Finding Random Integer Ideal Flow Network Signature Algorithms

• Published in: Jurnal Teknik Industri (Surabaya) Vol. 27; no. 1; pp. 105 - 120
• Main Authors: Teknomo, Kardi, Nababan, Erna Budhiarti, Bisono, Indriati Njoto, Lim, Resmana
• Format: Journal Article
• Language: English
• Published: Surabaya Universitas Kristen Petra / Petra Christian University 01.06.2025; Petra Christian University
• Subjects: Algorithms; Autonomy; Cycle; Graph theory; Ideal flow; Integers; Network management systems; Performance evaluation; Pivot; Power dynamic; Prediction models; Scaling factors; Signature; Signatures; Social networks; Strings; Sums; Term
• ISSN: 1411-2485; 2087-7439; 2087-7439
• DOI: 10.9744/jti.27.1.105-120

We propose a Random Integer Ideal Flow Network (IFN) Signature Algorithm that generates integral flow assignments in strongly connected directed graphs under uncertainty. Existing models often fail to incorporate the inherent randomness and integer constraints present in systems like social networks. Unlike traditional approaches that enforce integrality through large scaling factors, our method distributes integer coefficients across multiple canonical cycles, ensuring precise balance where the sum of inflows exactly equals the sum of outflows at each node. We introduce two pseudocode algorithms that uphold flow conservation while maintaining network irreducibility, ensuring autonomy through strong connectivity. Theoretical contributions include the decomposition of IFNs into canonical cycles and the construction of network signatures, string-based representations that allow efficient performance evaluation through direct string manipulation. These signatures enable quick validation of key network properties such as total flow, balanced link flows, and structural irreducibility. To demonstrate practical applications, we apply our algorithm to modeling family power dynamics, illustrating how IFN can create minimal yet resilient networks that balance autonomy with accountability. This framework lays the foundation for future advancements in predictive modeling and network optimization. To ensure reproducibility, we provide an open-source Python implementation on GitHub.