INSTALL TRELIS PLUGIN

[David Breitenmoser]

Dear Community

After managing to install the dependencies and DAGMC successfully, the last remaining thing I have to do is to install Trelis and the Trelis plugin. I encountered many issues, some of which I was able to solve and some of which I wasn't. In this context, I have the following issues/questions for you:

1) Cubit/Trelis Software

As you for sure know, very recently, the Trelis line ended and was rebranded it as “Coreform Cubit”. At the moment, I was only able to get a free license for "Cubit Learn". With this license, I could successfully download and activate "Coreform-Cubit-2021.5-Lin64" on my WSL Ubuntu 18.04 LTS system. Is this version in general ok for DAGMC/FLUDAG or do I need another version?

2) Trelis Plugin Link

The first issue with the Trelis plugin is the fact that I can't access the linked box on the Website: https://svalinn.github.io/DAGMC/install/plugin.html, i.e. I get the following error message: "This shared file or folder link has been removed or is unavailable to you". How can I access the linked plugin? Do I have to contact the administrator? Does anybody know who the administrator is? I have the feeling that there might be a newer version available on this box, which is also compatible with the new version of cubit (see next issue).

3) SVALINN/Trelis Plugin from Github

Because I was not able to use the linked plugin, I tried to install the developer version from Github (https://github.com/svalinn/Trelis-plugin). I tried to follow the building/installation instructions in the readme. Because I already installed HDF5, MOAB and DAGMC as described on the DAGMC website, I directly jumped to the section "Build the Plugin". I copied the Githup code into a subfolder "Trelis-plugin" under the same parent folder, where I've also built DAGMC, MOAB and HDF5. Then I run the following command:

cmake  -DCUBIT_ROOT=/opt/Coreform-Cubit-2021.5 \ 

 -DDAGMC_DIR=/home/bredavid/dagmc_bld/DAGMC  \ 

 -DCMAKE_BUILD_TYPE=Release \ 

 -DCMAKE_INSTALL_PREFIX=/home/bredavid/dagmc_bld/Trelis-plugin

In the first attempt, I got an error message that mcnp2cad could not have been found. I therefore copied the corresponding code from https://github.com/svalinn/mcnp2cad to the subfolder in Trelis-plugin/mcnp2cad. After that the building with the above command was successful. Then I tried to run the command "make -j`grep -c processor /proc/cpuinfo`" as described in the readme. At this point I got stuck because I got the following error message:

This seems to indicate that Coreform-Cubit-2021.5 is not compatible with the Trelis plugin provided by svalinn on Github, insn't it? Did I miss something? For your convenience, I've also printed the command output to a txt file attached to this post.

4) Windows Plugin

Is there a Trelis version available for Windows? Under WSl, I can't access the GUI of Cubit easily (only the command prompt). To my knowledge, the command promt is enough for the DAGMC workflow, but a GUI would be nice :-). Or is there a chance that in the near future, a Windows plugin will be provided?

I've invested almost two days to get Fludag working and I feel I'm very close. Any suggestions/tips and tricks are highly appreciated.

Thanks for your support,

Best

David

[David Breitenmoser]

[Baptiste MOUGINOT (BaM)]

Hello David, 

Thank you for putting this out, I’ll take a look.(this should not happens)

Could you share with us the cmake command line you used and the cmake log (basically everything that comes before you ran "make")

Some quick answers: 

1) Cubit Software:

I haven’t try the 2021.5 compatibility with ubuntu 18.04, so I am not sure if this will work (we build the 2021.5 for ubuntu 20.04 at the moment… I don’t know if any incompatibility will rise…)

2) thank for pointing this out, I’ll take a look and fix this.

3) The plugin should be compatible with 2021.5. we have successfully build it (see below) for 2021.5

I’ll take a deeper look to understand why you are experiencing this..

In the mean time, we started to build the plugin as part of our Continuous Integration…

You can find some of the plugin we build there:

https://github.com/svalinn/Trelis-plugin/actions/runs/909219700

Note that we build a plugin for Cubit 2021.5 on Ubuntu 20.04 only, but feel free to try it out it might be working…

Also to avoid having to “Copy/Paste” mcnp2cad you can run "git submodule update --init --recursive” from trellis-plugin folder that should get you the right mcnp2cad version...

4) Windows

We also have an beta plugin build for windows:

You can find it there:

https://github.com/svalinn/Trelis-plugin/actions/runs/926178182

(This has not being tested extensively tho….)

Let us know if you try any of this and run into any problem.

