printMatrix() in util.R?

[friendly]

I had forgotten about printMatrix() in util.R.
Should it be documented together with matrix2latex?

[philchalmers]

The defaults provide different outputs.

A <- matrix(1:12, 3, 4) / 6
printMatrix(A, fractions=TRUE)
     [,1] [,2] [,3] [,4]
[1,]  1/6  2/3  7/6  5/3
[2,]  1/3  5/6  4/3 11/6
[3,]  1/2    1  3/2    2

matrix2latex(A, fractions = TRUE)
\left[
 \begin{array}{llll}
  1/6 & 2/3 & 7/6 & 5/3 \\ 
  1/3 & 5/6 & 4/3 & 11/6 \\ 
  1/2 & 1 & 3/2 & 2 \\ 
  \end{array} \right]

however they are special cases of each other. printMatrix(A, latex=TRUE) is just a wrapper to matrix2latex, while the latter could be modified to reproduce the default behaviour of printMatrix() via matrix2latex(A, latex = FALSE) ...

I'm tempted to just use printMatrix() to do the latex presentation as it has a better name (and not export matrix2latex) and then set printMatrix(A, latex=TRUE) as the default.

[john-d-fox]

Hi Michael, A related question is whether matrix2latex() and symbolicMatrix() should be combined. E.g., symbolicMatrix() handles the examples in ?matrix2latex, though not all of the arguments (fractions, show.size, digits). Best, John

…

On 2024-08-07 9:47 a.m., Michael Friendly wrote: Caution: External email. I had forgotten about |printMatrix()| in |util.R|. Should it be documented together with |matrix2latex|? — Reply to this email directly, view it on GitHub <https://github.com/ friendly/matlib#51>, or unsubscribe <https://github.com/ notifications/unsubscribe-auth/ ADLSAQVV4MNMNXPINMKUS7LZQIQOTAVCNFSM6AAAAABMEOWN4OVHI2DSMVQWIX3LMV43ASLTON2WKOZSGQ2TGNJTGQ2DENA>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[friendly]

Hi John
Yes, I'm seeing that slight disconnect in my examples. The questions are:

• how to combine them (which to use as the base to add the other's functionality)?
• what name to use?

Maybe these should evolve into printMatrix()?

There is also a quite different approach sketched in dev/as_latex.Rmd that uses knitr print methods.

[friendly]

@philchalmers I think there is a function somewhere that can determine if a function is called from the console or inside a knitr document. There are also knitr::is_latex_output(), knitr::is_html_output()

These could help to determine what the appropriate default behaviors should be.

[john-d-fox]

Hi,

On 2024-08-07 10:28 a.m., Michael Friendly wrote: Caution: External email. Hi John Yes, I'm seeing that slight disconnect in my examples. The questions are: * how to combine them (which to use as the base to add the other's functionality)?

I think that the simplest thing to do is to add the missing arguments to symbolicMatrix(). Unless you see a reason not to, I'll do that -- it's harmless even if the functions aren't consolidated.

* what name to use? Maybe these should evolve into |printMatrix()|?

I haven't thought about it much, but that makes sense to me. printMatrix() could call an unexported, augmented, symbolicMathrix() if latex=TRUE). There is an issue of backwards compatibility if we get rid of functions that previously were in the package, but I suspect that there's not a whole of lot of use of these functions. We could deprecate one or more functions.

There is also a quite different approach sketched in |dev/as_latex.Rmd| that uses |knitr| print methods.

Interesting. I was thinking last night about how to implement something like that. It's probably a better approach if one is committed to using knitr. Implementing it generally for LaTeX equations would probably require quite a bit of work. Best, John

…

— Reply to this email directly, view it on GitHub <https://github.com/ friendly/matlib#51#issuecomment-2273613769>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ ADLSAQTQRZHHKASZSLM53NDZQIVJTAVCNFSM6AAAAABMEOWN4OVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENZTGYYTGNZWHE>. You are receiving this because you commented.Message ID: <friendly/ ***@***.***>

[john-d-fox]

Hi, I added show.size, digits, and fractions args to symbolicMatrix(). The implementation is a bit different from that in matrix2latex(). For example, \frac{}{} is used for rational fractions (with the interline spacing adjusted in the matrix). I did fairly extensive testing, but it would probably help if you took a look too. Best, John

…

On 2024-08-07 10:28 a.m., Michael Friendly wrote: Caution: External email. Hi John Yes, I'm seeing that slight disconnect in my examples. The questions are: * how to combine them (which to use as the base to add the other's functionality)? * what name to use? Maybe these should evolve into |printMatrix()|? There is also a quite different approach sketched in |dev/as_latex.Rmd| that uses |knitr| print methods. — Reply to this email directly, view it on GitHub <https://github.com/ friendly/matlib#51#issuecomment-2273613769>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ ADLSAQTQRZHHKASZSLM53NDZQIVJTAVCNFSM6AAAAABMEOWN4OVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENZTGYYTGNZWHE>. You are receiving this because you commented.Message ID: <friendly/ ***@***.***>

[philchalmers]

Hi John,

I was trying to test out Michaels issue in #52, but the new symbolicMatrix() fails with this input structure:

symbolicMatrix(diag(3))
## Error in format(m, digits = digits) : object 'm' not found

[philchalmers]

Continuing with the thread above, I think the main issue we're running into is which generator/printer function to use as they all serve a dual purpose. symbolicMatrix() seems to be the winner at this point in terms of the constructor/coersion function, though it's dual purpose of both constructing the symbolic matrix and printing/cat-ing the output is more difficult to track.

