stzMatrex: Declarative Pattern Matching for Matrices in Softanza

Introduction

stzMatrex brings this recognition ability into Softanza as a pattern language for matrices. Instead of writing conditional checks or nested loops, you can now express a matrix condition semantically:

oMx = new stzMatrex("{shape(square) & property(symmetric)}")

? oMx.Match([
    [1, 2, 3],
    [2, 4, 5],
    [3, 5, 6]
])
#--> TRUE

In one expression, you’ve defined a concept — “a square symmetric matrix” — and tested it instantly. Let’s explore how this pattern engine turns matrix semantics into readable, declarative logic.

1. Matching Matrix Size

The simplest patterns describe the size of a matrix. A {Size(rxc)} pattern specifies an exact shape, while {Size(mxn)} accepts any matrix.

oMx = new stzMatrex("{Size(3x3)}")

aMatrix = [
    [1,2,3],
    [4,5,6],
    [7,8,9]
]

? oMx.Match(aMatrix)
#--> TRUE

If we test a wrong size, the match fails:

oMx = new stzMatrex("{size(2x2)}")

aMatrix = [
    [1,2,3],
    [4,5,6],
    [7,8,9]
]

? oMx.Match(aMatrix)
#--> FALSE

When you need a generic matcher for any dimension, the notation mxn works elegantly:

oMx = new stzMatrex("{size(mxn)}")

aMatrix1 = [[1,2],[3,4]]
aMatrix2 = [[1,2,3],[4,5,6],[7,8,9]]
aMatrix3 = [[1,2,3,4]]

? oMx.Match(aMatrix1)
#--> TRUE
? oMx.Match(aMatrix2)
#--> TRUE
? oMx.Match(aMatrix3)
#--> TRUE

These size patterns provide the foundation for structural reasoning in numerical workflows.

2. Recognizing Matrix Shape

Shape patterns let you distinguish between geometric configurations — square, rectangular, tall, wide, or vector-like.

oMx = new stzMatrex("{shape(square)}")

aSquare = [[1, 2], [3, 4]]
aRect = [[1, 2, 3], [4, 5, 6]]

? oMx.Match(aSquare)
#--> TRUE
? oMx.Match(aRect)
#--> FALSE

The same idea extends naturally to other shapes:

oMxTall = new stzMatrex("{shape(tall)}")
oMxWide = new stzMatrex("{shape(wide)}")

aTall = [[1], [2], [3], [4]]
aWide = [[1, 2, 3, 4]]

? oMxTall.Match(aTall)
#--> TRUE
? oMxWide.Match(aWide)
#--> TRUE

For directional structures, you can match row and column vectors explicitly:

oMxRow = new stzMatrex("{shape(row)}")
oMxCol = new stzMatrex("{shape(column)}")

aRow = [[1,2,3,4,5]]
aCol = [[1],[2],[3]]

? oMxRow.Match(aRow)
#--> TRUE
? oMxCol.Match(aCol)
#--> TRUE

These shape-based expressions clarify algorithms that depend on the matrix’s orientation.

3. Detecting Matrix Properties

Beyond shape, stzMatrex can detect mathematical properties directly — such as whether a matrix is identity, diagonal, symmetric, or triangular.

oMx = new stzMatrex("{property(identity)}")

aIdentity = [
    [1,0,0],
    [0,1,0],
    [0,0,1]
]

aNotIdentity = [
    [1,0,0],
    [0,2,0],
    [0,0,1]
]

? oMx.Match(aIdentity)
#--> TRUE
? oMx.Match(aNotIdentity)
#--> FALSE

Other intrinsic properties follow the same syntax:

oMx = new stzMatrex("{property(symmetric)}")

aSymmetric = [
    [1,2,3],
    [2,4,5],
    [3,5,6]
]

aAsymmetric = [
    [1,2,3],
    [4,5,6],
    [7,8,9]
]

? oMx.Match(aSymmetric)
#--> TRUE
? oMx.Match(aAsymmetric)
#--> FALSE

The same logic identifies diagonal, upper, lower, and zero matrices, allowing you to encode classification rules directly in patterns.