Regards

Baptiste

On Jun 15, 2021, at 09:53, David Breitenmoser <breitenmo...@gmail.com> wrote:

Dear Community

After managing to install the dependencies and DAGMC successfully, the last remaining thing I have to do is to install Trelis and the Trelis plugin. I encountered many issues, some of which I was able to solve and some of which I wasn't. In this context, I have the following issues/questions for you:

1) Cubit/Trelis Software

As you for sure know, very recently, the Trelis line ended and was rebranded it as “Coreform Cubit”. At the moment, I was only able to get a free license for "Cubit Learn". With this license, I could successfully download and activate "Coreform-Cubit-2021.5-Lin64" on my WSL Ubuntu 18.04 LTS system. Is this version in general ok for DAGMC/FLUDAG or do I need another version?

2) Trelis Plugin Link

The first issue with the Trelis plugin is the fact that I can't access the linked box on the Website: https://svalinn.github.io/DAGMC/install/plugin.html, i.e. I get the following error message: "This shared file or folder link has been removed or is unavailable to you". How can I access the linked plugin? Do I have to contact the administrator? Does anybody know who the administrator is? I have the feeling that there might be a newer version available on this box, which is also compatible with the new version of cubit (see next issue).

3) SVALINN/Trelis Plugin from Github

Because I was not able to use the linked plugin, I tried to install the developer version from Github (https://github.com/svalinn/Trelis-plugin). I tried to follow the building/installation instructions in the readme. Because I already installed HDF5, MOAB and DAGMC as described on the DAGMC website, I directly jumped to the section "Build the Plugin". I copied the Githup code into a subfolder "Trelis-plugin" under the same parent folder, where I've also built DAGMC, MOAB and HDF5. Then I run the following command:

cmake  -DCUBIT_ROOT=/opt/Coreform-Cubit-2021.5 \ 

 -DDAGMC_DIR=/home/bredavid/dagmc_bld/DAGMC  \ 

 -DCMAKE_BUILD_TYPE=Release \ 

 -DCMAKE_INSTALL_PREFIX=/home/bredavid/dagmc_bld/Trelis-plugin

In the first attempt, I got an error message that mcnp2cad could not have been found. I therefore copied the corresponding code from https://github.com/svalinn/mcnp2cad to the subfolder in Trelis-plugin/mcnp2cad. After that the building with the above command was successful. Then I tried to run the command "make -j`grep -c processor /proc/cpuinfo`" as described in the readme. At this point I got stuck because I got the following error message:

<command_prompt_log.PNG>

This seems to indicate that Coreform-Cubit-2021.5 is not compatible with the Trelis plugin provided by svalinn on Github, insn't it? Did I miss something? For your convenience, I've also printed the command output to a txt file attached to this post.

4) Windows Plugin

Is there a Trelis version available for Windows? Under WSl, I can't access the GUI of Cubit easily (only the command prompt). To my knowledge, the command promt is enough for the DAGMC workflow, but a GUI would be nice :-). Or is there a chance that in the near future, a Windows plugin will be provided?

I've invested almost two days to get Fludag working and I feel I'm very close. Any suggestions/tips and tricks are highly appreciated.

Thanks for your support,

Best

David

--
You received this message because you are subscribed to the Google Groups "DAGMC Users and Collaborators" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dagmc-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dagmc-users/b80a6e95-0880-4f12-b453-97280f4ad430n%40googlegroups.com.
<log_plugin.txt><command_prompt_log.PNG>

[David Breitenmoser]

Hi Baptiste

Thank you very much for your promt and detailed reply. I really appreciate your support. The cmake command, which I've used is as follows:

cmake -DCUBIT_ROOT=/opt/Coreform-Cubit-2021.5 \

             -DDAGMC_DIR=/home/bredavid/dagmc_bld/DAGMC \

             -DCMAKE_BUILD_TYPE=Release \

             -DCMAKE_INSTALL_PREFIX=/home/bredavid/dagmc_bld/Trelis-plugin 

I've made a screen-shot of the command promt with the entered cmake command and the corresponding output. You can find it in the attachment of this post. Is this enough? Do you need something else?

In the meantime, I will try to setup ubuntu 20.04 and also a version on windows as you have described in https://github.com/svalinn/Trelis-plugin/actions/runs/926178182.

Please let me know, if you have found something.

Thank you again for your great support and help

Best,

David

[Baptiste MOUGINOT (BaM)]

Hi,

Yes this is all I need.

