Pattern matching computer programs

When a particular component eluting at a certain retention volume is to be estimated, this approach can be outlined as follows. Since SEC is extremely reproducible, the peak shape, peak width and peak height are dependent on the amount of the species in the sample volume injected, sample volume and retention time. From these factors the SEC peaks can be simulated or elution pattern of any species within the separation range can be plotted as a function of mass vs. retention volume. The analysis data supplies the concentration of this particular species over two or more 0.5 ml intervals. A match-up computer program has to be developed so that it can pick up the peak shape and concentration based on 3 or 4 data points at known Intervals. 

There are collections of calculated spectra that can be used to match complex, first-order, splitting patterns (see references to Wiberg 1962 and Bovey 1988). Alternatively, these spin systems can be simulated on the computer of a modern NMR spectrometer. For example see the NMRSIM computer program available from Bruker BioSpin. 

A regular expression (sometimes abbreviated regex) is a way for a computer user or programmer to express how a computer program should look for a specified pattern in text and then what the program is to do when each pattern match is found (www.whatis.com). 

ACT-R is composed of perceptual-motor and memory modules implemented as production rules. Both declarative and procedural memory are modeled. A pattern-matching mechanism is used to select production rules applicable to the situation. Buffers are used to interact with and represent the state of modules. ACT-R is implemented as a computer language. Users create a program containing relevant task data to model the task and task performance. An extension of ACT called SNIF-ACT that incorporates a Bayesian navigation mechanism was described earlier in this chapter (Fu and Pirolli 2007). 

Sea.rch-Ma.tch. The computer identifies which crystalline phases (components) match the unknown pattern by using a file of known powder patterns maintained by the International Center for Diffraction Data (ICDD). The Powder Diffraction File contains interplanar t5 -spacings d = A/(2sin0)] and intensities of the diffraction maxima for each crystalline powder pattern submitted to the ICDD. Currendy there are about 65,000 patterns in the file. Current search—match programs can successfully identify up to seven components in an unknown pattern. A typical diffraction pattern of an unknown sample and the components identified by the computer search-match program is shown in Figure 15. 

Fig.7 shows an example of the type of fit we obtain between experiment and theory. The experiment pattern was recorded in a CCD camera, and energy filtering was not used. The experimental and theoretical patterns are processed by a line detection program. To save computation times, only selected areas of the experimental pattern are matched. The areas are selected based on their sensitivity to lattice parameters. 

In a modern diffractometer, computer software performs the search-match task. All PDFs of the ICDD can be stored in computer. A search-match program can find all the possible matches for a specimen. The software algorithms also can identify more than one crystalline phase in a specimen. It searches the recorded pattern with its background, and adds candidate crystalline phases together to compose, rather than decompose, an observed multiphase pattern. 

The computer based identification of crystalline phases in powder diffraction patterns normally requires two separate components (a) a powder diffraction database containing reference information and (b) a search-match program that loads the diffractogram and accesses the database to attempt to match the diffraction data to known phases in the database. 

Raw X-ray diffraction data, either digital or acquired on a strip recorder, are used make mineral identifications as summarized above. Whole sample data collected from random powder mounts are compared to patterns of known minerals either manually or using a computer search-match program such as /zPDSM (Marquart, 1986). Because each component in a mixture of crystalline materials produces its own characteristic pattern that is independent of others, the identification process becomes one of simply unscrambling the superposed patterns. 

---