Chaos in random neural networks: the 1988 paper behind LRU's design
RobertTLange · x · 2026-11-15
Follow-up to the RNN eigenvalue thread, with deeper reading:
- Sompolinsky et al. (1988), "Chaos in Random Neural Networks": a mean-field theory for random asymmetric networks predicting a transition from stationary to chaotic phase past a critical gain, with autocorrelations and maximal Lyapunov exponent computed.
- Orvieto et al. 2023, the Linear Recurrent Unit paper: the work behind initialising eigenvalues on a thin ring inside the unit circle for stable long memory.
The animation is a toy sim: 100 units, random Gaussian matrix, eigenvalue ring |λ| ∈ [0.9, 0.999].
Related event: RNN Memory Explained: Recurrent Matrix Eigenvalues and LRU Initialization(2 posts)→
More from Research
- Stanford's Flash ALOHA robot arms catch a thrown ball in 1.85s with just 50 demos — k7agar · 2026-11-15
- Drone swarms tie knots in midair to lift objects autonomously — lukas_m_ziegler · 2026-11-15
- SO-Arena Benchmark: Top-Scoring Model Ranks Only 7th in Decision Reliability — kalyan_kpl · 2026-11-15
- Fleuret: v1.4 training was a disaster, but FCA-style models hit 41.17% on GSM8K — francoisfleuret · 2026-11-15
- World Model Hallucination Paper Wins Best Paper at CoRL 2026 PhysWM Workshop — LerrelPinto · 2026-11-15
- RNN memory is set by recurrent matrix eigenvalues — LRU's ring init fixes it — RobertTLange · 2026-11-15