4. Applying Element Constraints

Sometimes the structure is less important than what the matrix contains. Element constraints let you specify acceptable numeric values or ranges.

oMx = new stzMatrex("{element(0..10)}")

aInRange = [
    [1,2,3],
    [4,5,6],
    [7,8,9]
]

aOutOfRange = [
    [1,2,15],
    [4,5,6],
    [7,8,9]
]

? oMx.Match(aInRange)
#--> TRUE
? oMx.Match(aOutOfRange)
#--> FALSE

You can also define discrete sets or single-value matrices:

oMx = new stzMatrex("{element({0;1})}")

aBinary = [
    [1,0,1],
    [0,1,0],
    [1,1,0]
]

? oMx.Match(aBinary)
#--> TRUE

These patterns make it easy to validate numerical domains across large data structures.

5. Combining Logical Conditions

Matrix conditions can combine through logical operators &, |, and @!. This allows composite structural and numeric rules.

oMx = new stzMatrex("{shape(square) & property(identity)}")

aId = [
    [1,0,0],
    [0,1,0],
    [0,0,1]
]

aSquare = [
    [1,2,3],
    [4,5,6],
    [7,8,9]
]

? oMx.Match(aId)
#--> TRUE
? oMx.Match(aSquare)
#--> FALSE

Alternations are equally clear:

oMx = new stzMatrex("{(size(2x2) | size(3x3)) & property(diagonal)}")

aDiag2x2 = [
    [1,0],
    [0,2]
]

aDiag3x3 = [
    [1,0,0],
    [0,2,0],
    [0,0,3]
]

? oMx.Match(aDiag2x2)
#--> TRUE
? oMx.Match(aDiag3x3)
#--> TRUE

This logical composition is a natural way to describe high-level matrix requirements.

6. Matching Sets of Matrices

When working with collections, stzMatrex can search, count, and extract all matching matrices.

oMx = new stzMatrex("{property(diagonal)}")

aMatrices = [
    [[1,0],[0,2]],
    [[1,2],[3,4]],
    [[5,0,0],[0,6,0],[0,0,7]],
    [[1,1],[1,1]]
]

? oMx.CountMatchingMatrices(:In = aMatrices)
#--> 2

To retrieve them:

? @@NL(oMx.MatchingMatricesIn(aMatrices))
#-->
'
[
    [[1,0],[0,2]],
    [[5,0,0],[0,6,0],[0,0,7]]
]
'

And to analyze a set quantitatively:

oMx = new stzMatrex("{size(2x2)}")

aMatrices = [
    [[1,2],[3,4]],
    [[5,6],[7,8]],
    [[1,2,3],[4,5,6]],
    [[9,10],[11,12]]
]

? @@NL(oMx.AnalyzeMatches(aMatrices))
#-->
'
[
    ["pattern","{size(2x2)}"],
    ["totalmatrices",4],
    ["matchingcount",3],
    ["nonmatchingcount",1],
    ["matchrate",0.75]
]
'

Such analytical feedback transforms stzMatrex into a compact matrix quality assessment tool.

7. Extracting Matched Components

After a successful match, you can inspect what properties were recognized.

oMx = new stzMatrex("{size(3x3) & property(symmetric)}")

aMatrix = [
    [1,2,3],
    [2,4,5],
    [3,5,6]
]

oMx.Match(aMatrix)

? @@NL(oMx.MatchedParts())
#-->
'
[
    ["Size",[3,3]],
    ["Matrix",[[1,2,3],[2,4,5],[3,5,6]]],
    ["Properties",["Square","Symmetric"]]
]
'

This structured introspection is ideal for automated documentation or for building explainable AI models that interpret numerical patterns.