You may want to give a try to the plugin build using 20.04 on you 18.04 (if it works you won’t have to re-set it up :) )

The windows version is very new so we will be happy to get any feedback !

You can find instruction to install the plugin here: https://github.com/svalinn/DAGMC/blob/develop/doc/install/plugin.rst

(We are about to update the website with them)

For the windows plugin it is almost the same, you'll have to locate the Cubit installation folder.

You’ll need to unzip/untar (it might be a tar nested inside a zip file) in the Cubit folder.

Or need to copy the content of the bin/plugin/ folder (form the plugin) into the CubitXXX/bin/plugin folder of your installation.

(It is very similar to linux install by the way)

Again let me know if you experience any problem (I am also available for a quick video chat if needed…)

Baptiste.

To view this discussion on the web visit https://groups.google.com/d/msgid/dagmc-users/09f2d828-531b-4303-a485-081037bdb213n%40googlegroups.com.
<cmake_log1.PNG>

[David Breitenmoser]

Hi Baptiste

Thank you again for your prompt reply. Yes, I will try first to use the plugin for ubuntu 20.04 on my 18.04 system. I wonder, can I simply use the svalinn-plugin_ubuntu-20.04_cubit_2021.5.tgz file from https://github.com/svalinn/Trelis-plugin/actions/runs/909219700 for the installation? If yes, how can I install it? If not, I'm stuck with issue described in the first post.

For the Windows system, I have two questions:

1)  Do I have to install DAGMC and other dependencies as well on Windows or is the plugin something like a stand-alone feature together with a Windows version of cubit? Because at the moment, HDF5, MOAB and DAGMC are only installed on my WSL ubuntu system. 

2) Do I have to use the svalinn-plugin_Windows_2021.5.tar file from https://github.com/svalinn/Trelis-plugin/actions/runs/926178182 or one of the tar files from https://uwmadison.app.box.com/v/svalinn-cubit-plugin/?

Sorry for my many questions, I'm a novice in DAGMC and not yet familiar with all the details of the workflow. Regarding your proposition of a video chat, I think this would be very helpful, if I don't succeed with the mentioned approaches. Let's stay in contact.

Thanks again for your great support

Best

David

[Baptiste MOUGINOT (BaM)]

Hi David, 

Please find my answer inline

On Jun 15, 2021, at 11:43, David Breitenmoser <breitenmo...@gmail.com> wrote:

Hi Baptiste

Thank you again for your prompt reply. Yes, I will try first to use the plugin for ubuntu 20.04 on my 18.04 system. I wonder, can I simply use the svalinn-plugin_ubuntu-20.04_cubit_2021.5.tgz file from https://github.com/svalinn/Trelis-plugin/actions/runs/909219700 for the installation? If yes, how can I install it? If not, I'm stuck with issue described in the first post.

Yes you can, the instruction to install our pre-build plugin can be found here: https://github.com/svalinn/DAGMC/blob/develop/doc/install/plugin.rst

For the Windows system, I have two questions:

1)  Do I have to install DAGMC and other dependencies as well on Windows or is the plugin something like a stand-alone feature together with a Windows version of cubit? Because at the moment, HDF5, MOAB and DAGMC are only installed on my WSL ubuntu system. 

The plugin comes as a standalone (you’ll need Cubit tho :) ) so you don’t need to install either MOAB/DAGMC nor HDF5...

2) Do I have to use the svalinn-plugin_Windows_2021.5.tar file from https://github.com/svalinn/Trelis-plugin/actions/runs/926178182 or one of the tar files from https://uwmadison.app.box.com/v/svalinn-cubit-plugin/?

For windows our most recent plugin build is located in https://github.com/svalinn/Trelis-plugin/actions/runs/926178182 so I would suggest using one of those (I just tested 2021.5 and successfully loaded it)

Sorry for my many questions, I'm a novice in DAGMC and not yet familiar with all the details of the workflow. Regarding your proposition of a video chat, I think this would be very helpful, if I don't succeed with the mentioned approaches. Let's stay in contact.

Don’t worry we know this is not as simple as we’d like...

To summaries our installation workflow: 

To prepare your DAGMC/MCNP/FLUKA geometry input you’ll need:

- Cubit with our plugin installed. (You don’t have to build the plugin yourself, using a pre-build plugin is the recommanded path…)

To run fludag or any DAGMC related MonteCarlo transport:

- HDF5, MOAB and Physics code you want to use, in your case Fluka ?

- DAGMC installed and linked against those (specifically Fluka — as you can’t compile DAGMC without HDF5 or MOAB)

Hope this clarifies things a little...

