Take the 2-minute tour ×
Mathematica Stack Exchange is a question and answer site for users of Mathematica. It's 100% free, no registration required.

I have come up with some BlockMatrix Algebra for Mathematica to make notations easier.

I have the following:

Needs["Notation`"];
Notation[ParsedBoxWrapper[
RowBox[{"A_", "\[CenterDot]", "B_"}]] \[DoubleLongLeftRightArrow] 
   ParsedBoxWrapper[
RowBox[{"BlockMultiply", "[", 
RowBox[{"A_", ",", "B_"}], "]"}]]];
Notation[ParsedBoxWrapper[
SuperscriptBox["A_", "Inv"]] \[DoubleLongLeftRightArrow] 
   ParsedBoxWrapper[
RowBox[{"BlockInv", "[", "A_", "]"}]]];
Notation[ParsedBoxWrapper[
SuperscriptBox[
RowBox[{"(", "A_", ")"}], "T"]] \[DoubleLongLeftRightArrow] 
   ParsedBoxWrapper[
RowBox[{"BlockTransp", "[", "A_", "]"}]]];

And the actual matrix algebra functions:

ClearAll[BlockTransp]
(*SetAttributes[BlockTransp,OneIdentity]*)

BlockTransp[M_List] := Map[BlockTransp[#] &, Transpose[M], {2}] 
BlockTransp[M__ + B__ ] := BlockTransp[M] + BlockTransp[B]
BlockTransp[a_ /; NumberQ[a]] := a;
BlockTransp[a_Scalar ] := a;
BlockTransp[a_   M__ /; NumberQ[a]] := a BlockTransp[M];
BlockTransp[a_Scalar   M__] := a BlockTransp[M];

BlockTransp[BlockMultiply[x_, y_] ] := 
 BlockMultiply[BlockTransp[y], BlockTransp[x]]
BlockTransp[BlockTransp[x_]] := x

ClearAll[BlockMultiply]
BlockMultiply[mats1 : {{_ ..} ..}, mats2 : {{_ ..} ..}] := 
 Inner[BlockMultiply, mats1, mats2]
(*For numbers*)

BlockMultiply[a_ A_ /; NumberQ[a], B_] := a BlockMultiply[A, B]
BlockMultiply[A_ , a_ B_ /; NumberQ[a]] := a BlockMultiply[A, B]
BlockMultiply[A__, a_ /; NumberQ[a]] := a A
BlockMultiply[a_ /; NumberQ[a], A__ ] := a A
(*For scalar symbols*)

BlockMultiply[a_Scalar A_, B_] := a BlockMultiply[A, B]
BlockMultiply[A_ , a_Scalar B_ ] := a BlockMultiply[A, B]
BlockMultiply[A__, a_Scalar ] := a A
BlockMultiply[a_Scalar, A__ ] := a A



ClearAll[BlockInv]
BlockInv[mats1 : {{_ ..} ..}] := Map[BlockInv[#] &, mats1, {2}];
BlockInv[ a_ /; NumberQ[a]] := a

(*Make some formatting expression*)
Scalar /: MakeBoxes[Scalar[a_], StandardForm] := 
  MakeBoxes[a, StandardForm];

I can do now some funny cool thing like:
1. Example:
First Example
2. Example:
Second Example
3. Example:
Third Example
4. Example:
Fourth Example

Remark: Of course the BlockInverse Function is sensless in the sense that it does not really invert correctly. The input to BlockInverse should only be a diagonal block matrix! There is a general formula for block matrices Block Inversion, but it is not implemented here for the sake of simplicity.

I have two questions which I could not come up so far:

  1. How can I make the 3. example such that the scalar variable is treated like a scalar? Can I overload NumberQ[scalar]:= ... ? Or is there a better way? [SOLVED] See the above definitions with added _Scalar that matches any pattern with head Scalar. See example 4!

