Advancing software for ecological forecasting
1 like443 views
This document summarizes an upcoming workshop titled "Advancing Software for Ecological Forecasting" which will bring together scientists and developers working on improving model-data fusion, prediction workflows (scientific workflows), and facilitating inter-operability of different statistical engines (inference engines). The workshop aims to enable collaboration and development of improved software. The 3 day agenda includes plenary talks on the first day to establish goals. The second day will include keynote talks and workshops to identify opportunities for existing tools to interface and key gaps. The third day will focus on organizing outputs and conclusions from the workshop. Intended outcomes include working papers, funding proposals, community resources, and new networks.
Advancing software for ecological forecasting
- 1. Advancing Software for Ecological Forecasting National Center for Supercomputing Applications Urbana, IL March 25-27, 2014 conferences.igb.illinois.edu/ecological_forecasting Description Two distinct but overlapping challenges for advancing software for ecological forecasting are: improving model-data fusion and prediction workflows (scientific workflows) and facilitating inter-operability of different statistical engines (inference engines). This workshop aims to bring together scientists and developers working in these different areas to facilitate the development of improved software for ecological forecasting through enabling due-diligence, syntheses, generalization and collaboration. Each group will benefit from sharing ideas and experiences within their respective domains, but also from collaborations among domains. We expect developers of scientific workflows to benefit from knowing the latest inference methods, algorithms and outputs and inference engine developers to benefit from knowing how their engines might interoperate with the different data-types and analytical methods used in scientific workflows. Intended Outcomes 1) Working papers on the current state and future areas for improvement of software tools, 2) Initiate future collaborative research funding proposals, 3) Community websites featuring available and desired resources, and 4) new organizational networks. Agenda DAY 1: Plenary talks (1 hour total) by organizers stating scope, goals and intended outcomes. Aim will be to establish the key areas of environmental science and applications where we critically need to improve workflows and engines. Project overviews: (4 min each). Representative for each of the workshop invitees to introduce their tools and interests, interspersed with coffee breaks, concluding at lunch. Focus on understanding state of the art and what each group brings to the discussion. Afternoon will begin with a brief talk by Matt Jones on challenges and opportunities, followed by parallel track working groups focusing on 1) scientific workflows and 2) inference engines. At the conclusion of the day we will identify topics to focus on during the following day. DAY 2: Aaron Ellison will give a keynote talk. Then, workshops discussions organized around topics determined by the group on the previous day. The approach will be to cover opportunities for overlapping and interfacing existing tools, identifying key gaps in what we can currently do and any key opportunities and benefits to adopting common or compatible standards and protocols. The day will finish with sharing conclusions amongst the entire group and identifying areas of overlap or other topics needing critical discussion the next day. Focus of the day will be identifying new and important ways to develop scientific workflows and inference engines into the future. DAY 3: First 2 hours: groups will be shuffled to allow fresh ideas / criticism / contributions to the conclusions from Day 2. Either side of lunch dedicated to organizing and writing workshop outputs. Final session to bring together entire group to summarize workshop outputs. Focus of the day will be on developing tangible useful outputs from the workshop. Organizers: David LeBauer, Chair (Univ. Illinois) Perry de Valpine (Univ. California, Berkeley) Matthew Smith (Microsoft Research) Participating Projects: Scientific Workflows: ARES DART EcoPAD ESMF Kepler PEcAn Science Pipes VisTrials EcoPath and EcoSim Inference Engines OpenBUGS Distribution Modeller JAGS INLA STAN PyMC ADMB Filzbach POMP R-INLA Funding: DOE (RCN Forecast) Institute for Genomic Biology Microsoft Research Connections