<input file> | Source file from classesbymra |
<output file> | Destination file (ascii) |
[euler=<path>] | Path to the original image file for retreiving euler info |
[dim=<n>] | Dimensionality of the embedded space, generally 3 |
[start=<n>] | File containing starting positions |
[hemisphere] | Restrict the final points to lie on the surface of a hemisphere |
[fixq=<val>] | Output with a fixed quality factor rather than the calculated one |
[refpos=<file>] | A set of fixed positions that the distance matrix should be referenced to. |
[assigneuler] | In conjunction with the hemisphere option, this will assign eulers to each particle |
[bestref=<n>] | This will find a set of n images which are as different as possible for use as a reference set. |
matrixembed matrix.dat groel.loc
This is an experimental program for visualizing mutual distance matrices generated by the 'matrix' option of classesbymra. It generates an ascii file with the best (almost) possible representation of the distances in the input file in (usually) 3 dimensions. It also outputs a mean stress value for each point in the embedded space.
This is basically correspondence analysis.
Normally the input matrix file is the result of 'classesbymra proj.hed proj.hed matrix', but it is also possible to use this program to calculate positions referenced to a precalculated embedded space (the result of matrixembed on a square matrix). That is, once an embedded space has been determined, you can determine the embedded positions of individual particles using a matrix produced by 'classesbymra start.hed proj.hed matrix'. In this case, use the 'refpos=' option to specify the output coordinates from the initial matrixembed run.