[ibis-interconn] Re: [Sparse Matrix Mapping] Writeup Comments

  • From: <radek_biernacki@xxxxxxxxxxx>
  • To: <ibis-interconn@xxxxxxxxxxxxx>
  • Date: Tue, 17 Nov 2009 20:53:03 -0700

Hi All,

I have the following comments on the Draft 5 revision.

1. The complex value .999 .999 at the top of page 4 (top) is a good example of 
the problem with the current rules. We need to provide meaningless data that 
are to be ignored just to satisfy our a bit ill-conceived rules. We need to 
sort out the issue of [Number Of Sparse Labels], empty labels and the rigid 
format of the labels: 1: ... N:

2. Terminology:

Rows and columns (with a reference to the original matrix)  - these terms 
actually refer to the row and column indices and should be spelled out as "row 
index" and "column index", respectively.

Columns (with a reference to the network data) - should be avoided as the 
network data is not a matrix, nor it has to be arranged in columns

Larger matrix, full matrix - should be avoided - there is only one matrix

3. I would like to see a keyword like [End Of Sparse Matrix Mapping].

Radek

-----Original Message-----
From: ibis-interconn-bounce@xxxxxxxxxxxxx 
[mailto:ibis-interconn-bounce@xxxxxxxxxxxxx] On Behalf Of Bob Ross
Sent: Tuesday, November 17, 2009 4:08 PM
To: IBIS-Interconnect
Subject: [ibis-interconn] [Sparse Matrix Mapping] Writeup Comments

All:

As requested at the last meeting here is a top level review
of the Sparce Matrix Proposal.  I am assuming Draft 5 as the
reference, and some issues are listed first.  The editing then
will follow.

I agreement with nearly all of the technical content, but a few
details need to be agreed upon, and the list below needs to
be considered for editorial changes.

------------

Specific comments/issues to be resolveed only on the Draft5:

1. Nomenclature: Prefer changing notation to lower case hyphenated
    words instead of capitalized two word names:

    Integer Labels --> integer-labels
    Index Pairs    --> index-pairs

    Data Pair      --> data-pair  (actually not fully defined elsewhere)
    matrix pair    --> data-pair

                        does data-pair need to be defined?

2. Resolve whether empty integer-labels are permitted

    (this impact editing several rules and the examples.)
    (I am not sure we agreed on "No".)

3. Do we need [Number of Sparse Labels] ?

    Technically No, but for redundancy, cross checking, Yes.

    Must it appear before [Sparse Matrix Mapping]?

    No, can be an after-the-fact check with the maximum
    integer-label value.

4. Is there a better way to indicate "zero-valued" as in

"Data in the full matrix that is not indicated by a matrix
pair under [Sparse Matrix Mapping] is assumed zero-valued."

(in DB format 0 0 (zero-valued) means the magnitude in MA is "1"
and phase is 0 degrees.)

5. The formatting of the fundamental definitions (syntactical
definition and then brief description).  Put other rules elsewhere,
and we did define "n" earlier:

integer-label:  An integer between 1 and less than or equal to n^2
                 and without any space by a colon character.  For example: '5:'.
                 Integer-labels refer to the data-pair column under the
                 [Network Data] keyword where data-pairs are entered for
                 each frequency.

index-pair:     A pair of positive integers between 1 and n, separated
                 without any space by the comma character and enclosed
                 without space by the open-parenthesis and closed-parenthesis
                 characters.  For example: '(3,5)'.  Index-pairs refers
                 to the n-port row,column position where the specified data-pair
                 value exists in the fully populated n-port without the [Sparse
                 Matrix Mapping] keyword.

Rules for integer-labels and index-pairs (separate heading or just
the following paragraphs):

  sequencing, white-space, must be followed by one or more index-pair,
  maximum value of index-pair, cannot be repeated, highest index-pair
  match [Number of Sparse Labels], multiple lines, multiple labels on
  same line, etc.

This format matches better the format in the rest of Touchstone V.2

6.  More discussion regarding the special interpretation of
[Matrix Format] Full | Upper | Lower

   Position of [Matrix Format] before [Sparse Matrix Mapping]?  - Yes


-----------------------------------------

Bigger issues to be resolve later when this writeup is merged into
Touchstone 2.0 - analogous to a BIRD and then its merger and interaction
into the main document later.

1. Where in document (after what keyword)  or a new chapter?

2. Addition to [Matrix Format] number of entries rule.

3. Example re-numbering and Example format.

4. Other interacitons - e.g., page 3. for list of keywords.

5. Designation/distinction of Touchstone 2.0 2.1 or just 2.X for
all versions and growth.  Or does this become Touchstone 3.0?

------------------

Bob

-- 
Bob Ross
Teraspeed Consulting Group LLC     Teraspeed Labs
121 North River Drive              13610 SW Harness Lane
Narragansett, RI 02882             Beaverton, OR 97008
401-284-1827                       503-430-1065
http://www.teraspeed.com           503-246-8048 Direct
bob@xxxxxxxxxxxxx

Teraspeed is a registered service mark of Teraspeed Consulting Group LLC

------------------------------------------------------------------
      The IBIS Ad Hoc Interconnect Task Group Mailing List

Archives are available at:     
                //www.freelists.org/archives/ibis-interconn

TO UNSUBSCRIBE:
        Send a message to "ibis-interconn-request@xxxxxxxxxxxxx" 
        with a subject of "unsubscribe"

To administer your subscription status from the web, visit:
                //www.freelists.org/list/ibis-interconn





------------------------------------------------------------------
      The IBIS Ad Hoc Interconnect Task Group Mailing List

Archives are available at:
                //www.freelists.org/archives/ibis-interconn

TO UNSUBSCRIBE:
        Send a message to "ibis-interconn-request@xxxxxxxxxxxxx"
        with a subject of "unsubscribe"

To administer your subscription status from the web, visit:
                //www.freelists.org/list/ibis-interconn



Other related posts: