search

yaniyuval / Neural_nework_parameterization

Code used in:"Use of neural networks for stable, accurate and physically consistent parameterization of subgrid atmospheric processes with good performance at reduced precision"
MIT License
4 stars 6 forks source link

Instructions on building sam_code_NN #2

Open dorchard opened 1 year ago

dorchard commented 1 year ago

Would it be possible to get some instructions on building the core SAM+NN model? I've got some of my own process for generating a Makefile in the SCRIPTS directory (https://github.com/Cambridge-ICCS/Neural_nework_parameterization/pull/1/files) but having trouble then building. One thing that's not clear is the choice of the RAD modules (should an exclusive choice be made between RAD_DUMMY, RAD_CAM, RAD_CCM)?

yaniyuval commented 1 year ago

To my understanding these are different radiation schemes that you can choose to run the simulations with.

jatkinson1000 commented 1 year ago

Is there any documentation on the differences between the radiation schemes and which to choose?

I have also run into issues trying to compile this using gcc. Are there restrictions on the compilers that can be used? If you can provide an example of a makefile that you have used to successfully build this that would be useful.

yaniyuval commented 1 year ago

You can look on SAM's user guide. Also you can check references in Khairoutdinov and Randall (2003) which describes SAM whether it gives references for the radiation schemes (possibly later SAM publications will include also relevant references for the radiation schemes).

jatkinson1000 commented 1 year ago

OK, thanks. Do you have links to those?

Presumably you made some modifications to the SAM code to implement the neural net? Did you compile and run the code? If so do you have build information, or a makefile and compiler details at all?

yaniyuval commented 1 year ago

The SAM guide can be found with a simple search in Google (search for "Marat Khairoutdinov SAM guide system for atmospheric modeling"). I did change the code (all the associated code is in the sam library in here). I did compile and ran the code. Send me an email and I will send the build and makefile that worked for me on Cheyenne. I should say that when I moved the code to a different cluster it took a some work to recompile on a different system.