Louis Moresi bio photo

Louis Moresi

Professor of Geophysics University of Melbourne Victoria 3010 Australia +61 4 0333 1413

Email Twitter Facebook Google+ Github Youtube Researchgate

Summary of lessons learned in developing Underworld and building an Underworld community

I gave a presentations at the Fall AGU meeting to the computational geodynamics community and to the GeoPrisms meeting to provide a summary of lessons learned in building Underworld, supporting the user community, and our plans for the future. This is a good time to take a snapshot of the code because the timing is the end of the NeCTAR investment into making Underworld integrate with geological workflow tools (such as GPlates, GoCAD, Litmod3D, LeCode). This presentation includes many contributions from users of those tools and I have given contact details for the people responsible. The slides are here: Talks/AGU-SF-2013/Underworld-AncoraImparo.pdf.

I gave a modified version of this talk as a starter for discussion at the CIG long-term tectonics / Earthscope workshop in February 2014. The slides for this talk are available from CIG, but are essentially the same as the AGU slides above.

At the CIG workshop in Banff, Canada, May 2014, I gave a talk to reflect on 20 years of the CITCOM family of codes. Dave May asked me to include a look-forward of 20 years as well as all the “back in my day, sonny”, reminiscing. I tried to take this seriously. My argument is this: if you want to know what the landscape is going to look at in 5 years, just be more realistic about how long it is actually going to take to get all the things working that you are busy coding today. But 20 years ? I’m arguing for us to be more vigilant about carrying uncertainty information, make a big effort to treat “a model” as meaning a whole collection of models plus their fitness-for-purpose for a given application(i.e. in some sense, having model derivatives). I believe this will give us the possibility to calculate metrics in model space and build comparisons and that this will allow us to put models in categories and develop searchable libraries. That could take some time even if it is not 20 years: Talks/CIG-Banff-2014/CIG-CitcomBirthdayParty.pdf