Baptiste

To view this discussion on the web visit https://groups.google.com/d/msgid/dagmc-users/b56321a7-43f2-486e-ae5d-e92beca4b951n%40googlegroups.com.

[David Breitenmoser]

Hi Baptiste

Now I got it. Thanks for your great support. Yes, my plan is to use Fludag (DAGMC+Fluka). Fluka, MOAB, HDF5 and DAGMC are all setup, linked and ready to go on my ubuntu system. In addition, I've installed cubit 2021.5 on ubuntu as well as on windows. So the only thing missing is really the plugin. I will try your recommended approaches and if something exceeds me I would, with your agreement, come back to you (maybe in a bilateral form to minimize the spam of e-mails for the other group members). 

Best,

David

[Andrew Davis]

Hi David

Coming into this a little late, I'm very glad Baptiste could get you on the right path with regards to the plugin. When it comes to FLUDAG, you will need to build of my branch - https://github.com/makeclean/dagmc/ the FLUKA-CERN branch. This will include recent changes to allow building of DAGMC against either FLUKA-CERN or FLUKA-INFN.

There is a collection of tacit knowledge about running FLUDAG that you should be aware of, but when you get to that stage let me know. The biggest things are,

1) Plotting of geometry using Flair - it is possible to plot FLUDAG geometry using Flair, you need to make one or two modifications to the source code of Flair, im working on getting those changes upstream into Flair. It's pretty fiddly due to some bizarre FLUKA logic, but eventually with the right plot co-ordinates and widths its entirely possible.

2) If you're being super fancy with some custom sources, magnetic fields on so you'll need to modify the CMake but i've got some examples of that in a branch somewhere.

3) If you're just running FluDAG, then the easiest time you will have is with the 'legacy' workflow, where using Cubit you will create groups named "mat:IRON" and then you will in fluka have a MATERIAL IRON card.

I'm sure there's something more, but as a decent starting point that should get you going. Do you mind me asking about your end use application?

Thanks

Andy

To view this discussion on the web visit https://groups.google.com/d/msgid/dagmc-users/9cb725f9-d719-4ac9-a988-0b4f7ff621c0n%40googlegroups.com.

[David Breitenmoser]

Hi Andy

Thanks a lot for your input. I really appreciate that. I will answer first your questions and then I have a small problem as well as some some  questions for you:

----------------------

I) Answers: 

----------------------

1) End use application

The end use application will be spectral gamma-ray detector response  simulations of our NaI(Tl) detectors used in the Swiss airborne gamma-ray spectrometry system. The spectrometer is located in a storage bay inside a transport helicopter during the survey. Therefore, to account for the attenuation/scattering in the helicopter structure, I have to include also the helicopter geometry. As you might know, the helicopter shape is quite complex with a lot of higher order surfaces/bodies. Therefore, simulation with primitive CSG would require strong simplifications and a lot of work to build the CSG geometry in the first place manually or with some support from other software, e.g. MCCAD.

2) Current status

As you suggested and based on a post in the FLUKA forum, I've already installed your branch of FLUDAG/DAGMC ( https://github.com/makeclean/dagmc/). The tests, which are described on the website (https://svalinn.github.io/DAGMC/install/fludag.html) run all through successfully. In addition, thanks to Baptiste's help, I was able to install the plugin on my Windows system. I tested the most important commands in cubit (import mcnp test.i, imprint ,merge ,group , export acis  test.sat overwrite attributes_on, export dagmc "test.h5m" faceting_tolerance 1e-3 make_watertight verbose) and they all work well. 

----------------------

II) Problem:

----------------------

Although the fludag test (./tests/fludag_unit_tests) run through smoothly, I was not able to run the command "mainfludag /path-to-tests/cyl.h5m" to create a mat.inp file. I've checked the folder .../DAGMC/bin and there is indeed a file called mainfludag but I'm not able to run it. Did I something wrong? Do I have to compile something? I've attached two screenshots from the performed test and the error message from the mainfludag command.

----------------------

III) Questions:

----------------------

1) Flair

That would be awesome. So which format will Flair be able to read, acis or this h5m format?

2) custom user-routines

Actually, I'm using not custom sources but custom scoring, i.e. the user-routine usreou.f together with the detect card to get event-by-event energy deposition in our scintillation crystals (see attachment). Is this a problem? Can one run the mainfludag command with an additional executable -e, e.g. $FLUPRO/flutil/rfluka -N0 -M1 -e mainfludag -e usreou_exe -d test.h5m fluka.inp?

