BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20210916T132456Z
LOCATION:
DTSTART;TZID=Europe/Stockholm:20210706T173000
DTEND;TZID=Europe/Stockholm:20210706T190000
UID:submissions.pasc-conference.org_PASC21_sess182_post111@linklings.com
SUMMARY:P04 - Tasmania: Towards a Python-Based Approach to Atmospheric Mod
 eling
DESCRIPTION:Poster\n\nP04 - Tasmania: Towards a Python-Based Approach to A
 tmospheric Modeling\n\nUbbiali, Bianco, Gonzalez Paredes, Groner, Sawyer..
 .\n\nWe present the Tasmania framework to ease the development of atmosphe
 ric models in Python. Tasmania features a component-based architecture, wh
 ere each component represents either the dynamical core or a physical para
 meterization. The library provides different couplers which mold the compo
 nents into a model with the desired level of complexity. Each coupler purs
 ues a well-defined physics-dynamics coupling algorithm; six coupling schem
 es are currently supported. Within a model the components act on and commu
 nicate through the "state", i.e. the set of variables identifying the conf
 iguration of the physical system at any point in time. As a proof-of-princ
 iples, two concrete models have been implemented on top of Tasmania: the t
 wo-dimensional viscid Burgers' equations and the three-dimensional isentro
 pic model. Both models are settled on logically Cartesian grids and discre
 tized via finite differences. To work around the intrinsic slowness of the
  Python interpreter, all mathematical operators are encoded using the GT4P
 y library, which generates high performance implementations of stencil ker
 nels starting from a high-level definition. GT4Py harnesses the GridTools 
 (GT) framework to optimize the code for a given architecture at compile-ti
 me. We show that the GT backends designed for multi- and many-core process
 ors and GPUs deliver a significant performance improvement over a Numpy-ba
 sed implementation.
END:VEVENT
END:VCALENDAR