|Name||Modified||Size||Downloads / Week|
|WaveSorter ReadMe.txt||2013-06-07||2.5 kB|
|Build WaveSorter on a Mac.txt||2012-11-29||863 Bytes|
|Totals: 13 Items||9.6 MB||1|
WaveSorter by Matt Phillips Please see the 'About' menu item for credits and attributions. In addition see the manual (docs/Manual.pdf) for full instructions on how to get the most out of WaveSorter. WaveSorter works on all major 64-bit platforms. To get started download the .zip or tar.gz file to wherever you want to run WaveSorter and extract it. You should be able to launch the program directly by clicking on the binary; if not, email me (my email address is in the 'About' box). When WaveSorter launches for the first time, create a new profile. You will then be prompted for your default data and default data output directories. After that, just start playing around with the sliders, disks, transforms and clustering algorithms and you should be able to do pretty well on your own for now. One non-obvious feature worth noting is the ability to 'zoom in' on the waveforms (right panel) using the scroll wheel, and drag the view to the left or right. Click the 'Set Samples' button to restrict the clustering to just the visible samples. Another feature is using Centers for Manual clustering; these partition the unaccepted waveforms in the current session into regions as measured by 2D Euclidean distance. These are useful for using with Circles; separating the discarded waveforms into distinct clusters where appropriate can improve the statistical validity of the clustering. Change Log: 1.0.5 1) WaveSorter runs on the Mac 10.7 or greater directly from the download! 2) It is now possible to create multiple user transforms simultaneously, and save and reload these. 3) Numerous gui improvements. 1.0.4 1) WaveSorter now handles large files, >1,000,000 waveforms on a PC with 9GB RAM. 2) M-index no longer normalizes by number of dimensions used to sort. 3) Fixed bug in reading ascii files. 4) Cluster number now appended to saved file names, in order to avoid overwriting. 5) Minor gui improvements. 1.0.3: 1) PCA algorithm 10x faster. Now only first 8 components calculated. 2) Greatly reduced RAM consumption, down to 10%. 3) Minor gui improvements. 1.0.2: 1) Fixed bug in saving Simple Ascii, Simple binary, and REX E-files 2) Corrected mis-attribution of Klustakwik algorithm 3) Removed spurious error message occurring when user saved without specifying a clustering 4) Fixed issues with setting SPC parameters 1.0.1: 1) Fixed bug in reading .nev files 2) Improved profile creation gui operation 3) Minor gui improvements.