which their experiments lead.
-- Norbert Wiener

 

https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/>

Re: [petsc-users] Global Numbering

2018-03-21 Thread Matthew Knepley 
On Wed, Mar 21, 2018 at 8:18 AM, Mohammad Hassan Baghaei <
mhbagh...@mail.sjtu.edu.cn> wrote:

> Hello
>
> I am trying to create a global vector of coordinates based on the
> parallelized dm. I find it really hard to work with the local numbering,
> how can I manage to create a kind of numbering on global level so that I
> could easily access to the nodes based on the new numbering. Thanks for
> your time.
>
>
DMGetCoordinates() gives you a global vector.
DMGetCoordinatesLocal() gives you a local vector.

What are you trying to do?

  Thanks,

 Matt


> Amir
>



-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener

https://www.cse.buffalo.edu/~knepley/