Hello - I am currently using ASPECT in my geodynamics class. Thank you for having this forum – this is really great! I have two questions;
(1) Is there an example input file for modeling subduction?
(2) (more broadly) is there a set of example input files for class?
Regarding (1), I am hoping to cover it in class, so something simple would be fantastic.
Regarding (2), I am aware that great examples are listed in the cookbook, but some examples can be time consuming and therefore they may not be suitable for class. If some of you have already used ASPECT in class and are willing to share what examples you have used, that would be fantastic.
If you have questions please let me know. I would greatly appreciate your advice.
In terms of a specific subduction example, I don’t think any of the cookbooks in the making are available yet. There is an example in my paper (https://www.solid-earth.net/9/267/2018/) based on the benchmark of Schmeling 2008, that I also used in class, together with more complicated examples. Would that kind of simple setup be of interest to you?
Indeed, fantastic that you are using ASPECT in the classroom!
As Anne mentioned, a number of us are working on subduction problems but there is not a formal cookbook example in the repository yet. I’ll check with a user who I believe is planning to submit one.
There is likely a good deal of overlap with the 2016 tutorial posted above as well as newer material, so I would recommend starting with there.
On a related noted, there are tentative plans to hold a “geodynamic modeling summer school” based on ASPECT in 2020. As such, we would be very interested to hear your feedback and ideas on how to do further develop classroom teaching tools and example using ASPECT!
About question (2)
I also used ASPECT for teaching (master level) students last year. It was only for a couple of hours, so we only did some simple convection in a box models (also because these are the only ones that we were sure were running on the university’s computers!). The input files was thus based on the convection_2d.prm in the cookbook, and we modified it with the students to include some radiogenic heating, viscosity variations with temperature, etc. We managed to do some stagnant-lid convection, etc.
I’ll be happy to participate to create a place to share the input files and/or the slides and tutorials! Obviously, my slides were in French…
We also did a gradschool with ASPECT last year, where we did some runs with 2-phase flow (the melt example). It was able to run in about 2hours on a student computer with a virtual machine, so not easy but OK enough.
For what to use during class: I think it depends a lot on where the students are running the code, and if you expect them to do projects or not. For things that can only be run during the class, I found the 2D model without refinement OK enough, but clearly I asked them to also run stuff later, for longer times and playing with the parameters. And we had ASPECT directly installed on the Linux machines provided by the university for the duration of the tutorials (we cannot expect
The class in master level will be held this year again, but taught by one of the PhD student because I can’t do it this year.
Hello Anne,
Definitely! Thank you so much. Would it be possible to have copies of input files to reproduce Figure 12 and Figure 17? I understand that you have reported the input values in the paper, but it would be wonderful if we could have the input files for my class.
Some students will use ASPECT for a class project, so they will have more time to explore different models. I like using a virtual machine because students have very different environments.
Yes I would be happy to contribute to make materials available for classes.
the 2018 CIG/CGU that John linked to is the last tutorial I was involved in. But I will teach an Introduction to Geodynamics class next semester (based on the Turcotte & Schuber Geodynamics book) and I plan to make a set of basic example problems for that class (that would all run on a student’s laptop using a VM).
My idea was to make these example .prms part of the main ASPECT repository, because that’s the only way they will be updated whenever something changes in ASPECT. So they would probably become a separate category of cookbooks with documentation and example questions/problems for class. Each of these example cookbooks could also link to lecture slides (that I would probably host on my website) and reference a chapter in the book.
So @mikinakajima, @MarineLasbleis if you want to contribute your class problems/lecture slides, then let’s coordinate on how to set up a structure for that. We’d be really happy to have materials like this, as it would make teaching with ASPECT so much easier.
@mikinakajima I know that doesn’t really help you if you have to teach the class this semester, but I am happy to discuss/help setting up specific example problems because I have to prepare this for my class next semester anyway.
@jdannberg, thank you so much! That would be super helpful. If you have time later this week or next week, it would be fantastic if we could Skype. I’m hoping to set up a subduction model or something similar, but I’m still new to this and I haven’t been quite successful to make an input file using parameters listed on papers. I’ll try to educate myself more on this too.
Thank you!
I just realized that there is a benchmark for the slab detachment.
I am sorry I did not realize this right away - I was looking into cookbooks/benchmarks on my VM, which did not include this input file. I am running VM ASPECT with it now and it is working fine. I will play with this input file in my class probably next week – thank you so much for your input! I will keep you posted.
Dear All,
Thank you so much for the help! I successfully held the second ASPECT workshop.
This time, we covered the slab detachment and the latent heat problem. I think the students enjoyed it as well (a photo attached!). We played with particles (tracers) and the slab size etc as well as also plotted the depth-dependent temperature profile of the latent heat benchmark model. I would be happy to share the homework problem if it helps the community in any way Maybe I can provide feedback from a beginner’s point of view…
Some of the students are going to work on their final projects with ASPECT.
The challenge I can see is that it is not too straightforward to set up an initial condition from a paper table, if you are not familiar with ASPECT. Therefore, the easiest thing to do for us is to start with an input file in the cookbooks/benchmarks and make some changes to it. It would be amazing if the ASPECT GitHub stores (most of) input files used in published papers. This would make it easy for us to reproduce previous studies. In other words, we love the benchmarks/cookbooks and it would be fantastic to have more examples. What do you think? Maybe we could ask the authors to submit their example input files to the ASPECT community? (I have no idea if it is a good idea or not).