3) CAD preprocessing

As described on the papers and on the website, the CAD file needs preprocessing to remove small/non-essential parts. I wonder, can I do that also in Cubit (I've loaded some test.stp files in cubit and I think I can remove some volumes/parts)? The described workflow of DAGMC states that this should/can be done in SpaceClaim. But if it is also possible in cubit, why choose another software?

4) Parallel computing

I plan to use Fludag also on our cluster at the institute. I wonder, is this in theory possible (provided that DAGMC and the dependenies are installed)? I'm already using the plain Fluka code on our cluster and it works well. 

5) Groups in cubit

I wonder, by creating a group in trelis with a specific name X, does this imply the creation of an individual "region" X (in Fluka terms)? Is it problematic to group non-connected bodies (i.e. completely separated by other bodies) to a group in Fludag? 

6) Generall workflow

I've try to summarize the workflow in Fludag, which I plan to do:

  i) Load raw *.step file into cubit

  ii) Delete all non-essential parts (e.g. by using group "smallsurfaces" add surface with area < 1.e-3,...)

  iii) check overlaps & heal: validate vol all; autoheal problem vols

  iii) imprint body all

  iv) merge all

  v) Group bodies (to fluka regions) and add material names

  vi) export dagmc "test.h5m" faceting_tolerance 1e-3 make_watertight verbose ...

  vii) run: "mainfludag test.h5m" to create mat.inp file

  viii) insert mat.inp file in main fluka input file test.inp (which is done manually or using flair as described in https://svalinn.github.io/DAGMC/usersguide/uw2.html)

  ix) check materials/compounds and adapt maybe

  xi) run fludag: $FLUPRO/flutil/rfluka -N0 -M1 -e mainfludag -d test.h5m test.inp

Is this workflow more or less ok?

I know, it's a lot of questions. Maybe we can meet in a video chat to reduce long e-mail conversations.

Thanks a lot for your support

Best,

David

[David Breitenmoser]

[David Breitenmoser]

Hi Andy 

Short update, I've found a solution to my problem described in the last post:

With the following command, I was able to run mainfludag (cf. log in the attachment)

cd /path-to-DAGMC/bin

./mainfludag --dagmc /path-to-tests/dag_file.h5m

However, I didn't find the material snippet file mat.inp, only a vol_id_idx.txt file. In which folder would the mat.inp file be saved? WOuld it be the folder, in which mainfludag was called?

Best,

David

[Andrew Davis]

Hi David

I'll try go inline as best I can

1. Although the fludag test (./tests/fludag_unit_tests) run through smoothly, I was not able to run the command "mainfludag /path-to-tests/cyl.h5m" to create a mat.inp file. I've checked the folder .../DAGMC/bin and there is indeed a file called mainfludag but I'm not able to run it. Did I something wrong? Do I have to compile something? I've attached two screenshots from the performed test and the error message from the mainfludag command.

With the new FluDAG executable I've added a --dagmc flag, do mainfludag --dagmc <dagmc filename> this will then work as expected, producing a mat.inp file containing your ASSIGNMa cards.

1) Flair That would be awesome. So which format will Flair be able to read, acis or this h5m format?

Neither actually, but sort of h5m. So what the plot feature in Flair does is launch pseudoparticles called 'worms' in a 'special' FLUKA run, which basically returns a file containing a big grid of points, the distance between the next intersection and last and what mateiral it was in. Then then use Gnuplot to plot the geometry lines and colour by material. Since the mainfludag executable is DAGMC aware, it does this worm operation in the DAGMC geometry rather than the FLUKA geometry.

2) custom user-routines Actually, I'm using not custom sources but custom scoring, i.e. the user-routine usreou.f together with the detect card to get event-by-event energy deposition in our scintillation crystals (see attachment). Is this a problem? Can one run the mainfludag command with an additional executable -e, e.g. $FLUPRO/flutil/rfluka -N0 -M1 -e mainfludag -e usreou_exe -d test.h5m fluka.inp?

No, in this case you will need to build another mainfludag that includes your custom scoring file, as an example see the branch fluka-cern-source https://raw.githubusercontent.com/makeclean/DAGMC/fluka-cern-source/src/fluka/app/CMakeLists.txt you need to add your files in that CMakefile, which will then compile them in as needed.

3) CAD preprocessing As described on the papers and on the website, the CAD file needs preprocessing to remove small/non-essential parts. I wonder, can I do that also in Cubit (I've loaded some test.stp files in cubit and I think I can remove some volumes/parts)? The described workflow of DAGMC states that this should/can be done in SpaceClaim. But if it is also possible in cubit, why choose another software?

