PMF Evaluation Panel v2.03 Nov. 9, 2009 Update since PMF-AMS Analysis v2.02 Notes by Ingrid Ulbrich 8th Nov. 2009 ======================================== IMPORTANT CHANGES YOU SHOULD KNOW ABOUT ======================================== * No changes should interfere with old experiments. To upgrade, replace old .ipfs with new .ipfs and remake the main panel from the PMF pull-down menu. * Error preparation functions are now available in functions * pmf_ErrPrep_Q-AMS_OneIonEquiv_v2.3.ipf or pmf_ErrPrep_ToF-AMS_OneIonEquiv_v2.3.ipf * pmf_ErrPrep_AMS_v2_3.ipf * Convergence criteria for the PMF calculation can be set proportional to Q/Qexp using flags in the header of PMF_Execution_v2.03.ipf * Default SEED and FPEAK can be changed in the headers of PMF_Execution_v2.03.ipf * Alternate PMF executable file can be selected in the headers of PMF_Execution_v2.03.ipf * Download the updated version of mypmft.ini * Deleting NaNs from matrices has been made much faster. Changes made are backwards compatible with previous versions. * A new option has been added for viewing factor time series overlayed and mass spectra separately. Use Select Factor = -2. ======================================= DETAILED CHANGES ======================================= ----- PMF_Execution_v2.03.ipf ----- * Significantly improved the efficiency of pmf_ams_DeleteNaNs_mxWvs. NaNsLists are now stored as 2D NaNsWvs. * Created pmf_ams_NaNsList_2_NaNsWv for backward compatibility with old NaNsLists * Updated all pmf_ams_DeleteNaNs and pmf_ams_insertNaNs functions to use NaNsWvs *Changed the calculation of Qexp (for Qmx2d and Chi2NoDownwtMx2d) to be correct (mn-p(m+n)) (for Resid_as_TseriesFactors) to be correct (m-p) or (n-p) and for model error to be applied in these calcs * Correctly applied user-specified ModelError to calculation of Q-contribution time series/mass spectrum * Fixed SEED when running in Exploration Mode and changing FPEAK * Fixed problems with paths for Vista machines * Added a new strconstant line for choosing PMF executable file as pmf2wopt.exe; requires change to mypmft.ini! * Added constants DEFAULT_SEED and DEFAULT_FPEAK so that one can run e.g., a range of fpeaks at DEFAULT_SEED=2 * If PMF executable not found in PMF path, warning text is red * In View PMF Results Selection panel, pulldown menus for row and column waves only include waves with dimensions that match matrix size * Implemented James' bug fixes for bootstrap factor reordering * In Bootstrapping, if noNaNs_amus and noNaNs_t_series do not exist for plotting average factors, user is prompted for waves to use * In Bootstrapping, correlations between bootstrap cases and base case are made with uncentered correlation (instead of Pearson R) at recommendation of James Allan ----- PMF_ViewResults_v2.03.ipf ----- * Save Sol'n Space Waves button now creates subfolder Diurnals and save factor, total Resid, Q/Qexp diurnals * for Factor Profile plots, changed left axes to cross bottom at 0 instead of 10, narrowed bottom axis * made a global variable minY_Log_Profiles which the user can set with a min Y axis in log view; max in log view now = 1 * for Total Residual plots, changed left axes to cross bottom at left edge, narrowed bottom axis, changed axis label for Q-contribution to include /Qexp * for Credits, updated paper info to ACP 2009 version * added button to link to PMF-AMS Analysis wiki * for plots of species TS, total signal TS, resid TS, resid ratio TS, both axes of RR plot, make y-axis go from 0 or full scale if min values is negative * Factor Profile plots, y-axes cross the left axis at 0 (was at 10, a problem for non-AMS users who might start the indexes for their data at 0. * for species reconstruction and total reconstruction plots, put reconstructed trace on top of measured data trace * for TS overlay, made new option Select Factor = -2: TSS are overlayed, MS are plotted separately ----- PMF_Scatter_v2.03.ipf ----- * Updated scat_make_noNaNs_ms_OR_ts_DF to use NaNsWvs instead of NaNsLists. For compatibility with previous versions, if no NaNsWv is found, looks for NaNsList and converts it to a NaNsWv