  2. How can I pretty print these things such that superfluous brackets are neglected? Is there some elegant way for the function definitions such that, associativity and non-commutativity is handled well?

share|improve this question
1  
Per your comment on there being no general formula for the inverse, that may not be true. –  rcollyer Nov 26 '13 at 18:02

1 Answer 1

The idea about associativity is to use InfixNotation:

InfixNotation[ParsedBoxWrapper["\[CenterDot]"], Mult]

Then the associativity can be defined as

Mult[A___, mats1 : {{_ ..} ..}, mats2 : {{_ ..} ..}, D___] := Mult[A, Inner[Mult, mats1, mats2], D]
Mult[A___, Mult[B__], D___] := Mult[A, B, D];
Mult[A_] := A;

This allow you to get rid of some of the unnecessary brackets

Getting rid of the the brackets under the inverse operator is a bit more tricky. I don't know good answer, so I reformulated this question here. However I used the following workaround:

All the matrices are assumed to be enclosed in Matrix[] function. This allows me to make them bold in the output using the following code:

Notation[ParsedBoxWrapper[TagBox[StyleBox["A_", Rule[FontWeight, Bold]], ""]] \[DoubleLongLeftArrow] ParsedBoxWrapper[RowBox[{"Matrix", "[", "A_", "]"}]]]

Then the inverse of single matrix variable can be defined as InvSimple:

Inv[Matrix[A_]] := InvSimple[Matrix[A]];

and two different Notations for Inv and InvSimple are used

Notation[ParsedBoxWrapper[SuperscriptBox[RowBox[{"(", "A_", ")"}], RowBox[{"-", "1"}]]]\[DoubleLongLeftArrow] ParsedBoxWrapper[RowBox[{"Inv", "[", "A_", "]"}]]]
Notation[ParsedBoxWrapper[SuperscriptBox["A_", RowBox[{"-", "1"}]]]\[DoubleLongLeftArrow] ParsedBoxWrapper[RowBox[{"InvSimple", "[", "A_", "]"}]]]

The full code I am using is here (This code does not fully cover all the options that are implemented in your code, still it might interest you)

Notation[ParsedBoxWrapper[StyleBox["I", Rule[FontWeight, Bold]]] \[DoubleLongLeftArrow] ParsedBoxWrapper["IMatrix"]]
Notation[ParsedBoxWrapper[TagBox[StyleBox["A_", Rule[FontWeight, Bold]], ""]]\[DoubleLongLeftArrow] ParsedBoxWrapper[RowBox[{"Matrix", "[", "A_", "]"}]]]
InfixNotation[ParsedBoxWrapper["\[CenterDot]"], Mult]
Notation[ParsedBoxWrapper[SuperscriptBox[RowBox[{"(", "A_", ")"}], RowBox[{"-", "1"}]]]\[DoubleLongLeftArrow] ParsedBoxWrapper[RowBox[{"Inv", "[", "A_", "]"}]]]
Notation[ParsedBoxWrapper[SuperscriptBox["A_", RowBox[{"-", "1"}]]]\[DoubleLongLeftArrow] ParsedBoxWrapper[RowBox[{"InvSimple", "[", "A_", "]"}]]]

ClearAll[Matrix, Mult, Inv]
(*Matrix*)
Matrix[0] := 0;
Matrix[a_ /; NumberQ2[a]] := a IMatrix;
(*Multiplication*)
Mult[A___, mats1 : {{_ ..} ..}, mats2 : {{_ ..} ..}, D___] :=  Mult[A, Inner[Mult, mats1, mats2], D]
Mult[A___, Mult[B__], D___] := Mult[A, B, D];
Mult[A_] := A;

Mult[A___, a_ B_ /; NumberQ[a], D___] := a Mult[A, B, D];
Mult[A___, a_ /; NumberQ[a], D___] := a Mult[A, D];
Mult[A___, X_, X_, D___] := Mult[A, X^2, D];
(*Inverse*)
Inv[Inv[A_]] := A;
Inv[Mult[A_, B__]] := Mult[Inv[B], Inv[A]];
Inv[Matrix[A_]] := InvSimple[Matrix[A]];
Inv[InvSimple[A_]] := A;
Inv[IMatrix] := IMatrix;
Inv[a_ B__ /; NumberQ[a]] := Inv[B]/a;
share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.