[Users] Meeting minutes for 2020-03-12
babiuc at marshall.edu
Thu Mar 12 12:09:52 CDT 2020
Present: Helvi, Maria, Bill, Roland, Alois Peter Schaffarczyk, Zach, Peter, Christopher Evans, Steven, Bill Gabella, Beyhan, Liu Haoyang,
* Zach opens the meeting by requesting volunteers for call chairs and minute takers.
* After a brief discussion, the call chairs and minute takers for the next two weeks are decided, and Zach updates the Rota on wiki.
* Roland talks about the opportunity to submit a SPEC CPU proposal, Steve checks the deadline is, and consats that work at the proposal could begin Jan. next year.
* Peter urges for work to begin as early as possible, for review purposes
* Zack reports on Baikal & Baikal Vacuum: Helvi compiled successfully, and he has a student working in improving the c-code kernel of NRPy, for faster performance, and same results
* Helvi, as the reviewer for BaikalETK, asks for clarification on what part should review - will contact Zach directly by email
* Zach emphasizes that the update has no effect on the jupyter notebooks or thorn generation part
* Roland explains that when he reviews NRPy, he focuses on the generated notebooks, similarly on how it was done when reviewing Crank, it was never checked Crank itself, but the output.
* Steve points out that it makes no sense of reviewing binary codes
* Zach talks about the new updates at IllinoisGRMHD, which was completely rewritten as a collection of NRPy self-contained modules, that can generate the whole code by running the notebooks.
* Roland, who reviews IllinoisGRMHD, reiterates his method of reviewing and will get in touch with student that does the work
* Steven brings up the problem of missing documentation of how people code headers and macros
* Roland indicates that it should be contained in the User Guide, on the chapter How to write a thorn
* Roland reports on PreSync, Roland summarizes the discussion about code review, and reiterates that all three codes are in proper review mode
* Zach thanks for updates, and ask for timeline
* Roland confirms that the timeline did not change: freeze before APS april meeting, and major bugs are fixed before release. He reports on a weird bug in 2006 make, manifesting as an infinite loop, with no way to avoid it, unless the code is compiled serially.
* Zach’s suggestion that if make is like that, then add a note to people still using that version to please compile serially, is well received by Roland who agrees, with the alternative to encourage users to update make
* Zach goes on to request for a name for the new release and asks if any name should be taken out.
* A discussion about Heisenberg follows, and Alois explains that it is still perfectly appropriate, the only thing is that he did not do relativity. He suggests Hilbert, but Roland points out that it was already used.
* Helvi proposes Turing, and Roland comments that he kept this name for a special release.
* Alois proposes Goedel, and Steven likes this proposal.
* Zach proposes Levi-Civita, Maria suggests Marcel Grossman and Riemann, and Alois agrees that Marcel Grossman would be a good pick.
* Zach calls to vote, and Steven proposes that Roland, as release coordinator, to propose tow names.
* Roland proposes Turing and Riemann. Turing receives 4 votes, Riemann 2, and it is decided that Turing is the name of the next release.
* Zach goes on at the next item on the agenda, namely at how to handle incorrect Read/Write statements after this release.
* A debate follows between Steven and Roland, and it is decided to keep it as a warning now, but to change it to error next time, however to announce the upcoming change in this release
* Roland asks that if anything else need to be deprecated, please add at: https://docs.einsteintoolkit.org/et-docs/Deprecated_features
* Zach requests a report on minor or trivial changes
* Steven reports that malloc leaves holes in memory, and that seem to be inherent to platforms like Cactus, and there is nothing to do. The memory management might be fixed with Carpet.
* Zach asks Steve to volunteer to put a fix to the ticket, and raises the issue with the gallery testing data files not be able to be downloaded
* Roland explains that the data files need to be regenerated, especially for core-collapse, which needs proprietary code, and he offers to do it.
* Zach raises the problem of one test that failed in Hydro Excision
* Roland explains that the test is sensitive to round-off, it is hard to tweak, because it is excising a spherical region outside a feature, and it should remain open and push it to the next release
* Zach suggests increasing the tolerance, or shorten the test and increase the resolution
* Peter points out that the difference is 5 order of magnitude, and there is a shock that has to be resolves
* Roland concludes that the best approach is to decrease the region and increase the resolution, but it needs to evolve long enough, because is tests shock
* Zach reviews important roadblocks: PreSync fails to parse, infinite make loop.
* Steve fixed PreSync, and it needs to be tested
* Roland found a solution that might not work for Fortran codes
* Steven announces that the webpage for workshop https://www.cct.lsu.edu/Einsteintoolkitworkshop has Einstein in festive Mardi Gras attire
* Zach requests for student volunteers for gallery tests
* Roland mentions an unanswered question on Collapse boson stars
* Zach reports that he is taking care of the other question on calculating final spin for remnant, and thanks the ones who gave helpful suggestions
* Zach closes the meeting
Maria C. Babiuc Hamilton, Ph.D.
Professor, Department of Physics
College of Science, Marshall University,
1 John Marshall Drive, Huntington, WV, 25755
Room S 257, Phone: (304)696-2754
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Users