Skip to content

Publications

PySNE undergraduate theses

pysne originated as two companion undergraduate theses (Tugas Akhir) at Institut Teknologi Bandung, both supervised by the same advisor — who is also a co-author of the foundational clustering paper the library builds on (see Foundational literature below).

Note

Both theses have been submitted and are pending graduation assessment. They are not yet publicly available; this page will be updated with a repository link / DOI once they are.

Design of a Python Library for a Spiral Optimization Algorithm with Clustering for Finding All Roots of Systems of Nonlinear Equations

  • Author: Aldy Nugraha Hermawan (NIM 10822012)
  • Institut Teknologi Bandung, Actuarial Science Study Program
  • Advisor: Adhe Kania, S.Si., M.Si., Ph.D.
  • July 2026 — undergraduate thesis, submitted, pending graduation assessment

Covers the real- and integer-domain sides of pysne: the SNE root-finding implementation of Spiral Optimization with Clustering (SPOC), the library's low-discrepancy initialization methods, and its extension to integer (Diophantine) problems. Compares several low-discrepancy sequences for initializing the search and finds Korobov sequences give the most consistent results, which is why they were adopted as the library's default. Reports that SPOC in pysne recovers all reference roots across the benchmark systems tested, and that the integer extension reproduces known reference solutions while also surfacing additional numerically valid ones in some cases.

Design of a Python Library Based on the Spiral Optimization Algorithm with Clustering for Finding All Optimal Solutions of Multimodal Functions

  • Author: Azarya Benhanan Isriyanto (NIM 10822041)
  • Institut Teknologi Bandung, Actuarial Science Study Program
  • Advisor: Adhe Kania, S.Si., M.Si., Ph.D.
  • July 2026 — undergraduate thesis, submitted, pending graduation assessment

Covers the multimodal-optimization side of pysne: combining SPOC with Sobol-sequence sampling to locate every optimum (both peaks and valleys) of a multimodal function in a single run, rather than just the global optimum. Introduces the num_check_points multi-point-check parameter specifically to prevent over-merging of clusters on aggressive, closely-packed landscapes. Validated on standard benchmark functions from low dimensions up to 8-D, reporting a 100% solution discovery rate across the tested cases for both SNE and multimodal problems.

Foundational literature

pysne's core algorithm — Spiral Optimization with Clustering (SPOC) — is built directly on this line of work:

  • Tamura, K., Yasuda, K. (2011). Spiral Dynamics Inspired Optimization. Journal of Advanced Computational Intelligence and Intelligent Informatics, 15, 1116–1122.
    The original spiral optimization algorithm that SPO is based on.

  • Sidarto, K.A., Kania, A. (2015). Finding all solutions of systems of nonlinear equations using spiral optimization with clustering. Journal of Advanced Computational Intelligence and Intelligent Informatics, 19(5), 697–707.
    Introduces the clustering technique pysne's SNEProblem pipeline implements, for finding all roots of a nonlinear system in a single run.

The two theses above extend this same clustering idea in two directions that pysne implements as separate problem types: general multimodal optimization (MultimodalProblem) and integer-constrained problems (DiophantineProblem).

  • Algorithms — how SPOC/SPO and clustering work inside pysne.
  • References — additional background reading.