COMPGROUPS.NET | Search | Post Question | Groups | Stream | About | Register

### RGG question

• Email
• Follow

I am trying to use RGG to compute eigenvectors, and can't seem
to find them in the output matrix. Anyone out there know how to
do it? There are instructions in the commentary but I don't get
it. I have experimented with a matrix (from some gun pages) whose
eigenvalues and -vectors are known, and don't recognise any of the
numbers in the vector matrix I get. I am probably calling RGG
incorrectly but is there someone who knows how? It's an oldish
Fortran package.

I have asked in another ng but so far no real help.
--
Dieter Britz (dieterhansbritz<at>gmail.com)

 0
Reply dieterhansbritz (63) 10/20/2011 9:16:18 AM

See related articles to this posting

On 20.10.11 11:16, Dieter Britz wrote:
> I am trying to use RGG to compute eigenvectors, and can't seem
> to find them in the output matrix. Anyone out there know how to
> do it? There are instructions in the commentary but I don't get
> it. I have experimented with a matrix (from some gun pages) whose
> eigenvalues and -vectors are known, and don't recognise any of the
> numbers in the vector matrix I get. I am probably calling RGG
> incorrectly but is there someone who knows how? It's an oldish
> Fortran package.
>
> I have asked in another ng but so far no real help.

I have no idea what the answer to your question is, but perhaps you can

Also, what about using LAPACK, which is well supported?

Good luck
Paul

 0

> I am trying to use RGG to compute eigenvectors, and can't seem
> to find them in the output matrix. Anyone out there know how to
> do it?

See http://www.netlib.org/seispack/rgg.f :

c      on output
c
c        alfr  and  alfi  contain the real and imaginary parts,
c        respectively, of the numerators of the eigenvalues.
c
c        beta  contains the denominators of the eigenvalues,
c        which are thus given by the ratios  (alfr+i*alfi)/beta.
c        complex conjugate pairs of eigenvalues appear consecutively
c        with the eigenvalue having the positive imaginary part first.
c
c        z  contains the real and imaginary parts of the eigenvectors
c        if matz is not zero.  if the j-th eigenvalue is real, the
c        j-th column of  z  contains its eigenvector.  if the j-th
c        eigenvalue is complex with positive imaginary part, the
c        j-th and (j+1)-th columns of  z  contain the real and
c        imaginary parts of its eigenvector.  the conjugate of this
c        vector is the eigenvector for the conjugate eigenvalue.

he/she seems to know!

Regards,

Arjan

 0
Reply arjan.van.dijk (248) 10/20/2011 10:11:24 AM

Arjan wrote:

>> I am trying to use RGG to compute eigenvectors, and can't seem
>> to find them in the output matrix. Anyone out there know how to
>> do it?
>
> See http://www.netlib.org/seispack/rgg.f :
>
> c      on output
> c
> c        alfr  and  alfi  contain the real and imaginary parts,
> c        respectively, of the numerators of the eigenvalues.
> c
> c        beta  contains the denominators of the eigenvalues,
> c        which are thus given by the ratios  (alfr+i*alfi)/beta.
> c        complex conjugate pairs of eigenvalues appear consecutively
> c        with the eigenvalue having the positive imaginary part first.
> c
> c        z  contains the real and imaginary parts of the eigenvectors
> c        if matz is not zero.  if the j-th eigenvalue is real, the
> c        j-th column of  z  contains its eigenvector.  if the j-th
> c        eigenvalue is complex with positive imaginary part, the
> c        j-th and (j+1)-th columns of  z  contain the real and
> c        imaginary parts of its eigenvector.  the conjugate of this
> c        vector is the eigenvector for the conjugate eigenvalue.
>
>
> or ask E.S. Swanson, http://fafnir.phyast.pitt.edu/CompPhys/index.html,
> he/she seems to know!
>
> Regards,
>
>
> Arjan

Thanks, both of you; meanwhile it dawned on me what my problem was,
and I have sorted it out. What I found was multiples of the known
eigenvectors and didn't recognise them as such.
--
Dieter Britz (dieterhansbritz<at>gmail.com)

 0
Reply dieterhansbritz (63) 10/20/2011 12:47:09 PM

In article <j7p5s0$v83$2@dont-email.me>,
Dieter Britz <dieterhansbritz@gmail.com> wrote:

> Thanks, both of you; meanwhile it dawned on me what my problem was,
> and I have sorted it out. What I found was multiples of the known
> eigenvectors and didn't recognise them as such.

