📄 changes.html
字号:
<html><head><title>Bug fixes</title></head><body bgcolor=white><a name="top"></a><h1 align=center>Changes, bug fixes and new programs</h1><hr><ul><li><a href="#tisean2.0">TISEAN_2.0 -> TISEAN_2.1</a><li><a href="#tisean1.2">TISEAN_1.2 -> TISEAN_2.0</a><li><a href="#tisean1.0">TISEAN_1.0 -> TISEAN_1.2</a><li><a href="#tisean0.9">TISEAN_0.9 -> TISEAN_1.0</a><li><a href="#tisean0.5">TISEAN_0.5 -> TISEAN_0.9</a><li><a href="#tisean0.4">TISEAN_0.4 -> TISEAN_0.5</a><li><a href="#tisean0.3">TISEAN_0.3 -> TISEAN_0.4</a><li><a href="#tisean0.2">TISEAN_0.2 -> TISEAN_0.3</a><li><a href="#tisean0.1">TISEAN_0.1 -> TISEAN_0.2</a></ul><hr><center><font size=4 color=blue><b><a name="tisean2.0">TISEAN_2.0 -> TISEAN_2.1</a></b></font></center><hr><h4 align=center>Changes</h4><ul><li>I continued to modify programs in such a way that they are able tohandle multivariate data. Modified in this release: <ahref="ar-model.html">ar-model</a>, <a href="extrema.html">extrema</a>,<a href="nstep.html">nstep</a>, <a href="boxcount.html">boxcount</a>,<a href="lyap_spec.html">lyap_spec</a>, <ahref="recurr.html">recurr</a>, <a href="makenoise.html">makenoise</a>and <a href="rescale.html">rescale</a>.</li><li>To achieve consistency with the other programs I renamed someflags. The affected programs and flags are: <a href="d2.html">d2</a>flags <font color=red>-e to -r</font>, <font color=red>-E to -R</font>and therefore, <font color=red>-R to -E</font>, <ahref="av-d2.html">av-d2</a> flag <font color=red>-e to -E</font>,<a href="histogram.html">histogram</a> flag <font color=red>-e to-r</font> and <a href="boxcount.html">boxcount</a> flag <fontcolor=red>-e to -r</font> and <font color=red>-E to -R</font>.</li><li> I gave unique <a href="../../source_c/routines/tisean_cec.h">exitcodes</a> to all the exceptions in the routines. I'm pretty sureThomas will have doubts about the usefulness of this, but who knowswhat it is good for.</li><li>Again <a href="recurr.html">recurr</a>: It does no longer printthe diagonal terms. That was really nothing than a waste of discspace. Furthermore, it now uses a two-dimensional box assisted searchalgorithm. Should be a bit faster, therefore.</li><li><a href="makenoise.html">makenoise</a> has got a new flag <fontcolor=red>-0</font> which, if set, allows to produce random numberswithout reading any data.</li><li><a href="zeroth.html">zeroth</a> was replaced by the old <ahref="zeroth.html">mzeroth</a>. It got one newfunctionality. Depending on the Verbosity level it now prints theindividual forecast error for each component of each reference point.</li><li><a href="nstep.html">nstep</a> got a new flag <fontcolor=red>-0</font>. Setting this flag causes nstep to perform azeroth order fit instead of a local linear one.</li><li><a href="nstat_z.html">nstat_z</a> the flag <fontcolor=red>-S</font> was renamed <font color=red>-s</font> to beconsistent with the other programs.</li><li>The programs <a href="nstat_z.html">nstat_z</a>, <ahref="zeroth.html">zeroth</a>, <a href="onestep.html">onestep</a>and <a href="ll-ar.html">ll-ar</a> got a new flag <ahref="causality.html">-C</a>.</li><li>The program <a href="nrlazy.html">nrlazy</a> got a new verbositylevel. If this level is set the number of neighbors used for thecorrection is printed to the output file. If this number is 1, nocorrection was done.</li></ul><hr><h4 align=center>Bug Fixes</h4><ul><li>There was a slight inconsistency in the <ahref="xcor.html">xcor</a> man page. The page stated the output filecontains <x<sub>t</sub>*y<sub>t-d</sub>>, instead it is<x<sub>t</sub>*y<sub>t+d</sub>>. REMOVED (Thanks to Marioagain)</li><li><a href="d2.html">d2</a> crashed when the maximal dimension wasONE. FIXED. (Results for 'non crashing' cases were not influenced, atall)</li><li>There was a bug in <a href="mutual.html">mutual</a>, which couldlead to negative mutual information in case that the delay time waslong compared with the length of the time series. FIXED (Thanks toEckehard) </li><li>Though stated somewhere on this page, the output of <ahref="histogram.html">histogram</a> was not produced in such a waythat the integration over the whole data interval gave 1. Now itdoes.