[maybee]

hi,

I m currently have to make some matrix calculations and therefore created a matrix like

Matrix<scalarField,scalar> D(2,1);
D[1][1] = pow(phase.Size()[celli],0); //addressing of first row first element?

The code compiles, but I don t think I am addressing the first element in the first row.

Simple question: How can I address the different elements of the matrix? I ve already searched the forum, but could not find the answer.

Additionally: How can I do a element/element multiplication with two matrices? Example:

[A1 A2 A3] * [B1 B2 B3] = [A1B1 A2B2 A3B3]

[chrisb2244]

To answer the second question, what you have there is the dot product, right?

Openfoam implements the dot product of two vectors using

 VectorA & VectorB

In terms of your first question, I imagine what you want is D [0] [0] but I haven't tried explicitly using matrices in OF, so give me a minute to find relevant documentation.

However, your `pow' call is going to return 1 every time - you have x^0.

[maybee]

hi,

first of all thx. I really need to know which mathematical operations are available for objects of class "matrix" . Is there anywhere an overview? I already installed "armadillo", because I could not do

Term[celli]= -(D*Prod.T()).value();

where "D" and "Prod" are "Matrix<Field<double>,double>" and "T()" should return the transpose of "Prod".

and got:

/home/USER/SolverNEW/NEW/NEW_TransportEqns.H:141: error:
no match for ‘operator*’ in ‘D * Foam::Matrix<Form, Type>::T() const [with Form = Foam::Field<double>, Type = double]()’

I really need to know which matrice methods OpenFoam provides otherwise I ll have to use armadillo.

EDIT: I know about the pow returning always 1 - I implemented several similar functions which differ in the exponent - I ll probably change this one later for saving a little bit time when running the program.

EDIT2: I also tried now

Prod = SF & w;

where Prod,SF and w are of type "Matrix<Field<double>,double>"

only getting:

/home/USER/SolverNEW/NEW/NEW_TransportEqns.H:137: error: no match for ‘operator&’ in ‘SF & w’

greetings maybee

[chrisb2244]

Dear Maybee,

Firstly, sorry for the problems you're having with multiplication - I have checked the documentation more closely and noticed that the dot product is (unsurprisingly) not defined for matrices - only vectors. Hence

 A & B

will work for and , as you asked, but not for matrices more generally. I have edited my answer earlier to reflect that.

The documentation for OpenFOAM is available online at
http://www.openfoam.org/docs/cpp/

There you will see a search bar in the top right - you can use this to search for functions, or classes, etc. In your case, you would want the class Matrix. (Classes are displayed in the dropdown list as eg

  Matrix  Foam.

)

I'm unsure what acceptable "Form" is for a matrix - if you or someone else can give a link to documentation explaining this, or alternatively, explain this, that'd be really useful I imagine. My initial guess is that Form describes eg SquareMatrix, SymmetricSquareMatrix, RectangularMatrix etc.

As a result of this, part of me feels like you're getting undefined operators because your template substitution is messing you up. If you type

Matrix<Banana<double>, double> D(2,1)

Does that fail in a way that

Matrix<scalarField, scalar> D(2,1)

does not?

I ask because GeometricField<scalar> (== scalarField via typeDef) is not listed as a Form as far as I can see, so I'm wondering if that is because I can't find a suitable list of "Form"s, or if it is not a valid construction.

Best,
Christian

[maybee]

Hi,

I ve already used the FOAM Docs to search for the matrix class:

http://www.openfoam.org/docs/cpp/

But there is not much documentation about mathematical methods that can be used on matrices!

I also could not find any list for valid "Forms", only:

http://www.cfd-online.com/Forums/ope...m-1-6-ext.html

I'm unsure what acceptable "Form" is for a matrix - if you or someone else can give a link to documentation explaining this, or alternatively, explain this, that'd be really useful I imagine. My initial guess is that Form describes eg SquareMatrix, SymmetricSquareMatrix, RectangularMatrix etc.

As a result of this, part of me feels like you're getting undefined operators because your template substitution is messing you up. If you type

Code:
Matrix<Banana<double>, double> D(2,1)
Does that fail in a way that
Code:
Matrix<scalarField, scalar> D(2,1)
does not?

I ask because GeometricField<scalar> (== scalarField via typeDef) is not listed as a Form as far as I can see, so I'm wondering if that is because I can't find a suitable list of "Form"s, or if it is not a valid construction.

Same over here . Initially I had the same thoughts. I also tried "int" for the "Form" which does compile fine.
I ll use armadillo now, perhaps OpenFOAM s Matrix class is not finished yet (?)....

greetings maybee

[jherb]

I guess this might help:
http://www.foamcfd.org/Nabla/guides/...sGuidese4.html

[maybee]

nice, but still I ll use armadillo since I am working with matrices and not tensors.

[l_r_mcglashan]

Of course the Matrix class works. It's not there for nothing.

Originally Posted by maybee

hi,

I m currently have to make some matrix calculations and therefore created a matrix like

Matrix<scalarField,scalar> D(2,1);
D[1][1] = pow(phase.Size()[celli],0); //addressing of first row first element?

What do you expect to happen with the first template parameter as 'scalarField'?

What is the form of you matrix? Rectangular/Square/Diagonal/Symmetric? You would then do:

Matrix<RectangularMatrix<scalar>, scalar> D(2,1);

or shorthand (see scalarMatrices.H):

scalarRectangularMatrix D(2,1);

Most standard matrix operations are available.

[maybee]

Ok,

I missread the translation of "rectangular" when translating it :/ .
Is there anywhere a list/reference of the matrix operations in OpenFOAM?

Furthermore is there defined a method for this operation

A * B = AB
[A1 A2 A3] * [B1 B2 B3] = [A1B1 A2B2 A3B3]

Note, that I just used * here for the unknown method/operator.

[purnp2]

for future help...
For [A1 A2 A3] * [B1 B2 B3] = [A1B1 A2B2 A3B3], use scale(A,B).
See reference in:
http://foam.sourceforge.net/docs/Gui...mmersGuide.pdf

[mrishi]

Originally Posted by purnp2

for future help...
For [A1 A2 A3] * [B1 B2 B3] = [A1B1 A2B2 A3B3], use scale(A,B).
See reference in:
http://foam.sourceforge.net/docs/Gui...mmersGuide.pdf

For future reference,

The scale(a,b) function has been replaced by cmptMultiply(a,b).

See here:
scale(a, b) function