($Id: cc523e423547ec1e335b86a2ead55e4b2e6b3040 $) Log-file Millepede II-P Tue Nov 22 10:20:58 2022 Text and data files: 1 text mp2con.txt 2 Fortran_binary mp2tst.bin LOOP1: starting LOOP1: 20 labels from txt data stored LOOP1: reading data files LOOP1: 700 is total number NTGB of labels/parameters LOOP1: 0 is number of pre-sigmas LOOP1: 683 is number NVGB of variable parameters LOOP1: 700 is total number NTPGRP of label/parameter groups NREC = 10000 = number of records MREQENF = 25 = required number of entries (eqns in binary files) MREQENA = 10 = required number of entries (from accepted fits) LOOP1: ending LOOP2: starting LOOP2: start event reading LOOP2: event reading ended - end of data NTGB = 700 = total number of parameters (all parameters, appearing in binary files) NVGB = 683 = number of variable parameters (appearing in fit matrix/vectors) NAGB = 683 = number of all parameters (including Lagrange multiplier or reduced) NTPGRP = 700 = total number of parameter groups NVPGRP = 683 = number of variable parameter groups NFGB = 679 = number of fit parameters NOFF = 232903 = max number of off-diagonal elements NCGB = 4 = number of constraints NAGBN = 18 = max number of global parameters in an event NALCN = 28 = max number of local parameters in an event NAEQN = 38 = max number of equations in an event NCACHE = 25000000 = number of words for caching cache splitting 56.7 % 3.0 % 40.2 % Solution method and matrix-storage mode: METSOL = 1: matrix inversion with 3 iterations MATSTO = 1: full symmetric matrix, (n*n+n)/2 elements Convergence assumed, if expected dF < 0.1000E-02 Constraints handled by elimination Rank of product matrix of constraints is 4 for 4 constraint equations Cut values of Chi^2/Ndf and Chi2, corresponding to 2 and 3 standard deviations Ndf Chi^2/Ndf(2) Chi^2(2) Chi^2/Ndf(3) Chi^2(3) 1 4.000 4.0 9.000 9.0 2 3.090 6.2 5.915 11.8 3 2.675 8.0 4.718 14.2 4 2.429 9.7 4.063 16.3 5 2.263 11.3 3.641 18.2 6 2.141 12.8 3.344 20.1 7 2.048 14.3 3.121 21.8 8 1.974 15.8 2.947 23.6 9 1.912 17.2 2.806 25.3 10 1.861 18.6 2.690 26.9 12 1.779 21.3 2.508 30.1 14 1.716 24.0 2.371 33.2 16 1.666 26.7 2.263 36.2 18 1.625 29.2 2.176 39.2 20 1.590 31.8 2.104 42.1 25 1.523 38.1 1.967 49.2 30 1.474 44.2 1.868 56.0 35 1.428 50.0 1.757 61.5 40 1.399 56.0 1.702 68.1 LOOP2: ending Solution algorithm: ======================================== solution method: matrix inversion convergence limit at Delta F= 0.10E-02 maximum number of iterations= 3 using pre-sigmas: no regularization: no Checking feasibility of parameters: parameters are feasible (i.e. satisfy constraints) === local fits have bordered band matrix structure === NBNDR = 10000 = number of records (upper/left border) NBDRX = 0 = max border size NBNDX = 4 = max band width No rank defect of the symmetric matrix it fc fcn_value dfcn_exp slpr costh iit st ls step cutf rejects hhmmss FMS -- -- ----------- -------- ---- ----- --- -- -- ----- ---- ------- ------ --- 0 1 0.19854E+07 0.0 0 00000 FMS PTLINE: line-search method based on slopes with sufficient slope-decrease PTLDEF: slope ratio limit= 0.50000000000000000 PTLDEF: maximum step = 10.000000000000000 PTLDEF: 1 <= nr of calls <= 10 i x F(x) F'(X) minimum F'(X) 0 0.000000 L 0.00000 -0.188279E+07 0.000000 ratio 1 1.000000 *R -941397. 0.139983E-07 1.000000 0.000 PTLINE: INFO=1 convergence reached 1 2 0.10259E+06 0.19E+07-0.000 0.657 0 0 1 1.000 0.0 0 00000 F 00000 S PTLINE: line-search method based on slopes with sufficient slope-decrease PTLDEF: slope ratio limit= 0.50000000000000000 PTLDEF: maximum step = 10.000000000000000 PTLDEF: 1 <= nr of calls <= 10 i x F(x) F'(X) minimum F'(X) 0 0.000000 L 0.00000 -0.188279E+07 0.000000 ratio 1 1.000000 *R -941397. 0.139983E-07 1.000000 0.000 PTLINE: INFO=1 convergence reached 2 3 0.10259E+06 0.28E-14 0 0 0.0 0 00000 F Sum(Chi^2)/Sum(Ndf) = 102589.35168164221 / ( 99992 - 679 ) = 1.0329901592101960 In total 2 + 3 loops through the data files Times [in sec] for text processing 0.039 LOOP1 0.087 LOOP2 0.067 func. value 0.130 * 2. func. value, global matrix, solution 0.260 * 1. new solution 0.001 * 1. Total time = 1 seconds = 0 h 0 m 1 seconds end Tue Nov 22 10:20:58 2022 Peak dynamic memory allocation: 0.102192 GB Explanation of iteration table ============================== it = iteration number. Global parameters are improved for it > 0. First function evaluation is called iteraton 0. fc = number of function evaluations. fcn_value = value of 2 x Likelihood function (LF). The final value is the chi^2 value of the fit and should be about equal to the NDF (see below). dfcn_exp = expected reduction of the value of the Likelihood function (LF) slpr = ratio of the actual slope to inital slope. costh = cosine of angle between search direction and -gradient ls = line search info < 0 recalculate function = 0: N or STP lt 0 or step not descending = 1: Linesearch convergence conditions reached = 2: interval of uncertainty at lower limit = 3: max nr of line search calls reached = 4: step at the lower bound = 5: step at the upper bound = 6: rounding error limitation step = the factor for the Newton step during the line search. Usually a value of 1 gives a sufficient reduction of the LF. Oherwise other step values are tried. cutf = cut factor. Local fits are rejected, if their chi^2 value is larger than the 3-sigma chi^2 value times the cut factor. A cut factor of 1 is used finally, but initially a larger factor may be used. A value of 0.0 means no cut. rejects = total number of rejected local fits. hmmsec = the time in hours (h), minutes (mm) and seconds. FMS = calculation of Function value, Matrix, Solution.