</li><li>I removed a lot of 'index out of range segmentation faults' in theprograms. (Thanks to a lot of people)</ul><hr><h4 align=center>New programs</h4><ul><li> <a href="sav_gol.html">sav_gol</a>: Performes a Savitzky-Golayfilter on, possibly multivariate, data to either clean the data or toget better estimates for their derivatives.</li><li><font color=red>Suggestions for further (not too specific)programs are welcome.</font></li></ul><hr><a href="#top">top</a> of the page.<hr><center><font size=4 color=blue><b><a name="tisean1.2">TISEAN_1.2 -> TISEAN_2.0</a></b></font></center><hr><h4 align=center>Changes</h4><ul><li>A new routine <a href="../../source_c/routines/exclude_interval.c">exclude_interval</a> was added to <ahref="routines.html">tsa.h</a>. It allows to exclude an interval ofindexes from a list of neighbors found.</li><li><a name="lyap_k"></a>When the parameter of <font color=red>-m</font> is larger thanthat of <font color=red>-M</font> in <a href="lyap_k.html">lyap_k</a>,now the <font color=red>-m</font> value is used. (Thanks to JensTimmer)</li><li>Furthermore, <a href="lyap_k.html">lyap_k</a> was rewritten in C.</li><li>The programs <font color=blue>delay, make_ps</font> wereremoved from the supported part of the package. In case you area fan of these programs, you'll find the sources and the html-files inthe directory <font color=blue>unsupported</font>. As thename says, there will be no support, what so ever, for theseprograms. Maybe <font color=blue>make_ps</font> will come back in alater version when it is useful for more people, not only for me.</li><li>The program <a href="histogram.html">histogram</a> was rewrittenin C. This means the package is now C++ free.</li><li>The program <a href="nrlazy.html">nrlazy</a> has an additionaloption <font color=red>-v</font> which allows to set the neighborhoodsize in units of the standard deviation of the data.</li><li>The program <a href="d2.html">d2</a> is now able to handle multivariatedata, also with mixed embeddings. Further, the default for the <fontcolor=red>-N</font> is now 1000 instead of all. To get all back, set thevalue to 0. Furthermore, columns 1 and 3 in the output files areinterchanged. The first column now contains the length scales in unitsof the data.</li><li>The program <a href="av-d2.html">av-d2</a> was changed in a waythat it is now able to handle the output of the files of the program. Since the natural unitsfor length scales are the default units now, the option <fontcolor=red>-r</font> was replaced by <font color=red>-e</font>.</li><li>Some of the messages of the programs were changed to avoid thatthey look like error messages. (Thanks to ChristianRödenbeck)</li><li>Two new routines to handle multivariate time series were added to<a href="routines.html">tsa.h</a>: <ahref="../../source_c/routines/make_multi_box.c">make_multi_box</a> and<a href="../../source_c/routines/find_multi_neighbors.c">find_multi_neighbors</a>.</li><li>To reduce the number of <i>segmentation faults</i> anotherroutine <ahref="../../source_c/routines/check_alloc.c">check_alloc</a> was addedto <a href="routines.html">tsa.h</a>.</li><li>A new flag <font color=red>-S</font> was added for the <ahref="zeroth.html">zeroth</a> program.</li><li>The <font color=red>-F</font> was replaced by <fontcolor=red>-c</font></li>.<li>To make the <font color=red>-c</font> and <fontcolor=red>-m</font> flags more flexible to use I changed the routine<ahref="../../source_c/routines/get_multi_series.c">get_multi_series</a>in <a href="routines.html">tsa.h</a>.