I have another specific question. Would it be hard for them to do a 3D viscoelastic subduction model (Section 4, https://www.solid-earth.net/9/267/2018/se-9-267-2018.pdf)? Most of the students have geophysics/geochemistry backgrounds and they are very interested in real-Earth examples. If it would be possible to reproduce @anne-glerum 's work, would it possible for you to send us a copy of the input file?
If not, we totally understand it.
Great to hear the workshop went well! I’m not sure what the right place to share the homework sets is, but it would be fantastic if we can start of a compilation of these kinds of materials. Perhaps a different github repository?
Subduction models - I believe all of the subduction input files from the paper by @anne-glerum et al. are located on this repo:
The input files would need to be added to the current ASPECT structure and to use the visco_plastic material model (very similar structure), but I think all of that should be doable.
Benchmarks/cookbooks - yes, the goal is to add as many as possible! If you have an idea for a new benchmark or cookbook and would like to discuss setting it up, do not hesitate to reach out.
Hi all,
I think collecting more parameter files is a great idea, we just need to find the right place to store them. We currently have:
benchmarks: models that have some reference results (analytic or published) and can be used to verify correctness of ASPECT models
cookbooks: example models with an educational component and a section in the manual.
And additional things we want to collect:
published models: these are hard, because they often rely on additional plugins and data that we can not include in the main repo (it would grow immensely). But I think a lot of studies already publish their input files and data in a github repository or as supplemental information. Maybe we can start a folder with links to those materials?
example .prm files: these are not quite like cookbooks, because they are not specifically teaching something, and maybe the author does not want to write a full page or two for the manual. We already have a few such files in cookbooks/future/ that never made it into real cookbooks. Maybe we can rename that to cookbook/model_gallery/, or just model_gallery/ ? We just need to set up a similar infrastructure as for the existing cookbooks that we test they are still working with the current ASPECT version.
So in short: I would be in favor if someone wants to:
move cookbooks/future to a different folder, and make that a place for collecting models that are good start points for studies of different processes
start a list of published example models (we already collect publications in doc/manual/citing_aspect.bib, but the new list should only contain publications with runnable parameter files, and should be easier to find)
should these be encouraged to be archived somewhere e.g. zenodo, figshare and encouraged to get a separate doi? These would then be easy to link to and discover. In addition, zenodo has added download and citation tracking which could benefit their creators.
teaching resources;
do we want to keep these in the main repo (how much “weight” is this)? Is there a better place? We do have the wiki but is there something better?
I think that the right place for input files and data would actually be the
supplementary material associated with a particular journal article, assuming
journals allow for this. This would be the easiest for a reader to find since
they already know about the paper.
Thank you for the discussion!
If the input files are stored somewhere else (such as on journal websites), having link to these files on the ASPECT publication list may be helpful.
I attached part of my homework as below. The majority of the students who are taking this course is Earth science graduate students. Some students have strong backgrounds in physics and coding, while others do not. The main purpose of this homework is to have students become more familiar with ASPECT. I know this is very basic … but if this can be helpful for the community, I would be happy to contribute to building teaching resources. Also it would be great to have feedback from the community as well. I will also keep you updated about the final projects in class.
Thank you so much!
Slab detachment model
Run the code and plot the slab at t = 6 Myr and t = 12 Myr
Make a movie (gif format) of the slab between t = 0 Myr and 40 Myr.
The original slab width was 80 km. Now, change it to 40 km. Plot the slab at t = 6 Myr and 12 Myr.
The default rheology is non-Newtonian. Instead, use a Newtonian rheology. Describe your rheology model (i.e. explicitly write your viscosity formula providing all the parameters you are using). Plot the slab at t = 40 Myr. Describe the difference of the slab behaviors between (a) and (d). What is causing this difference?
Let’s get back to the original input file setting. Currently, the distance between two tracers is 25 km. Make it 10 km. Keep the edges of the tracer locations [(x,z)=(46km, 58km), (46km, 33 km), (54km, 58km), (54km, 33km)]. Plot the tracer locations (x, z) at time t=6 Myr and 12 Myr.
Miki, thank you for sharing this. And I agree, it’s useful to collect a list of links to published input files. I’ll put that on my ToDo-List. And sorry for coming so late to this, I was traveling and completely missed your post about having a skype call
One of the reasons why published models are not all in the ASPECT repository is that for a publication, the main reason for publishing the file is to make the model reproducible, so it’s the version of the file that worked with the version of ASPECT that was used to generate the results. This means that it may not work with a current ASPECT development version, or may produce different results. For the cookbooks, we make sure they are updated and always work with the development version. So I like @gassmoeller’s idea of a model_gallery, that could then contain updated version of these published models. So if you wanted to run the model setup of a paper that was published a few years ago (and doesn’t need any shared libraries/etc.) you wouldn’t need to update any of the files, only the results may be slightly different.
@bangerth In most cases it doesn’t work well to share input files as supplementary material. I’ve had journals not accept files that ended with .prm, rename the files after I uploded them to supp0001.txt or something like that, unpack them if I uploaded a zipped folder, etc. I would recommend just uploading them as a separate github repo that also includes all the code, and sharing that link in the paper.
