[quarkz]

Hi,

I'm trying to run my simulation from coarse to fine grid. I used to run:

mapFields ../"source dir" -consistent -mapMethod interpolate -sourceTime 0.001

which works fine.

However now my 3D fine grid is rather large @ 2130x900x150

Using mapFields with 1 processor doesn't seem to work due to large memory req.

I'm trying to use either mapFields with parallel source and target or to use mapFieldsPar instead. However, I can't get them to work. I wonder what's wrong with my syntax.

For e.g. I use:

mapFields -mapMethod interpolate -consistent -sourceDecomposeParDict ../sphere-heat-transfer_80x80_test/system/decomposeParDict -parallelTarget -targetDecomposeParDict system/decomposeParDict -sourceTime 5.000001e-05 ../sphere-heat-transfer_80x80_test

or

mpirun -n 4 mapFieldsPar ../sphere-heat-transfer_80x80_test/ -parallel -sourceTime 5.000001e-05

In the 1st, I got a seemingly correct output:

Source mesh size: 6400

Target processor 0
mesh size: 2065

Mapping fields for time 5.000001e-05

interpolating wallHeatFlux
interpolating p
interpolating T
interpolating U

Target processor 1
mesh size: 2066

Mapping fields for time 5.000001e-05

interpolating wallHeatFlux
interpolating p
interpolating T
interpolating U

Target processor 2
mesh size: 2030

Mapping fields for time 5.000001e-05

interpolating wallHeatFlux
interpolating p
interpolating T
interpolating U

Target processor 3
mesh size: 2031

Mapping fields for time 5.000001e-05

interpolating wallHeatFlux
interpolating p
interpolating T
interpolating U

End

But there's nothing inside the processor* folders.

Can someone guide me?

Thanks!

[lebc]

Hi,

Do you have a source that was run in parallel? I mean, are you trying to map from a reconstructed case to a parallel case?

The cases I run have parallel source and target, the command I run in the target path is:

mapFields path/to/source -sourceTime latestTime -parallelSource -parallelTarget -consistent

My cases have a much smaller mesh (~10^6 elements, your case is almost 300 times bigger...).

Best Regards,
Luis

[quarkz]

Thanks for the help lebc!

I tried your mtd and it seems to work. I need to 1st create the processors' directories by using decomposePar with the same time directory. I just renamed my 0 dir to that same time directory. It ran but I got the error:

Source processor 2

Source time: 0.0001
Target time: 0.0001
mesh size: 900

Target processor 0
mesh size: 3600
--> FOAM Warning :
From function void Foam::meshToMesh0::calcAddressing()
in file meshToMesh0/calculateMeshToMesh0Addressing.C at line 155
Source patch wall has no faces. Not performing mapping for it.

Mapping fields for time 0.0001

interpolating p
interpolating T
interpolating U

Target processor 1
mesh size: 3600
--> FOAM Warning :
From function void Foam::meshToMesh0::calcAddressing()
in file meshToMesh0/calculateMeshToMesh0Addressing.C at line 155
Source patch wall has no faces. Not performing mapping for it.

Mapping fields for time 0.0001

I also tried another command:

mpirun -n 4 mapFieldsPar -consistent -parallel -sourceTime latestTime ../path/to/source

for 4 processors. The output are different although there's no error msg. The file size for both are the same but "diff" says they are different in contents. I will have to test to see if either or both works...

[lebc]

Hi,

I do the same thing, I create a folder with dummy files in the time step I want to map.

What is your OF version? I ask because Iḿ using 4.1 and there was a correction of a bug right after I compiled my version, I think I was receiving this kind of messages before.

You can check this post (number 5), just to make sure that you have this fix compiled in your version:

mapFields

Let me know if it works!

Best Regards,
Luis