Centring, Normalisation

There are numerous publications proposing a glut of data treatment methods prior to PCR/PLS. Well established, tested and essentially universally applied are mean-centring and normalisation of the data. We have seen in Mean Centring, Closure (p.239) that mean centring reduces the dimensionality by one, which of course cannot harm. In PCR/PLS it is also common to normalise to the standard deviation of the signals. Both are implemented in Main PCR. m. 

Mean-centring and normalisation are optional. The PCR (and PLS) algorithm are essentially independent of the nature of pre-treatment of the data, only the centring has to be reversed in the prediction step. In the programs we 

---

Fig.2. Comparison of the SIF of a single notch under pure bending (Ki ) and that from finite element calculation of the unotched G1 configuration (Ki). The different x values represent increasing off centre, normalised by the inner span.

The difference between these two spectra is shown in Figure 5.42(b). The shape of the difference spectrum depends on the magnitude of the scaling factor used to normalise the two spectra. These authors chose a factor so as to match the integrated intensities in a 50 eV window centred at 980 eV, 50 eV above the edge threshold. 

An indication of the environmental effects of hydrogen deployment for each of six European countries (France, Germany, Italy, the Netherlands, Spain, and the UK) is given in Fig. 19.10, where the fine dust emissions are shown for the hydrogen high-penetration scenario. Similar trends are found for other pollutants. The data are normalised in respect to the baseline scenario and show a trend very similar for the analysed countries with a reduction of more than 70% in 2050. The results are an average per country. At a local level, higher reductions can also be achieved if nontechnical measures, such as limitation of city centre access for non-zero emission vehicles, are taken. 

Now we use the information gathered so far for the prediction of the 10 test samples Ys removed from the complete data set at the veiy beginning. The function PCR PLS pred. m does the work according to equation (5.70). Importantly the mean-centring and normalisation have to be performed in exactly the same way as in the calibration. 

In direct methods calculations we use normalised structure factors E(hkl), which are the structure factors compensated for the fall-off of the atomic scattering factors f hkl). In fact this procedure tries to simulate point-like scattering centres. 

Fig. 3.13. Partial least-squares (PLS) calibration of the API data set (5 s accumulation time). Spectra were baseline corrected, normalised to unit length and mean centred. The data set was randomly split into a calibration set (two-thirds) and a prediction set (one-third) obvious outliers from the PCA analysis were excluded from the analysis. The graph shows predicted versus measured API concentration of the prediction set. The straight line represents the 45° diagonal (this figure was published in [65], Copyright Elsevier (2008))...

Fig. 9.7. REE patterns and mantle-normalised incompatible element patterns (restricted to mafic rocks for most centres) for Plio-Quatemary volcanic rocks from Sardinia.

Figure 2.6 PCA of the top 200 small molecule drugs in 2006 including natural product drugs (red) and 595 Merck natural products (blue). The descriptors MW, HBA, HBD, AlogP98, PSA, normalised bond flexibility, nitrogen count, oxygen count and chiral centre count were used.

Figure 3 Normalised single particle velocity correlation functions and their normal mode approximations ip (<)/i/) (0) and i/ ° (<)/0 (O). Left panel centre-of-rnass velocity correlation function for CO, 80 K, 1 bar right panel rotational velocity correlation function for CSj, 293 K, 10 kbar. Thick lines Molecular dynamics result dotted lines normal mode result deluding unstable modes thin lines normal mode result including unstable modes.

Baklanov, A. (1986) Numerical modelling for normalisation of atmospheric environment on industrial sites, in Numerical solution of atmospheric hydrothermodynamics problems, Novosibirsk Computing Centre RAS, 30-38. 

Centre of Environmental Science - Leiden University (CML) (2001) Characterisation and normalisation factors. Leiden, The Netherlands... 

Figure 7.16 Views of an organic small molecule crystal solved with wavelength normalised Laue data and MULTAN and comparison with monochromatic result. The 20 hydrogen atoms were located from difference Fourier syntheses after anisotropic temperature factor refinement of the 29 nonhydrogen atoms. This work illustrates the sensitivity of the Laue data to weak scattering centres such as hydrogen and shows it is as good as monochromatic data. From M. Hel-liwell et al (1989) with the permission of IUCr.

The one-centre expansion (6.15) is specialised to the case where R = 0, and is valid inside the atomic sphere centred at the origin. It may be used to derive the LMTO equations and with the normalisation implied by (6.11) it is consistent with the secular matrices (5.46,47) in the ASA. In linear methods in band theory [6.2] Andersen presented the one-centre expansion in the form (6.15) and derived the LMTO formalism from that assumption. His LMTO formalism is equivalent to that presented here apart from the normalising factor [/S/2 (- -1)] 1 appearing in the definition (6.11) of the energy-dependent muffin-tin orbital. 

In the LMTO method, attention is focussed upon an energy range centred around some energy which we are free to choose to suit the problem at hand. Therefore, for each value of l we choose an E, and use the energy-independent basis set formed by the normalised radial function... 

Fig. 3 23. (a) Grofh of energy vs volume (scale normalised to the diamond structure) for eleven phases of silicon (b) Enthalpy-pressure plot for the same eleven phases relative to the body-centred cubic phase (Figures redrawn from Needs R f and A Mujica 1995. First-principles pseudopotential study of die structural phases of silicon. Physical Review B51-9652-9660.)... 

The normalisation of the basis functions is nearly as simple and involves the (one-centre) overlap integral between two primitives of the same type but different exponents ... 

Because in ASA the basis functions (oX ) assumed to be normalised Is GTO fimctions, with centre at the position R, one can write the gradient vector as ... 

Fig. 2.10 Simulation of an observation of a binary system consisting of 2 unresolved monochromatic point sources separated approximately 9 mas (left) and / = 0.5. The normalised squared visibility amplitude in units of spatial frequency 1 /X (centre) has a cosine shape given by the source separation. Three baselines have been simulated aligned with the point sources (blue), perpendicular to the line connecting the point sources (red), and in an angle between red and blue (green)...

Note that the simulation space is now constrained to the region 1 < i < 1 + VTmax, R = 0 is the point at the centre of the electrode (its interior) so naturally the simulation space begins at i = 1 (r = rg), the electrode surface. The time is likewise normalised against the radius ... 

The noise generated from moving vehicles on open-textured surfaces, similar to PA, has been found to be reduced when compared with conventional surfacing materials of the same macro-texture and skid resistance coefficient. The reduction measured was 3 to 4 dB(A), when the pavement surface was dry, and 7 to 8 dB(A), when the surface was wet (Nelson and Ross 1981). Similar results were also found by the Belgium Road Research Centre (Decoene 1989) and by Phillips et al. (1995). At a later study, resurfacing an old concrete pavement with PA led to wind-normalised reductions in noise exposure of the order of 4.5 to 6 dB(A) L/ io,i8h at villages approximately 0.5 km from the motorway (Baughan et al. 2002). 

Memorandum on the Bases for Normalisation of Relations Between the Republic of Moldova and Transdniestria, 8 May 1997, at para. 2. Full text available at http //www.intstudies.cam.ac.uk/ centre/cps/documents moldova memo.html,... 

One relatively simple extension of RIS theory is to deduce / , and matrices that are not normalised, from computer-based modelling. From this, one can deduce the relative probabilities of formation of various tacticity sequences during polymer propagation. These may be compared with the experimental probabilities, obtained in polymers where the asymmetric centres can be part-epimerised by chemical means [26]. 

---