LargeScale Bayesian DataAnalysis forNuclear Fusion Experiments

Thereiscurrently aparadigm shift taking place in current tomography, toinfer theinternal magnetic field that thefield ofscientific methodology. Methods fortheextraction of provides thecommoncoordinate system forthemappingof underlying physicsfrom observations and the other diagnostic information. Thismodelcanthenbefully falsification/co nfirmation ofscientific hypothesis areundergoingintegrated withindividual Bayesian models ofnon-magnetic asignificant change through theuseofageneric approach to diagnostics tomakeoptimal inference onfurther physics inference fromobservations: socalled 'Bayesian' Probability tt Theory. Thefirst partofthis paperwilloutline andexemplifyquanT ies. howthismethodischanging dataanalysis innuclear fusion: iThe combination of a large number ofdianostic How alluncertainties (systematic, statistical andmodel instruments fora jointBayesianinversion leadsto uncertainties) canbetreated inaunified way,andhowdata challenging architectural problems. A large nuclear fusion analysis methods canbeunderstood andunified through experiment hasontheorderof100diagnostic instruments probability theory. Thepractical advantage herefornuclear andatleast asmanycomputer codes formodelling different fusion experiments isthepossibility toutilise this method fora aspects oftheplasma. Eachdiagnostic isonitsownafairly morecomprehensive understanding oftheinternal state of complex piece ofequipment withassociated statistical and fusion plasmas asinferred frommeasurements frommultiplesystematic uncertainties, viewinggeometry, calibration heterogeneous diagnostics. Thesecondpartofthepaperwill procedures andassociated physics models. Thesecond part of discuss architectural issues relating totheverycomplex analysis systems thatmight emergefromasystematic application OfthilSthipae del ihteacietua rbesta rs method inlarge scientific experiments, fromaconsistent application ofthis method ofjoint Bayesian analysis infusion experiments. There will betwoparts tothe