</li><li>Now <a href="mutual.html">mutual</a> uses the standard norm forits output.</li><li>A new routine <ahref="../../source_c/routines/what_i_do.c">what_i_do</a> was added to<a href="routines.html">tsa.h</a>. Itdisplays kind of a welcome message when the program is started.</li><li>A new routine <ahref="../../source_c/routines/myfgets.c">myfgets</a> was added to <ahref="routines.html">tsa.h</a>. It handles loooong input lines.</li><li>A new flag <font color=red>-V</font> was added for allprograms. It allows to add an <a href=../general.html#verbosity>verbosity level</a> like in Thomas'programs. See the corresponding help pages for more info. According tothis change, I had to modify the routine <fontcolor=blue>get_series</font> and <font color=blue>get_series</font> in<a href="routines.html">tsa.h</a>.</li><li><a href="lyap_spec.html">lyap_spec</a> is now able to handledelays other than 1.</li></ul><hr><h4 align=center>Bug Fixes</h4><ul><li><a name="ll-ar"></a>When using the routines <a href="ll-ar.html">ll-ar</a>, <ahref="onestep.html">onestep</a> and <a href="zeroth.html">zeroth</a>with the <font color=red>-s</font> flag set to a value larger than 1the points mapped into the interval [actual point,actual point + valueof -s] were not excluded from the neighborhood. This caused causalityproblems especially for oversampled flow data, which could result ina severe underestimation of the forecast errors. FIXED (Thanks to MarioRagwitz and Holger)</li><li>The <font color=red>-r</font> flag in <ahref="xzero.html">xzero</a> was buggy. When used, the result wasusually (depending on the architecture) a <font color=blue>floatingexecption</font>. (Thanks to Mario Ragwitz)</li></ul><hr><h4 align=center>New programs</h4><ul><li><a name="mzeroth"></a><a href="zeroth.html">mzeroth</a>: Estimates the forecast errors fora multicomponent time series using a zeroth order model.</li><li><a name="nstat_z"></a><a href="nstat_z.html">nstat_z</a>: Takes a scalar time series,divides it into pieces and estimates the zeroth order cross-forecasterrors as a test for nonstationarity.</li></ul><hr><a href="#top">top</a> of the page.<hr><center><font size=4 color=blue><b><a name="tisean1.0">TISEAN_1.0 -> TISEAN_1.2</a></b></font></center></b></font></center><hr><h4 align=center>Changes</h4><ul><li><a name="ar-model"></a> In <a href="ar-model.html">ar-model</a> now the mean is first subtracted which should be less confusing.</li><li>A new routine <a href="../c/routines/test_outfile.c">test_outfile</a> was added to <a href="routines.html">tsa.h</a> which tests whether a file for writing the output into can be created or not.</li><li>Another new routine <a href="../c/routines/invert_matrix.c">invert_matrix</a> was added to <a href="routines.html">tsa.h</a>. It inverts a square matrix and is based on <a href="../c/routines/solvele.c">solvele</a>.</li><li><a href="lyap_k.html">lyap_k</a> wanted to have the epsilon values in units of the data, the output was in rescaled units, though. Since this was a bit confusing, it was changed. Now output is also in units of the data.</li><li>All programs were modified such that they use the routine <a href="../c/routines/test_outfile.c">test_outfile</a>.</li><li><a name="rbf"></a> <a href="rbf.html">rbf</a>: The distribution of the centers and the variance of the gaussians was changed to make the program more effective. Now less centers should be necessary. Furthermore, the program should be much faster now and thus the default is now: write to stdout.</li><li><a name="histogram"></a> <a href="histogram.html">histogram</a>: The output was changed in a way that now the integral over the second column should be 1.</li></ul><hr>⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -