Spacing of data

<< Click to Display Table of Contents >>

Navigation:  Tutorials > Artefacts and stochastic realisations >

Spacing of data

The spacing of input data is another thing that PoreXpert takes literally.  The goodness of fit, or Distance, is measured averaging the distance from each experimental percolation point to the nearest simulation point.  So if your datafile has a large number of points in a particular region, it will assume that you think that region is most important.  However, as commercial instruments tend to work from a table of equally spaced pressure, suction or tension values, there are usually fewer points along the point of inflexion part of the percolation curve rather than in other regions.  Some commercial porosimeters can give out a very large number of data points, but there is no advantage to PoreXpert beyond having a hundred or so spaced appropriately along the intrusion curve.  For datafiles comprising very many intrusion points, we have a Python application to thin them - it selects points 1% apart along the percolation curve in two-dimensional parameter space defined by scaling the both the logarithmic axis and the intrusion range to 100%.  An example is shown below - note that the spacing is equal along the curve itself, so the steeper part of the intrusion curve is more closely spaced with respect to the logarithmic diameter axis.  

data_thinning_example

For the purposes of this tutorial, we will just carry out the data thinning exercise manually on our example file, although in this case there is not much of a problem. Delete the slightly bunched data points at 22.45 and 16.40 μm, as shown at the bottom of the graph below.

data_thinning_medicinal_tablet

Your datafile should now look like this, ready for the next part of the tutorial, on how to find the optimum structure type.

 

Medicinal_Tablet_datafile_truncated_thinned