______________________________________________________________________________ #020-INFO: Choosing the Right TN #020 Igor Technical Notes ______________________________________________________________________________ WaveMetrics Technical Support #020-INFO: Choosing the Right TN #020 Written by: James Prouty Jun 1992 Updated by: James Prouty Dec 2008 TN #020: Peak Measurement and Fitting In the beginning, there was Technical Note #020, the Peak Measurement experiment. Well, okay, it was actually October 1990. It fit the sum of Gaussian or Lorentzian peaks and a baseline function to your data. It handled 1-6 peaks. The experiment was way cool, but impervious to customization. TN#020-B: Peak Areas And lo! Other customers saw that six peaks was too few, and couldn't we help them with their two hundred peaks? Yes, we could. Thus was born the Peak Areas experiment. This experiment is useful when you have data containing a large number of peaks and need "rough and ready" estimates of peak parameters. This experiment does not fit a peak function to your data; it just measures the amplitude, width, and areas of peaks. There is no programmed limit to the number of peaks, but each peak measurement takes about 1 second on a Mac IIfx. NOTE: The Peak Areas experiment's automatic peak finder assumes the data has regularly spaced periodic BIPOLAR PEAKS (peaks that have easily distinguished minima and maxima). If your data is unipolar (gaussian-like peaks on a baseline), use TN#020-C Unipolar Peak Areas. The automatic peak identifier is crude, but can be modified by the knowledgeable user. If the automatic peak finder doesn't work with your data, you can still make peak measurements. Just use the cursors to identify the peak start, center, and end. The experiment does the rest: it can create and subtract a piece-wise linear baseline, and make the measurements automatically. TN#020-C: Unipolar Peak Areas This is essentially a re-write of TN#020-B to handle unipolar peaks. Instead of using peak minimums, thresholds are used to locate the peak start and peak end. This experiment doesn't have its own documention; please refer to TN#020-B documentation. TN#020-H: Holding Peak Fitting Estimates Constant This is a small modification of the original Technical Note #020, the Peak Measurement experiment. The 'Peak Measurement' experiment estimates the position, width and amplitude of peaks automatically or semi-automatically. If you know any of these parameters and don't want Igor to vary them during peak fitting, TN020 does not provide an easy way to to that. This experiment (Peak Meas Enter Est & Hold) provides those capabilities, in addition to all of the capabilities of the original TN020 Peak Measurement experiment. It remains difficult to modify. [use a fixed-width font for this table] SUMMARY of FEATURES ----------------- by Technical Note ------------------ 020 & 020-H 020-B 020-C Number of Peaks: 1-50* 1 - inf 1-inf Peak Fitting: Yes No No Peak Measurements Based On: Fit Parameters Direct Measurement Direct Measurement Peak Identification: FindAPeak Local Min/Max Min Threshold (FindPeaks XOP) Local Maxima Compatible Peak Shape: Unipolar peak Alternating Unipolar peak on baseline +peak -peak on baseline Peak Shape Functions: gauss, lor (n/a) (n/a) Ease of Modification: Difficult Easier Easier Documentation: TN020 TN020 TN020 TN020-H TN020-B TN020-B TN020-INFO * 50 peaks is a suggested practical peak fitting limit. More peaks may be analyzed, but the peak fitting will become increasingly slow and unlikely to converge. This problem would abate if constrains are added to the peak fitting process at some future date. TN020-INFO: Continuing the Peak Fit (more iterations) This note describes how you can change TN020 Peak Measurement and Fitting to cause the fitting of peaks to proceed after Igor gives up trying to fit. Sometimes this results in a successful fit in very difficult situations. These changes do not provide you with more iterations of 'FitBaselineAtRegions'.