For complex eigenvectors with nondegenerate eigenvalues, there are
really two different kinds of ambiguity in the computation.  There
is an arbitrary normalization factor that scales the real and
complex parts of the vector, and there is an arbitrary complex phase
factor which mixes the real and complex parts together. When
comparing computed outputs with reference values, you must ensure
that the computed vectors use the same normalization and phase
conventions.  For example, you might choose to normalize the vectors
to Euclidian norm of 1 and to rotate the real and imaginary parts to
minimize the norm of the imaginary component.  There are many such
choices (probably an infinite number) that you could use.

When computing real eigenvectors with nondegenerate eigenvalues,
there is the normalization factor that scales each vector and a sign
factor of +1 or -1.

For degenerate eigenvalues, the set of associated eigenvectors can
be mixed arbitrarily among themselves.  This is another source of
problems when comparing computed outputs with reference values.  If
the computational algorithms are different, then you are almost
certain to compute different eigenvectors within these degenerate
subspaces, but also even with the same algorithm, different
implementations may compute intermediate results in a slightly
different order, and due to these small roundoff error differences,
completely different vectors can be computed.  Something as simple
as switching BLAS libraries can result in what looks like completely
different eigenvectors.

$.02 -Ron Shepard   0 Reply ron-shepard (1197) 10/20/2011 3:16:16 PM 4 Replies 59 Views Similar Articles 12/13/2013 8:15:11 PM [PageSpeed]  Reply: Similar Artilces: Questions, Questions... Hi! I took the routine from Gastronor.mdb: In the frmOrderDetails -> ITemID_AfterUpdate: Dim strFilter as String strFilter = "ItemID = " & Me!ItemID but it didn't work on win 2000 with access 2000. However, after playing around for a while, I found: Me!UnitPrice = ItemID.Column(2) to work just fine, even after updating prices in the tblItems. Anybody knows why? The routine in win 98/win nt in the footer of frmOrderDetails: =Sum((Price*Amount*(1-[Discount])/100)*100)) didn't either work. After testing and trying I finally discovered that =Sum(Price*Amount*Discount)... Questions, Questions, Questions -- Well, really only two This newsgroup seemed distribution non-specific so here goes: 1. Many companies who want to use linux, want it for email and web service. These systems don't need xwindows, and, in fact, as I understand it, not installing it not only saves resources but reduces security risk. Is there anywhere freely available a how-to install various current distributions of linux using text only, and installing mail, web and domain name service that is fairly easy for the uninformed to follow? Especially helpful would be a debian install. 2. Does anyone know of an encyclopedia, preferrably text-b... Question and bonus question Given an integer between 0 and 5000 (characters 0-9; no leading zero) can anyone suggest a non-iterative, non-recursive way of summing the individual digits? I'm just looking for a clever way to do what is easy with the obvious loop: Sum = 0 Do I = 1 to length(n) Sum = sum + substr(n,I,1) End Bonus Question: Can anyone suggest a mechanism to allow reading the TOC (Table of Content) of an Audio CD? The connection? The TOC contains frame numbers where each track starts. Do an integer divide by 75 and you get the time offset in seconds. Sum the digits of each of the start times... design question Hello all - I'm working on a database that stores data for a survey. We are storing several different types of surveys, each of which can have any number of different question types. For instance, a survey could have a multiple choice question, a yes/no question, numerical question, or a free-form text answer. When a user logs on to fill out a survey, the data from the survey template will be copied to a set of tables which look basically the same, except that they will store the survey questions and the answers from the user. Each question type would have a different field or fields fo... All the interesting questions are software questions A presentation attributed to Intel was posted in another group: http://www.beyond3d.com/content/articles/31 While the presentation is framed as a CPU-GPU confrontation, it really just presents as policy some trends that could have been predicted quite some time ago: 1. More simple (in-order cores) get you more bang for the buck than do mainstream OoO cores. 2. They also get you more bang per watt. 3. GPU's are already infringing on CPU territory and the trend is bound to continue. 4. Some problems are better suited to a GPGPU (general purpose GPU) than others. 5. The big question is... solve question/simplify question Hi, I have a list called equns of linear equations which are independent of each other. The according variables are in a list called params. I solve the equations into a list called sol using sol = Solve[equns, params]; Can I expect the solution procedure to be faster or in favor of memory usage if I iterate through the elements of the lists and solve for each equation separately? Same question goes for Simplify. I don't want to try and find out myself because I have noticed big differences in running times with Mathematica and I am not sure how to test for improved... Question I have a history that I have written in MS Word. It is composed of text and images. Is Pagemaker the best program to use for setting up the pages images etc and if so how do I import a MS document into Pagemaker? Thanks folks, I know the questions are probably rather dumb but...... ;-) Wolf wrote: > I have a history that I have written in MS Word. > It is composed of text and images. > Is Pagemaker the best program to use for setting up the pages images etc and > if so how do I import a MS document into Pagemaker? > > Thanks folks, I know the questions are probably... Question Can anyone define the difference between ITS vs. simulation? Anne L. Michaels almichaels@dynasims.com "simarts" <almichaels@crashzoom.com> wrote in news:1125937604.488797.222670@g14g2000cwa.googlegroups.com: > Can anyone define the difference between ITS vs. simulation? Probably, if you'd care to say what "ITS" means. All the best, John. On Mon, 05 Sep 2005 15:06:43 -0400, "simarts" <almichaels@crashzoom.com> wrote: >Can anyone define the difference between ITS vs. simulation? What is ITS?... A.L. A.L. <alewando@alamakota.co... Question Hi, Can someone tell me why this little piece of php/mysql code fails?$row = $result->fetch_assoc(); if ($clog) fwrite($logfile,"row['nametag'] {$row['nametag']}\n"); if ($row['nametag'] != 0) { if ($clog) fwrite($logfile,"name {$row['nametag']}\n"); } I've got this logfile where I am putting debug output. In the above, the first fwrite works -- it writes the name contained in the row value 'nametag'...

