To edit this file or the web pages, please do: svn checkout https://github.com/pencil-code/website/trunk/ www Topics for Pencil code User Meeting in Newcastle 2017: ====================================================== % Notes: "-" means that this item is still open (nothing has happened) % "+" means that it is already in progress (people working on it) % "*" same, but high priority % "/" postponed % "x" means that it has been dealt with % [assigned Persons] feel free to add yourself % # http://pencil-code.nordita.org/UserMeetings/2017/ - Pencil Code Steering Committee (report & elections) Approval of ToR - Updating various data bases: pencil-code/license/developers.txt ~/public_html/pencil-code/PCSC/committers/list.tex ~/pencil-code/doc/citations/notes.pdf http://pencil-code.nordita.org/doc/citations.pdf + pencil GPU/CUDA, astaroth: implementation for basic hydro problem (samples/gputest) almost complete. Plans for future convergence of astaroth and pencil/gpu code. [Matthias, Johannes] + particles' modules: new 'particles_inertial' (or 'particles_lagrangian') module, without swarms, to be created from particles_dust. most particles_* modules to be moved to a subdirectory src/particles (cf. src/magnetic). Also some streamlining of the particles code to be implemented, to reduce the amount of interpolation and allocation/deallocation. [Nils, Chao-Chin, Dhruba, Andrew] o - also re: particles: some 'direct' unformatted writes from within particles modules should be replaced by calls to routines in the i/o modules, to facilitate the implementation of the HDF5 file i/o option for particles. + 'decoupling sample from website': Create new 'highlights'/'production_runs' directory in repository, for 'serious' simulations (cf. the smaller problems considered in samples). Also expand highlights on website with additional examples (chase up individuals currently with unexpanded listings). Adopt *.md format for README files on github. [Many?] - python/pencilnew: these (more object-orientated) python scripts now functional for most purposes; users encouraged to use them. More functionalities will be ported soon. Sim objects largely functional. - HDF5: currently implemented, beyond minor issue with persistent variables used by some modules (particularly any processor-dependent variables: e.g. random seed?). And some particles files not yet dealt with: needs work on particles modules (some 'direct' writes to unformatted fortran output should be replaced by calls to routines in the i/o modules.) - git/svn. Given problems with svn-bridge on github, bitbucket was discussed as an alternative. They provide free hosting for academic users, and have a facility for mirrored git and svn repositories (using SubGit). To be tested with the website repository. If ultimately adopted, github repository would be maintained as a read-only mirror. x code stability/timestep control: additional dt-control functionality added for momentum equation, via lcdt_tauf. [Fred] - pencil-code boundary conditions: one-sided b.c.s implemented, avoiding overspecification of b.c.s by current 'a'/'s' options. But currently prone to instability -- Joergen has some experience of this? (Matthias, Joergen) - reorganising code to allow multiple special modules: in principle possible, but more concrete applications where this is needed are requested, before proceeding. (And before this is done, there is no need to link to special modules dynamically.) + auto-test: stop using g95 compiler, start using pgi as an additional option? (On norlx51 or scylla?) To be discussed further with Wolfgang et al. (NB: current gfortran compilers have more debugging options than currently being used.) [Axel, Boris, Wolfgang?] + PC paper: the idea of a special issue was revived. Likely destination GAFD, after discussion with editor (Andrew Soward). (Physica Scripta, SIAM J Sci Comp, Comp Phys Commun, & JPP were earlier discussed). Likely guest-editors: Axel, Fred, Simon. A revised list of likely submissions to be chased up; likely deadline Feb 2018? [Axel, Simon, Fred] - chase up alternative ways to track pencil-code-using publications (cf. ADS bumblebee); any possibilities on Web of Science? - chase up whitelisting of norlx51b, for people currently being blacklisted [Graeme, w.r.t. Newcastle] - documentation: some (many?) modules not currently documented in manual. + registering: do all registering from register_* subroutines: i.e. do all put_shared_variable calls from register_*; and do all get_shared_variable calls from initialize_*. Boris concerned this is not so simple (entropy modules very messy)? [Matthias, Boris] - code style: it is proposed that we avoid using labelled-if structures shorter than, say, 20 lines of code. - yin-yang grid: progress currently stalled due to interpolation issues at grid boundaries. - non-equidistant grid: upwinding still not currently possible, due to glitch in deriv_alt. - other topics (not discussed in detail at this meeting): - reorganising the code to separate 'engine' from 'toolbox' (involving moving more code into special modules?) - reducing size of compiled src directory. (dynamical linking?) - f2003 - future PC user meetings: probably Boulder for 2018; a doodle-poll will be arranged for dates. (Future possibilities discussed include Aalto and Goethenburg (t.b.c.), possibly also Ventura, Zhejiang?) Also considering the possibility of running alongside an associated meeting, for a wider science community. (Axel)