8. Filtering and Selection

When analyzing heterogeneous datasets, you can filter entire lists by pattern match.

oMx = new stzMatrex("{shape(square)}")

aMatrices = [
    [[1,2],[3,4]],
    [[1,2,3],[4,5,6]],
    [[5,6,7],[8,9,10],[11,12,13]]
]

aFiltered = oMx.FilterMatrices(aMatrices)
? len(aFiltered)
#--> 2

Filtering based on structural rules reduces preprocessing complexity in multi-matrix computations.

9. Analytical Operations on Similarity and Constraints

Pattern analysis often extends beyond exact matches. stzMatrex can measure similarity and adjust constraints dynamically.

oMx = new stzMatrex("")

aMatrix1 = [[1,2],[3,4]]
aMatrix2 = [[1,2],[3,4]]
aMatrix3 = [[1,0],[0,4]]

? oMx.SimilarityScore(aMatrix1, aMatrix2)
#--> 1.0
? oMx.SimilarityScore(aMatrix1, aMatrix3)
#--> 0.5

It can also find the most similar matrix among candidates:

oMx = new stzMatrex("")

aTarget = [[1,2],[3,4]]

aCandidates = [
    [[1,2],[3,0]],
    [[5,6],[7,8]],
    [[1,2],[3,4]]
]

? @@(oMx.MostSimilarMatrix(:To = aTarget, :In = aCandidates))
#--> [[1,2],[3,4]]

And patterns can evolve interactively:

oMx = new stzMatrex("{size(3x3)}")

aMatrix = [
    [1,2,3],
    [4,5,6],
    [7,8,9]
]

? oMx.Match(aMatrix)
#--> TRUE

oMx.AddConstraint("property(symmetric)")

? oMx.Match(aMatrix)
#--> FALSE

This adaptability makes stzMatrex well-suited for progressive data validation and iterative model checking.

10. Debugging and Pattern Inspection

For deeper insight, stzMatrex offers verbose introspection tools.

oMx = new stzMatrex("{Size(3x3)}")

aMatrix = [
    [1, 2, 3],
    [4, 5, 6],
    [7, 8, 9]
]

oMx.EnableDebug()
? oMx.Match(aMatrix)
#--> TRUE

The debug trace shows how each token of the pattern is parsed and evaluated. You can also query the current pattern or inspect its internal tokens:

? oMx.Pattern()
#--> {size(3x3) & property(identity)}

? @@NL(oMx.Tokens())

These features make stzMatrex as transparent as it is expressive.

11. Comparative Advantages

While other frameworks provide mathematical functions, stzMatrex provides semantic intelligence — the ability to describe and recognize matrix patterns directly.

Conclusion

stzMatrex redefines how matrices are understood in code. Where stzMatrix performs computations, stzMatrex interprets structures. Together, they unify the computational and semantic dimensions of matrix work in Softanza.

By writing clear, human-readable expressions such as:

{size(3x3)}
{shape(square) & property(symmetric)}
{element(0..10) & @!property(diagonal)}

you bring the expressive clarity of pattern languages into the numeric world. stzMatrex turns structural logic into a readable art — bridging mathematics and meaning in one line of code.

Citer

Mansour Ayouni. « stzMatrex: Declarative Pattern Matching for Matrices in Softanza ». Softanza narrations, 2025. https://mayouni.github.io/stzsite/fr/narrations/stzmatrex-declarative-pattern-matching-for-matrices.html

Non exécuté : son code ne se charge pas tel qu'il est écrit. L'article est publié tel qu'il est écrit ; ses blocs sont des illustrations. Le code des articles est du code Haro. Là où un article a été écrit avant que la langue prenne son nom actuel, le nom a été changé et rien d'autre. Genre, domaine et série lus dans le nom du fichier, année tirée du premier commit : dérivés, jusqu'à ce que l'article porte son propre en-tête. Le fichier sur GitHub(s'ouvre dans un nouvel onglet)