snha4any

SNHA algorithm implemented in various programming languages.

Already existing are implementations in:

• R - by Detlef Groth
• Python - Tim Hake
• Julia - Roberto Zarazua

References

• Groth, D.; Scheffler, C.; Hermanussen, M. (2019): Body height in stunted Indonesian children depends directly on parental education and not via a nutrition mediated pathway - Evidence from tracing association chains by St. Nicolas House Analysis. In Anthropologischer Anzeiger. DOI: 10.1127/anthranz/2019/1027
• Hermanussen, M., Aßmann, C., & Groth, D. (2021). Chain reversion for detecting associations in interacting variables—St. Nicolas House Analysis. International journal of environmental research and public health, 18(4), 1741. DOI: 10.3390/ijerph18041741

We will use the check following language transpilers to create code for several programming languages from one code base:

• Fusion - Fusion programming language, it should create code for C, Cpp, C#, D, Java, Javascript, Python, Swift, TypeScript and OpenCL C
• Haxe - Haxe programming language it should create code for C++, C-sharp, the Haxe interpreter, Java, Lua, Python and Neko
• Temper - Temper programming language transpiles to C-Sharp, Java, Javascript, TypeScript, Lua, Python and Rust (C++)
• Py2many - using the Python programming language it should create code for C, C++, D, Dart, Go, Julia, Kotlin, Nim, Rust and V - where C++ and Rust are currently considered production ready, however not ready to be used yet
• Nim - Nim programming language compiles to C, C++ and Javascript, C code seems hard to use
• V - V programming language compiles to C and can import C
• Dafny - Dafny programming language compiles to C#, Go, Python, Java or Javascript

Further the following direct implementations in other programming languages which seems not to have a high level programming language transpiler will be placed here:

• julia
• octave

Using generated C code, for instance create by Fusion, we might then use the Swig interface generator to add support for more languages, here an example for that workflow:

The following features will be implemented:

• Data structures:
  • Data frame with row and colnames features.
  • Matrix with rownames and colnames features.
  • Graph structure with degree and harmonic centrality measures.
  • Reading data from CSV files into data frames.
• Graph and data generation:
  • Graph generation for regular, random and explicit graph types.
  • Generate data with a normal distribution.
  • Data generation for graphs using Monte Carlo simulation.
  • Command line support for generating graphs and data.
  • Generating of GraphViz code for creating plots.
• Graph prediction:
  • SNHA algorithm implementation.
  • Quality measures for graph prediction if the real structure is known.
  • Command line support for data reading and prediction

Here is the process diagram:

Progress

Language	Compiler / Interpreter	Rand.fu	Stats.fu	Matrix.fu	Mgraph.fu	Mcgraph.fu
C	gcc 14-15	[x]	[x]	[x]	[x]	[x]
C++	g++ 14-15	[x]	[x]	[x]	[x]	[x]
C#	dotnet 10	[x]	[x]	[x]	[x]	[x]
D	ldc2 1.40 / dmd 2.112	[x]	[x]	[x]	[x]	[x]
Java	javac 21-25	[x]	[x]	[x]	[x]	[x]
JavaScript	node 20-22	[x]	[x]	[x]	[x]	[x]
Python	python 3.13	[x]	[x]	[x]	[x]	[x]
Swift	swift 6.0-6.3	[x]	[x]	[x]	[x]	[x]
TypeScript	bun 1.3	[x]	[x]	[x]	[x]	[x]

Language	FileSystem.fu	CSVParser.fu	SNHA.fu	Timeit.fu	ListUtils.fu
C	[x]	[x]	[x]	[x]	[x]
C++	[x]	[x]	[x]	[x]	[x]
C#	[x]	[x]	[x]	[x]	[x]
D	[x]	[x]	[x]	[x]	[x]
Java	[x]	[x]	[x]	[x]	[x]
JavaScript	[x]	[x]	[x]	[x]	[x]
Python	[x]	[x]	[x]	[x]	[x]
Swift	[x]	[x]	[x]	[x]	[x]
TypeScript	[x]	[x]	[x]	[x]	[x]

Class Diagram (WIP)

Authors

The following persons contribute to the project:

• Alba de Prada Hernandez (AP), University of Potsdam (graph and data generation modules)
• Christopher Olvera (CO), University of Potsdam (implementation of SNHA algorithm and accuracy metrics)
• Detlef Groth (DG), University of Potsdam (supervisor and random data generation)
• Harshini Praveen Kumar (HP), University of Potsdam (statistics module)
• Roberto Zarazua (RZ), University of Potsdam (Julia code, used for implementation)