Energy-efficient Brain-like Forgetting computing via tunable relaxation for linear-time scientific solving and training-free creativity (BFC)

Here we demonstrate a hardware-software co-design paradigm, brain-like forgetting computing (BFC), which harnesses the intrinsic, tunable conductance relaxation of memristive devices not as a flaw, but as a core computational feature. By mapping algorithm dynamics to physical relaxation, we show that complex non-linear equations including multi-dimensional electron cloud orbitals can be solved with linear complexity (O(N)) and near-zero post-stimulation energy consumption.

🌤️ Highlights

• 🏆 Proposed for the first time a bio-inspired computing approach based on a brain-like forgetting mechanism.
• 🏆 Designed and fabricated an artificial synaptic array for implementing bio-inspired computation.

🔥 Updates

• (2026.02.01) BFC core numerical calculation code release
• (2026.02.10) BFC Image and text feature extraction code release
• (2026.02.15) BFC Electron cloud orbital calculation code release
• (2026.12.24) BFC Demo release

👉 Visualization