It may be best to switch the result from symbolicMatrix() to return the object information only (the print=FALSE behaviour), and assign it an S3 object like class sMatrix. That way we can build a suitable print() generic which could be used to cat() the output and detect the suitable output to build (@friendly's point about detecting the output type for compilation). After that I'm in favour of deprecating printMatrix() and matrix2latex() as they should be obsolete.

[john-d-fox]

Hi Phil,

On 2024-08-07 1:22 p.m., Phil Chalmers wrote: Caution: External email. Continuing with the thread above, I think the main issue we're running into at this point is which generator/printer function to use as they all serve a dual purpose. |symbolicMatrix()| seems to be the winner at this point in terms of the constructor/coersion function, though it's dual purpose of both constructing the symbolic matrix and printing/cat- ing the output is more difficult to track. It may be best to switch the result from |symbolicMatrix()| to return the object information only (the |print=FALSE| behaviour), and assign it an |S3| object like class |sMatrix|. That way we can build a suitable | print()| generic which could be used to |cat()| the output and detect the suitable output to build ***@***.*** <https://github.com/friendly>'s point about detecting the output type for compilation). After that I'm in favour of deprecating |printMatrix()| and |matrix2latex()| as they should be obsolete.

I agree that this is a more elegant, flexible, and sounder approach. I'm in favour of it. BTW, I inadvertently sent my previous message only to you, so for Michael, the bug Phil identified is now fixed. Best, John

…

— Reply to this email directly, view it on GitHub <https://github.com/ friendly/matlib#51#issuecomment-2273960348>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ ADLSAQXHWUNUTQ7ZVSZCFZDZQJJVJAVCNFSM6AAAAABMEOWN4OVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENZTHE3DAMZUHA>. You are receiving this because you commented.Message ID: <friendly/ ***@***.***>

[john-d-fox]

Hi again Phil, symbolicMatrix() now returns an object of class "symbolicMatrix", with a new print.symbolicMatrix() method, and no longer has a print argument. The print() method is very elementary, cat()ing to the console if the arg onConsole=TRUE, which is currently the default. As per your suggestion, the print() method could be made more sophisticated, but I think that you're in a better position to know what to do with it than I am. Best, John

…

On 2024-08-07 1:22 p.m., Phil Chalmers wrote: Caution: External email. Continuing with the thread above, I think the main issue we're running into at this point is which generator/printer function to use as they all serve a dual purpose. |symbolicMatrix()| seems to be the winner at this point in terms of the constructor/coersion function, though it's dual purpose of both constructing the symbolic matrix and printing/cat- ing the output is more difficult to track. It may be best to switch the result from |symbolicMatrix()| to return the object information only (the |print=FALSE| behaviour), and assign it an |S3| object like class |sMatrix|. That way we can build a suitable | print()| generic which could be used to |cat()| the output and detect the suitable output to build ***@***.*** <https://github.com/friendly>'s point about detecting the output type for compilation). After that I'm in favour of deprecating |printMatrix()| and |matrix2latex()| as they should be obsolete. — Reply to this email directly, view it on GitHub <https://github.com/ friendly/matlib#51#issuecomment-2273960348>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ ADLSAQXHWUNUTQ7ZVSZCFZDZQJJVJAVCNFSM6AAAAABMEOWN4OVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENZTHE3DAMZUHA>. You are receiving this because you commented.Message ID: <friendly/ ***@***.***>

[philchalmers]

Perfect! Thanks so much for making these changes. I see no need to keep the other constructor functions at this stage, though rather than breaking Michaels workflow will leave them in. I'll update Eqn() to work with this new object-oriented approach.

[friendly]

Wow! Thank you guys for taking this forward. This has been diversionary for me, but it's been so much fun...

[john-d-fox]

Hi Michael,

--

On 2024-08-07 3:36 p.m., Michael Friendly wrote: Caution: External email. Wow! Thank you guys for taking this forward. This has been diversionary for me, but it's been so much fun...

Diversionary for me too, though probably in a different sense. FWIW, I learned some more about \boldsymbol{} vs. \mathbf{}. It turns out that the latter works for uppercase but not lowercase Greek letters, while the former works for both. It seems that l.c. Greek letters are considered italic, which \mathbf{} doesn't bold. So both \mathbf{\Lamba} and \boldsymbol{\Lambda} work but \mathbf{\lambda} doesn't. OTOH, \boldsymbol{} applied to Latin letters renders them as bold italic, which generally looks awful. Best, John

…

— Reply to this email directly, view it on GitHub <https://github.com/ friendly/matlib#51#issuecomment-2274209817>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ ADLSAQSPG3TXHEGM32KQ5J3ZQJZK3AVCNFSM6AAAAABMEOWN4OVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENZUGIYDSOBRG4>. You are receiving this because you commented.Message ID: <friendly/ ***@***.***>

[john-d-fox]

Hi, Once we sort out the other functions, I suggest that we uniformly rename symbolicMatrix() to something like latexMatrix(), which better reflects what it actually does now. Best, John

…

On 2024-08-07 2:55 p.m., Phil Chalmers wrote: Caution: External email. Perfect! Thanks so much for making these changes. I see no need to keep the other constructor functions at this stage, though rather than breaking Michaels workflow will leave them in. I'll update |Eqn()| to work with this new object-oriented approach. — Reply to this email directly, view it on GitHub <https://github.com/ friendly/matlib#51#issuecomment-2274139991>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ ADLSAQXJWDZRFRX46VNGDTLZQJUQ7AVCNFSM6AAAAABMEOWN4OVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENZUGEZTSOJZGE>. You are receiving this because you commented.Message ID: <friendly/ ***@***.***>