SpaceClaim is really good at doing it, its Raison D'Etre is defeaturing and repair. Cubit's existing is for FEA mesh pre-processing, which is a slightly different use case than ours. I would say only defeature if you have to, if you're happy with the detail and it doesnt break the geometry, you can leave it. But you can do everything you need to do in Cubit, its just a little more tedious.

4) Parallel computing I plan to use Fludag also on our cluster at the institute. I wonder, is this in theory possible (provided that DAGMC and the dependenies are installed)? I'm already using the plain Fluka code on our cluster and it works well. 

Yes, FluDAG will parallelise as well as Fluka, which is not brilliantly, but I presume you have a series of different inputs, each with a different random number seed which you run 'at once' then average the results across? The only issue you might face is how you've implemented it. If you launch a single FLUKA calculation per node on your cluster then you'll be fine. If you've been clever, and written some scripts which actually take the entire node, but then submit NCPU jobs on that node, then you might have problems. This is due to the large memory requirements of CAD problems, and depending on how RAM this problem will take, will determine if that method will work for you.

5) Groups in cubit I wonder, by creating a group in trelis with a specific name X, does this imply the creation of an individual "region" X (in Fluka terms)? Is it problematic to group non-connected bodies (i.e. completely separated by other bodies) to a group in Fludag?

You can think of it as convenience layer, all volumes with a common property go into a group. Each volume is equiavlent to a region in FLUKA terms. It isnt problematic to group disparate volume since they will share a common material property. Its only metadata, it doesnt necessarily imply any topological links.

6) yep that all sounds fine, except i would go for 1e-4 facettng tolerance.

If your interested in cross comparison with say, MCNP6, then it would be worth looking at the UWUW route for metadata.

--
You received this message because you are subscribed to the Google Groups "DAGMC Users and Collaborators" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dagmc-users...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/dagmc-users/ceb546aa-fd4c-4304-bd52-79ed097646cbn%40googlegroups.com.

[Andrew Davis]

Hmm weird on that one, could you screenshot the left hand pane of your cubit window showing your "mat:" groups?

--
You received this message because you are subscribed to the Google Groups "DAGMC Users and Collaborators" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dagmc-users...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/dagmc-users/4c1f08df-b157-4df6-88a2-7cf8e70b8221n%40googlegroups.com.

[David Breitenmoser]

Hi Andy

Thanks for your detailed reply. This helps me a lot. Regarding custom user-routines, I still have some small open questions. For my current installation, I've used the following commands to get your branch of dagmc:

$ git clone https://github.com/makeclean/DAGMC.git

$ cd DAGMC

$ git checkout fluka-cern

So, if I'm not wrong, I've downloaded the fluka-cern branch. So my question is, can I stick with this branch, simply exchange the CMakeLists.txt in the folder DAGMC/src/fluka/app by the one attached to this post and copy the usreou.f file to this folder? There are other CMakeLists.txt in other folders, e.g. DAGMC/src/fluka/CMakeLists.txt, but they don't have to be altered, do they?

After building, I guess I can run Fludag with the new mainfludag executable (including compiled usreou.f) by :

$FLUPRO/flutil/rfluka -N0 -M1 -e mainfludag -d test.h5m fluka.inp

correct?

Thanks for your great support

Best,

david

[Andrew Davis]

Hi David

Only the one in the App dir needs to change I think, you can use the fluke-cern and modify the cmakefile to your own ends. Then yes, you run as normal. This is the same mecanism if you’ve use lfluka before, this is the way you need to build a custom executable with your own usr* stuff.

Thanks

Andy

To view this discussion on the web visit https://groups.google.com/d/msgid/dagmc-users/e1ed418f-b7b1-4b68-8b77-75977505c7a2n%40googlegroups.com.
<cmake.txt>

[David Breitenmoser]

I've used the dag_file.h5m from the tests folder from DAGMC (I didn't have time to build my own example). So I actually don't know the materials... Maybe this file is not compatible with Fludag?

[Andrew Davis]

Quite possibly

To view this discussion on the web visit https://groups.google.com/d/msgid/dagmc-users/e1ed418f-b7b1-4b68-8b77-75977505c7a2n%40googlegroups.com.
<cmake.txt>

[David Breitenmoser]

I see, I will create my own example then. In addition, I will try to run some comparison simulations in fluka and fludag to test the response. If something exceeds me, I will come back to you.

Thanks for your help and support

David