Week 3 - Java Solver Libraries

This past week marked the transition from theoretical preparation to hands-on coding for my GSoC project. Since I have an exam coming up on Monday, my available time was a bit limited. I will resume my regular workload after that. My main focus this week was evaluating Java-based solver libraries to see how well they might integrate with SBSCL.

I started by working with Apache Commons Math (specifically the SimplexSolver). I recreated minimal working examples to understand its functionality and API structure. This involved dynamically defining variables, constraints, and objective functions. Based on these tests, I was able to confirm that it's possible to dynamically map SBML-derived models to this solver’s API, which aligns perfectly with the flexibility required for SBSCL.

Then I explored the ojAlgo library. Following a similliar approach as with Commons Math, I set up basic examples in Maven to try out ojAlgo's functionalities, explore its API, and see how result extraction is handled. It is worth noting that the newer ojAlgo's API (v55) introduces certain structural differences, such as handling variables and constraints via model methods. 

Comparing both Apache Commons Math and ojAlgo gave me a clear understanding of their strengths and differences. Both libraries proved suitable for dynamically building linear optimization models, although their API styles and solution handling vary significantly.

Finally, I started drafting the solver abstraction interface (ILPSolver), drawin inspiration from the Optlang Python package we discussed earlier. The idea is to provide SBSCL users with the ability to switch seamlessly between solvers. While I took an initial look at Google OR-Tools, I decided to wait with the practical implementation until I have completed and tested the first interface version using Commons Math and ojAlgo.

Looking ahead, I plan to focus on implementing and refining the solver abstraction layer. This will involve initial tests with Commons Math and ojAlgo to make sure our solver interface is practical and flexible enough for future solver integrations. As of now, I am on schedule and expect to have a first working model of the interface ready soon.

That’s it for this week, see you next week!

Project Details:

Kommentare

Kommentar veröffentlichen

Beliebte Posts aus diesem Blog

OptSolvX in SBSCL: A Solver-Agnostic LP Core (GSoC 2025 Final Report)

Bild
Enhancing the Systems Biology Simulation Core Library (SBSCL) with a Solver-Agnostic Framework for Constraint-Based Simulation and Analysis National Resource of Network Biology (NRNB) Google Summer of Code 2025 Final Report Name Michael Gaas Mentors           Prof. Dr. Andreas Dräger, Matthias König Overview This work modernizes constraint-based simulation in the Systems Biology Simulation Core Library (SBSCL) by introducing a solver-agnostic linear programming (LP) core (OptSolvX), a thin integration layer within SBSCL, and a bridge from SBML/FBC to linear programs. The objective is to restore robust, portable Flux Balance Analysis (FBA) on current platforms, including ARM-based systems, while preserving SBSCL’s public API and keeping solver-specific concerns decoupled from biological model handling. Implementation Summary OptSolvX provides a minimal, solver-agnostic LP domain model with AbstractLPModel, Variable, Constraint, OptimizationDirection,...
Mehr anzeigen

Week 2 – Community Bonding Wrap-Up

Community Bonding Period - Getting Ready for SBSCL The community bonding period is coming to an end, and the actual coding phase is about to begin. Over the past two weeks, I've spent my time getting familiar with the Systems Biology Simulation Core Library (SBSCL) and preparing for my GSoC project. During this initial phase, I focused on understanding how SBSCL uses SBML (Systems Biology Markup Language) to simulate biological models. My primary interest was in constraint-based simulations, particularly Flux Balance Analysis (FBA). My project aims at enhancing SBSCL by developing a solver-agnostic framework, ensuring compatibility with different solvers and hardware platforms, including newer ARM-based systems like Apple's M-series processors. I gained deeper insight into FBA as a computational method used to calculate the optimal flow of metabolites through metabolic networks to achieve specific biological goals, such as maximizing biomass or minimizing nutrient usage. Howeve...
Mehr anzeigen

Week 5 - Unit Testing, OjAlgo Integration, and Code Cleanup

In the past week, my primary goal was enhancing the test infrastructure, finishing the integration of the second solver, OjAlgo, and maintaining overall code quality and clarity. Progress: Unit Tests for CommonsMathSolver Implemented unit tests covering maximization, minimization, variable bounds, equality constraints, unsolvable and unbounded LP scenarios, and precision checks. Standardized test logic and provided clear comments aligned with Commons Math documentation and architecture. Test Environment Stabilization Fully synchronized Maven and IntelliJ configurations, eliminating issues with dependencies (JUnit, etc.) and resolving ClassNotFoundException permanently. Cleaned up the project structure by separating test and main code, removing obsolete files, and ignoring IDE-specific files ( .idea , .iml ) from version control. Development of Second Solver Adapter ( OjAlgoSolver ) Created a new adapter implementation to the ILPSolver   interface. S...
Mehr anzeigen