Question
I have this statement: DELETE FROM TMP_TABLE_MERGE WHERE EXISTS (SELECT 1 FROM TMP_TABLE_1 T1 WHERE TMP_TABLE_MERGE.OBJECT_ID = T1.OBJECT_ID) In both the table TMP_TABLE_MERGE and TMP_TABLE_1 the field OBJECT_ID is the primary key but in the Explain Plan the table TMP_TABLE_MERGE is acessed in full scan mode. Do someone knows how can I force the use of the primary keys? Thanks Pier On Sat, 4 Jan 2003 14:20:33 +0100, "Pier Gaetano Novara" <pgaetanoNoSpam@csc.com> wrote: >I have this statement: >DELETE FROM > TMP_TABLE_MERGE >WHERE EXIS...

Question
Hello, I have just subscribed to this group today and am reading some of the posts. I've tried looking for the answer to this question but can't seem to find it with any of the search functions. I'm looking for a game new game to play and I don't play many nor do I really keep up with trends so I'm looking for advice and suggestions. My favorite games are rpgs with turn based combat and extensive dialouge trees. My favorite game ever was Planescape: Torment though I had fun with Neverwinter Nights, mostly because of some of the downloaded modules for it. Genre is not imp...

question
Hello, I would like to know what EDIF and SDF are? Thanks, Neil They are my cousins from the side of the family that we don't talk about much. Thy don't spll wll ether. Did you try a web search? www.EDIF.org SDF = standard delay format Regards, Jim Neil Zanella wrote: > Hello, > > I would like to know what EDIF and SDF are? > > Thanks, > > Neil > -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Jim Lewis Director of Training mailto:Jim@SynthWorks.com SynthWorks Design Inc. http://www.SynthWorks.com 1-503-59...

Question
Hi all I have a very naive question.Is there any analogy between manchester encoding and scrambling using PN sequence. Thanks sanjana ...

