I have installed R-2.8.1 right now and will proceed with my tests. So far I haven't used it. The fault is presumably caused by a sort of misuse of the memory on my part. The attached R script (whose last Step is not debugged yet) implements the algorithm for signal features extraction described in the attached paper. Needless to say, to get good statistics we need plenty of data.Therefore I started out with a minimum of 30 signals (please find attached 48 detrended, denoised, centralized, standardized, and compressed signals I am using with the script). The algorithm generates many big matrices and vectors. Roughly a matrix is generated at each step. Of course in my last script version I will remove the useless data structures dynamically. At the time being, though, I keep them all because I have to trace back each data elaboration step. Between successive runs I clear the console but rarely I clear the workspace. Even if the data structures are generated again at each run, maybe the system memory stack gets fragmented ? ... just my worthless guess ... I am afraid I cannot prune the code without changing the algorithm. The detected fault seems to be a memory problem. Therefore, the allocation of big data structures over and over again is suspicious ...
About the algorithm... my implementation seems to perform according to the prescription. However, I am not enthusiastic with results depending on vague choices of the parameters. Basically the number of signal features extracted depends on an arbitrary input threshold. Maybe someone in the R-help list might suggest a better, more robust, way to extract features from mono-channel signals ... Kind regards, Maura E ======================================================================== ==== Benvenuto in Alice Giga Mail! mau...@alice.it tramite il servizio Giga Mail ha messo a tua disposizione i seguenti allegati: * Archive.zip ( 5980513 bytes ) * Signals-Features-Extraxtion-through-DWT-Coefficients.txt ( 10148 bytes ) * wave.pdf ( 256758 bytes ) per scaricarli, fai click sul seguente link che ti portera' su una pagina dove troverai i comandi per visualizzare o scaricare gli allegati sul tuo PC: http://gigamail.rossoalice.alice.it/messages/readMessageFrameset.aspx?De liveryID=19412684-adf5-4e53-a1fe-a8e58848cb8e Ti ricordiamo che gli allegati saranno a tua disposizione fino al 08-01-2009 alle ore 11.35.04 e che il mittente potrebbe ricevere le informazioni relative alla tua apertura della Giga Mail e all'avvenuto download degli allegati. GigaMail è il nuovo servizio gratuito di Alice che ti permette di inviare a chi vuoi, allegati di grandi dimensioni, fino a 2GB, in modo semplice e veloce, senza occupare spazio utile nella tua casella di posta. Per saperne di più visita il sito www.alice.it Ti ringraziamo per aver utilizzato il servizio Alice GIGA MAIL. [[alternative HTML version deleted]]
______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
