krj@na.toronto.edu (Ken Jackson) (02/14/90)
Matlab Digest Tuesday, February 13, 1990 Volume 2 : Issue 1 Today's Editor: Chris Bischof Today's Topics: Response to request for moving average filter Possible bug in PCMATLAB Re: Matlab Surface Plots Matlab on Lap-Tops new submissions to the matlab library ---------------------------------------------------------------------- >From mathworks!matlab!jnl@apple.com Tue Jan 2 18:16:35 1990 Subject: Response to request for moving average filter In response to the newsgroup request for a box car or moving average filter algorithm, try the following: b = ones(1,5)/5; y = filter(b,1,x); This implements a five point MA filter. John Little The MathWorks, Inc. jnl@mathworks.com info@mathworks.com Ph 508-653-1415 FAX 508-653-2997 ---------------------------------------------------------------------- >From hgf@lakisis.umd.edu Fri Jan 5 10:25:51 1990 Subject: Possible bug in PCMATLAB I would like to report on a bug which I discovered today. Playing around with PCMATLAB (on a 386 machine of Gateway) I ended up typing plot(fft(1,y)); where y was a (1,1024) vector. (hoping to get the real part,....OK). The result was surprising: I was back to DOS without warning. Tracing back it turned out, that fft(1,y) gives an empty matrix, and that plot(x) is OK, if x=[]; but the combination I used turned out to be fatal. Of course I don't know what happend in the background, but this should not be so much my problem, I hope. Best regards, Hans G.Feichtinger. ---------------------------------------------------------------------- >From lozier@blaze.cam.nist.gov Wed Jan 17 08:04:45 1990 Subject: Re: Matlab Surface Plots Thanks for suggesting a way around my blacking-in problem. If I follow the suggestion and reduce the data by a factor of 4, so that the matrix is 250X16 instead of 1000X16, the blacking-in is not as serious a problem. The data suffers a slight but noticeable loss of detail. If I reduce it by a factor of 8, the loss of detail is unacceptable. The data arises from a partial deconvolution process. A smooth signal is deconvolved progressively, leading to signals that become less and less smooth. The surface illustrates the result of going from the original signal (no deconvolution) to the fully deconvolved signal in 16 equal steps. The fully deconvolved signal exhibits sharp spikes that can only be resolved by taking lots of points. Making xy plots of 1000 points presents no problem for matlab graphics. Even though 1000 points may be more than are needed in many cases, it is very convenient not to have to decide which ones are unnecessary. It seems to me that the same convenience should be available in three dimensions. All that is needed is a way of telling the mesh command not to draw mesh lines in one of the coordinate directions. In fact, we are already exhibiting the data in this way with an Evans & Sutherland graphics terminal. Dan Lozier lozier@cam.nist.gov ---------------------------------------------------------------------- >From lnt@math.mit.edu Thu Jan 25 16:28:15 1990 Subject: Matlab on Lap-Tops I'm planning a sabbatical journey but can't survive without Matlab. Does anyone have experience with Matlab on lap-tops? Nick Trefethen ________________________________________________________________________ mail: L. N. Trefethen, Dept. of Math., M.I.T., Cambridge, MA 02139 e-mail: lnt@math.mit.edu OR na.ntrefethen@na-net.stanford.edu phone: 617-253-4986 (office), 617-861-3921 (home) ---------------------------------------------------------------------- >From bischof@mcs.anl.gov Subject: New submissions to the matlab library We have some new submissions in the matlab library: In `approximation': lagrange: Lagrange interpolation of arbitrary order (Micheal Saucier) In `control': hankel: Hankel model reduction (Richard Chiang, USC) lyapunov: Solution of a Lyapunov or Sylvester equation (Richard Chiang, USC) modelred: Some files for model reduction (M. Steinbuch, KEMA & Delft University of Technology) orbit: orbit mission analysis tasks, using two body theory (Don Mackison, University of Colorado, Boulder) ztransform: subroutines for converting from the Laplace domain to the discrete domain (Arthur Jutan, University of Western Ontario) stuff1, stuff2, stuff3: miscellaneus stuff from the old matlab user group toolbox distributed by the Mathworks In `dataanalysis': stat: several statistics functions (A. T. McAdams) In `graphics': blockdiagram: Files for drawing a block diagram (Garth Milne) In `linearalgebra': csdgsvd: CS decomposition and generalized SVD (S. J. Leon) osb: diagonal scaling of a complex matrix (C. N. Nett, GE CRD/CTB) testmatrices: A collection of fourty-five parametrized test matrices. (Nick Higham, University of Manchester) In `misc': aero: some files related to air vehicle modelling (from the old matlab user group toolbox distributed by the Mathworks) shareware: miscellaneous stuff from the old matlab user group toolbox distributed by the Mathworks tax: various files for computing income taxes (from the old matlab user group toolbox distributed by the Mathworks) In `optimization': leasqr: Levenberg-Marquardt nonlinear regression (Richard I. Shrager) There also was a typo in the directory of contents distributed in v1no3. Sigmon Kermit's matlab primer is `primer.tex' in `teaching', not 'teaching.tex'. -- Chris Bischof ---------------------------------------------------------------------- <end of matlab digest> Reposted by Prof. Kenneth R. Jackson, krj@na.toronto.edu (on Internet, CSNet, Computer Science Dept., ARPAnet, BITNET) University of Toronto, krj@na.utoronto.ca (on CDNnet and other Toronto, Ontario, X.400 nets (Europe)) Canada M5S 1A4 ...!{uunet,pyramid,watmath,ubc-cs}!utai!krj (Phone: 416-978-7075) (on UUCP) (FAX: 416-978-4765)