Question
I want to know if there are companies than bring support of PostgreSQL. __________________________________ Do you Yahoo!? Yahoo! SiteBuilder - Free, easy-to-use web site design software http://sitebuilder.yahoo.com ---------------------------(end of broadcast)--------------------------- TIP 4: Don't 'kill -9' the postmaster Manuel Haroldo writes: > I want to know if there are companies than bring > support of PostgreSQL. Searching for "PostgreSQL" and "support" on Google will show you hundreds of them. -- Peter Eisentraut peter_e@gmx.net ------...

ipfw question & portscan question
Using BrickHouse to set ipfw so that only port 80 is open (I serve several Web sites), I note lots of the following sort of stuff in console.log and system.log: Jan 27 06:17:34 Stella-Blue kernel: ipfw: 55005 Deny UDP 64.122.147.30:137 216.243.177.155:137 in via en0 Jan 27 06:25:29 Stella-Blue kernel: ipfw: 55005 Deny ICMP:8.0 216.243.190.220 216.243.177.155 in via en0 Jan 27 06:56:48 Stella-Blue kernel: ipfw: 55005 Deny UDP 210.13.22.79:1160 216.243.177.155:1434 in via en0 Jan 27 07:06:11 Stella-Blue kernel: ipfw: 55005 Deny ICMP:8.0 216.244.14.36 216.243.177.155 in via en0 Jan 27 07:09:...

Two related question. Question 1
I have two related question. Let me introduce a pure function f = #1^2 + #2 & Now. I want to make an operation over the function, for example to find its square and to call the result (the expected function f = (#1^2 + #2)^2 & ) c: c=f^2 However, I do not obtain this, as c[a,b] does not evaluate to (a+b)^2. Can anybody advise me how to obtain such a function without long substitutions. I would like to obtain something which is made for derivatives : In[11]:= Derivative[1][f] Out[11]= 2 #1& In[12]:= Derivative[2][f] Out[12]= 2& Regards ...

factoring book question and cast question.
What is recommended as the best practical factoring and coding in C++ book? (...if exists...) related topic: If I have well factored and designed C++ code in a fairly complex system, where if any would dynamic_cast<> statements be? Should dynamic_cast<> be the only really proper type of cast ideally that would exist?? Thanks a lot! todd. -- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] todma wrote: > What is recommended as the best practical factoring and coding in C++ > book? (...i...

64HDD setup questions (and RS232 question)
Yesterday I got fired from a real lousy job so had time to reformat my 64hdd setup. Originally it was a msdos box and worked fine. Then I formatted it to get rid of all the win 3.1 crap. I used freedos .9 version I think. 64hdd wouldn't boot right. I used an msdos 6.2 boot disk from then on. I wasn't sure if the msdos boot disk could install dos. Is it that 6.2 can only 'sys' FAT32? So I made two boot disks. One with msdos 5.0 and one with msdos 6.2. Then I formatted the hd from a FreeDos 1.0 boot cd and installed that os. (Is there an easier way to do this?) When I booted...

[ntp:questions] PPSkit patch questions
There are two different linux kernel patches for PPS support. There is a standalone patch file for 2.4.20: ftp://ftp.kernel.org/pub/linux/daemons/ntp/PPS/PPS-patch-2.4.20.gz There is also the PPSkit that contains the file patch-2.4.20: ftp://ftp.kernel.org/pub/linux/daemons/ntp/PPS/PPSkit-2.1.1.tar.gz These two patches are very similar but have differences. What's the deal with these two patches? Which is the correct one for use for locking a machine to a PPS source with NTP? jeff Jeff Mock <jeff-ml@mock.com> writes: > There are two different linux kernel patches fo...

exec question and a windows path question
Does exec wait for another exec to finish? If I have two exec statements: exec /path/to/app exec /path/to/app Does the second one fire off after the first one completes? Also on Windows are the pathnames valid: c:\\path\\to\dir c:/path/to/dir Thanks! Robert sigzero@gmail.com wrote: > Does exec wait for another exec to finish? If I have two exec > statements: > > exec /path/to/app > exec /path/to/app > > Does the second one fire off after the first one completes? If you add an ampersand (&) to run these commands in the background, like this-- exec /path/to/a...