coulomb excitation data analysis codes; gosia 2007 - Physics and ...

5.6 OP,ERRO (ERRORS)ThemoduleofGOSIAactivatedbyOP,ERROisdesigned primarily for estimating the error bars to beassigned to the set of matrix elements corresponding to the minimum value of χ 2 . However, this option alsocan be helpful in checking the existence of better solutions by providing a relatively fast way of scanning theχ 2 hypersurface. Error estimation is performed in two separate stages. First, the “diagonal“,or uncorrelatederrors are found, next, the diagonal errors are used in the estimate of the “overall“, or correlated errors.This procedure is described in detail in Section 4.6. Section 7.4.10 shows the correct location of OP,ERROin the input stream.The input to OP,ERRO is as follows:OP,ERROIDF,MS,MEND,IREP,IFC,RMAXwhere:IDFestimation.Mode flag. IDF =0sets diagonal error calculation mode, IDF =1causes overall errorMS,MEND The range of matrix elements indices for which error estimation is to be performed,i.e. the calculation will be carried out for matrix elements with indices fulfilling MS ≤ INDEX ≤MEND.MS can also be entered as 0 or −1. MS =0implies that the calculation will be performedfor all matrix elements (excluding fixed ones - see below), thus providing a short form of specifying thefull range. MS = −1 can be used only for “overall“ error calculation (IDF =1). This allows the userto select an arbitrary set of the matrix elements defined by additional input required only if MS = −1.In this case the input to OP, ERRO is as follows:OP, ERRO1, −1, 0, IREP, IF C, RMAXMS 1 ,MEND 1··MS n ,MEND n0, 0where MS i and MEND i define the range of matrix elements indices similarly to MS and MENDwhen both are positive. MS i = MEND i selects a single matrix element. Two zeros terminate thisportion of input. This feature allows repeating the overall error estimation for a subset of matrixelements or to include the matrix elements that previously have been skipped without modifying theME setup. A given value of MEND is redundant if MS =0or MS = −1 has been selected.IREPRepetition flag, assuming values of 0, 1 or 2. IREP =0implies a new calculation, i.e.no previously stored errors are read in and the error file will be created. Obviously, IREP =0shouldbe used in conjunction with IDF =0for a first calculation of the diagonal errors. IREP =1causespreviously stored errors to be read in and used for the continuation of the error estimation. The errorsare stored on file TAPE15, which is updated during each execution of OP,ERRO. IREP =2meansthat the sum-rules file, TAPE3, has already been created during a previous “overall“ errors calculationand causes the code to read it in and update it during the current run. The CONT switch SMR,should be set for both creation and update of TAPE3. Note that in this case the experimental yieldsmust reside on TAPE4 to avoid a multiple definition of the input/output files. TAPE3 is created withIREP=1 and updated with IREP =2only if SMR, was specified. IREP =2reduces to IREP =1if the SMR, switch was not encountered. Both TAPE15 and TAPE3 are required by the sum-rulescode SIGMA - see Section 6.2.2. The proper combination of MS,MEND and IREP makes it possible tosplit the time-consuming error estimation into several runs which is important when running GOSIAon heavily loaded or unreliable computers. Note that in OP,YIEL NTAP must equal 4.71