Presentation is loading. Please wait.

Presentation is loading. Please wait.

Tuning GENIE Earth System Model Components using a Grid Enabled Data Management System Andrew Price, Gang Xue, Andrew Yool, Dan Lunt, Tim Lenton, Jasmin.

Similar presentations


Presentation on theme: "Tuning GENIE Earth System Model Components using a Grid Enabled Data Management System Andrew Price, Gang Xue, Andrew Yool, Dan Lunt, Tim Lenton, Jasmin."— Presentation transcript:

1 Tuning GENIE Earth System Model Components using a Grid Enabled Data Management System Andrew Price, Gang Xue, Andrew Yool, Dan Lunt, Tim Lenton, Jasmin Wason, Graeme Pound, Simon Cox and the GENIE team. http://www.genie.ac.uk/ UK e-Science – All Hands Meeting 3 rd September 2004

2 03/09/04UK e-Science - All Hands Meeting, Nottingham, 20042/22 Outline Introduction Scientific aims of GENIE e-Science tools –Data Management System –Geodise Toolboxes –OPTIONS Design Search and Optimisation Results Future work Conclusions

3 03/09/04UK e-Science - All Hands Meeting, Nottingham, 20043/22 Introduction The GENIE project is developing a Grid-based system to: Flexibly couple together state-of- the-art components to form a unified Earth system model Execute the resulting model on the Grid Share the distributed data produced in simulations Provide high-level open access to the system, creating and supporting virtual organisations of Earth system modellers

4 03/09/04UK e-Science - All Hands Meeting, Nottingham, 20044/22 Scientific Aims ice-age Orbital parameters affect incident radiation and climate Biological and geological processes interact with, and feedback upon, the climate (via, for instance, CO 2 )

5 03/09/04UK e-Science - All Hands Meeting, Nottingham, 20045/22 The target GENIE Model 3D atmosphere 3D ocean 2D sea ice Atmospheric CO 2 2D land surface Land vegetation Ocean biogeochemistry Ocean sediments 3D ice sheets Atmosphere – Bristols IGCM3 Ocean – SOCs GOLDSTEIN Land – Met. Offices TRIFFID Land ice – Bristols GLIMMER Ocean biogeochemistry and sediments – UEAs BioGEM

6 03/09/04UK e-Science - All Hands Meeting, Nottingham, 20046/22 Initial GENIE experiments Initial studies in GENIE performed parameter sweeps to investigate the properties of the model

7 03/09/04UK e-Science - All Hands Meeting, Nottingham, 20047/22 e-Science Tools Data Management System (augmented version of the Geodise Database System) Matlab scripting environment Geodise Toolboxes XML Toolbox OPTIONS Design Search and Optimisation package Template and Example scripts

8 03/09/04UK e-Science - All Hands Meeting, Nottingham, 20048/22 Data Management System Java Client Code Apache Axis CoG Jython Functions Globus Server Geodise Database Toolbox Metadata Database ClientGrid SOAP Matlab Functions Database Web Services Authorisation Service Location Service Metadata Archive & Query Services Jython XML Schema GridFTP Portal

9 03/09/04UK e-Science - All Hands Meeting, Nottingham, 20049/22 Grid Computation National Grid Service (GT2) OxfordLeeds RALManchester Jython Local Resources (GT2) Java CoG Imperial Condor PoolSouthampton Condor Pool Flocked Condor Pools

10 03/09/04UK e-Science - All Hands Meeting, Nottingham, 200410/22 Geodise Toolboxes Geodise Compute Toolbox gd_createproxy.m Creates a Globus proxy certificate for the user's credentials gd_destroyproxy.m Destroys the local copy of the user's Globus proxy certificate gd_jobsubmit.m Submits a compute job to a Globus GRAM job manager gd_jobstatus.m Gets the status of a Globus GRAM job gd_putfile.m Puts a remote file using GridFTP gd_getfile.m Retrieves a remote file using GridFTP gd_rmfile.m Deletes a remote file using GridFTP gd_makedir.m Creates a remote directory using GridFTP gd_rmdir.m Deletes a remote directory using GridFTP Geodise Database Toolbox gd_archive.m Archives a file or data structure to the database gd_query.m Query the database for data matching specified criteria. gd_retrieve.m Retrieves a file or data structure from the database

11 03/09/04UK e-Science - All Hands Meeting, Nottingham, 200411/22 Scripting a Tuning Study GENIE Database Grid Resource MATLAB function RMS_Error = cgoldstein(params) config file results file CG binary return RMS_Error optimum = fminsearch( … @cgoldstein, params, … ) gd_query(results) gd_putfile(CG binary) gd_putfile(config file) gd_jobsubmit(RSL) gd_getfile(results file) gd_archive(results)

12 03/09/04UK e-Science - All Hands Meeting, Nottingham, 200412/22 Matlab Optimisation Toolbox % ************************ % Specify a starting point % ************************ parameters = [ 0.5 ]; % ************************ % Perform the minimisation % ************************ optimum = fminsearch( @cgoldstein_1D, parameters, optimisation_parameters ) % ************************ % Specify a starting point % ************************ parameters = [ 420 5000000 ]; % ************************ % Perform the minimisation % ************************ optimum = fminsearch( @cgoldstein_2D, parameters, optimisation_parameters )

13 03/09/04UK e-Science - All Hands Meeting, Nottingham, 200413/22 OPTIONS Matlab interface to the Options design exploration system –http://www.soton.ac.uk/~ajk/options/welcome.html State of the art design search and optimisation algorithms –Design of Experiment methods –Response Surface Modelling –Over 30 search methods including: Adaptive Random Search (ADRANS), Powell's Direct Search (PDS), Simplex Method (SIMP), Genetic Algorithm (GA), Simulated Annealing (SA), Evolutionary Programming (EP)

