Latest News
Deep dives into evolutionary computation, project updates, and community news.
EvoX Genesis Completes WASM Render in About Two Days for $36
Using DeepSeek V4 Flash, EvoX Genesis completed WASM Render in about two days for just $36 in model charges, becoming the first AI system worldwide to submit a result for the challenge. Team test records show 2,017/2,019 Khronos CTS tests passed (99.90%), including all 1,184/1,184 WebGL2.0 conformance tests run.
EvoCoCo: Automatic MATLAB-to-EvoX Code Conversion for GPU-Accelerated Multiobjective Optimization
EvoCoCo converts MATLAB multiobjective evolutionary algorithms into GPU-accelerated EvoX implementations through semantics-guided tensorization, execution feedback, and multi-agent validation. Evaluated on 48 algorithms, it achieves 88.2% optimization-fidelity coverage.
AutoPSO: GPU Computing Power × Meta-Evolution → Fully Automatic Particle Swarm Optimization
The EvoX team's AutoPSO, published in IEEE TEVC, encodes the parameter configuration, learning mechanisms, and swarm structure of PSO into a searchable evolutionary space, and automatically searches for the optimal PSO variant for a given target problem through meta-evolution.
EvoX Genesis: A Recursive AI System for Long-Horizon Autonomous Software Evolution That Built a 250K-Line C Compiler from Scratch
The EvoX team from the Department of Data Science and Artificial Intelligence at The Hong Kong Polytechnic University has released EvoX Genesis, a recursive AI system for long-horizon autonomous software evolution. Rather than relying on a persistent agent to sustain long-horizon development, it lets the software world itself keep evolving: starting from an empty repository, the system built a 248,989-line C compiler in 123.4 hours, with a total model token cost of just $44.38.
Codex for EvoX: Design Algorithms, Migrate Code, and Run Reproducible Experiments Through Conversation
Evolutionary computation is stepping into the AI-native era. With Codex as a representative AI coding agent, EvoX turns natural-language intent into runnable evolutionary-computation workflows—covering problem modelling, algorithm migration, backpropagation-free neuroevolution, and GPU tensorization.
EvoGM: Autonomous LLM Merging via Population Evolution Without Retraining
The EvoX team, in collaboration with Peng Cheng Laboratory, proposes EvoGM, a generative evolutionary model merging framework that reformulates coefficient search as a learnable generative optimization problem, enabling autonomous model merging without retraining large models.
EvoGP: A GPU-Native Framework for Tree-Based Genetic Programming at 10^11 GPops/s
EvoGP reorganizes tree representation, genetic operators, and parallel execution from the ground up, achieving peak throughput exceeding 10^11 GPops/s and up to 304× speedup over existing GPU implementations.
EvoGO: GPU Compute × Generative Learning → A New Paradigm for Evolutionary Algorithms with 10-Generation Convergence
EvoGO is a fully data-driven evolutionary optimization framework that learns how to generate better solutions from past search experience and achieves fast, strong performance on complex high-dimensional tasks.
iStratDE: GPU Computing x Ultra-Large Populations Unlock the Full Potential of Differential Evolution
The EvoX team introduces iStratDE, a GPU-accelerated differential evolution method that assigns fixed strategies at the individual level, enabling communication-free large-scale parallel search with strong empirical performance and theoretical convergence guarantees.
EvoX Quickstart: Run GPU-Accelerated Evolutionary Computation in Just 10 Minutes
A beginner's tutorial to get started with GPU-accelerated evolutionary computation using EvoX in just 10 minutes.