14 03/09/04UK e-Science - All Hands Meeting, Nottingham, 200414/22 Grid Computation OptionsMatlab National Grid Service (GT2) OxfordLeeds RALManchester Local Resource (GT2) GENIE Database objfun.mobjfun_parse.m optjobparallel.m

15 03/09/04UK e-Science - All Hands Meeting, Nottingham, 200415/22 OptionsMatlab >> OptionsInput = createOptionsStructure(4.0) DNULL: -777 OLEVEL: 2 MAXJOBS: 100 NVRS: 12 VNAM: {'SCLTAU' 'INVDRAG' 'OCNHORZDF'... } LVARS: [1.3000 2.0000 2500... ] UVARS: [2.1000 4.8000 5700... ] VARS: [1.7000 3.4000 4100... ] ONAM: 'RMSERROR' OMETHD: 4.0000 DIRCTN: -1 NITERS: 1000 OPTFUN: 'cgoldstein_12D' OPTJOB: 'optjobparallel' GA_NPOP: 100 >> OptionsOutput = OptionsMatlab(OptionsInput); Available Optimisation Methods: 1.1 for OPTIVAR routine ADRANS 1.2 for OPTIVAR routine DAVID 1.3 for OPTIVAR routine FLETCH 1.4 for OPTIVAR routine JO 1.5 for OPTIVAR routine PDS 1.6 for OPTIVAR routine SEEK 1.7 for OPTIVAR routine SIMPLX 1.8 for OPTIVAR routine APPROX 1.9 for OPTIVAR routine RANDOM 2.1 for user specified routine OPTUM1 2.2 for user specified routine OPTUM2 2.3 for NAG routine E04UCF 2.4 for bit climbing 2.5 for dynamic hill climbing 2.6 for population based incremental learning 2.7 for numerical recipes routines 2.8 for design of experiment based routines 3.11 for Schwefel library Fibonacci search 3.12 for Schwefel library Golden section search 3.13 for Schwefel library Lagrange interval search 3.2 for Schwefel library Hooke and Jeeves search 3.3 for Schwefel library Rosenbrock search 3.41 for Schwefel library DSCG search 3.42 for Schwefel library DSCP search 3.5 for Schwefel library Powell search 3.6 for Schwefel library DFPS search 3.7 for Schwefel library Simplexsearch 3.8 for Schwefel library Complexsearch 3.91 for Schwefel library two­membered evolution strategy 3.92 for Schwefel library multi­membered evolution strategy 4 for genetic algorithm search 5 for simulated annealing 6 for evolutionary programming 7 for evolution strategy

16 03/09/04UK e-Science - All Hands Meeting, Nottingham, 200416/22 Twin-Test Experiment Attempt to recover a known state of the model using a Genetic Algorithm. Performed 10 generations of a 100 member population. Then applied a local Simplex search of the best candidate. Population too small to find optimal solution – suitable for finding local minima

17 03/09/04UK e-Science - All Hands Meeting, Nottingham, 200417/22 Tuning using Observational Data Model Sea Surface Temperatures NCEP Sea Surface Temperatures Model Air Temperatures NCEP Air Temperatures Apply the same method but calculate the RMS error statistic by comparing the model state with NCEP observational data. The lack of a land surface in the model means tuning cannot match the observational data.

18 03/09/04UK e-Science - All Hands Meeting, Nottingham, 200418/22 IGCM3 Atmosphere Model The objective function is a weighted sum of the RMS differences between the model state and NCEP data. Winter and Summer averages for a number model fields.

19 03/09/04UK e-Science - All Hands Meeting, Nottingham, 200419/22 IGCM Results 25% reduction in error statistic compared to default parameters Similar result to a parallel study performed using the Ensemble Kalman Filter Model physics insufficient to perfectly match observational data.

20 03/09/04UK e-Science - All Hands Meeting, Nottingham, 200420/22 e-Science Summary

21 03/09/04UK e-Science - All Hands Meeting, Nottingham, 200421/22 Conclusions Provided the environmental scientist with a toolset for tuning GENIE models: –Scripting environment –Database repository –Computational Grid interface –Suite of generic optimisation algorithms A Global minimum can reliably be found in low dimensional problem space. For higher dimensional problems, the tools are appropriate for locating local minima in the state space.

22 03/09/04UK e-Science - All Hands Meeting, Nottingham, 200422/22 The GENIE Team Coordinator: Tim Lenton– CEH Edinburgh Principal investigator: Paul Valdes – Bristol Research Team and Collaborators: James Annan – FRSGC, Japan Chris Brockwell – UEA Norwich David Cameron – CEH Edinburgh Peter Cox – Hadley Centre (UKMO) Neil Edwards – Bern, Switzerland Murtaza Gulamali – London e-Science Centre Julia Hargreaves – FRSGC, Japan Phil Harris – CEH Wallingford Dan Lunt – Bristol Bob Marsh – SOC Andrew Price – Southampton e-Science Centre Andy Ridgwell – UBC, Canada Ian Rutt – Bristol Gang Xue – Southampton e-Science Centre Andrew Yool – SOC Management Team: Melvin Cannell – CEH Edinburgh Trevor Cooper-Chadwick – Southampton e-Sci. Centre Simon Cox – Southampton e-Sci. Centre John Darlington – London e-Science Centre Richard Harding – CEH Wallingford Tony Payne – Bristol John Shepherd – SOC Andrew Watson – UEA Norwich Thanks to Steven Newhouse


Download ppt "Tuning GENIE Earth System Model Components using a Grid Enabled Data Management System Andrew Price, Gang Xue, Andrew Yool, Dan Lunt, Tim Lenton, Jasmin."

Similar presentations


Ads by Google