arXiv:2604.14332cs.LGcs.AI2026-04被引 1

用物理热力学实现扩散模型推理,能耗降万倍,无需数字计算。

Thermodynamic Diffusion Inference with Minimal Digital Conditioning

论文配图:Thermodynamic Diffusion Inference with Minimal Digital Conditioning
图 1 · 摘自论文原文
  • 用分层双线性耦合模拟网络跳连,连接数从平方级降至线性级。
  • 仅用2560参数的极简数字接口,使输出相似度达0.9906,接近理论上限。
  • 首次实现可量产的热力学扩散推理,适合低功耗边缘设备部署。

扩散模型推理与过阻尼朗之万动力学在数学上完全等价。若物理介质能编码得分函数,系统将仅靠热力学自然达到正确输出,推理过程无需数字计算,理论上可比GPU节省高达10⁴倍能耗。但两大障碍长期制约其规模化应用:非局部跳连无法由局部耦合的模拟硬件表示;输入条件控制中耦合常数信号量不足原需求的2600倍,难以锚定特定输入。本文提出两种解决方案:层级双线性耦合通过编码器与解码器格拉姆矩阵的奇异结构,将U-Net跳连表示为秩-k模块间交互,仅需O(Dk)物理连接而非O(D²);设计4维瓶颈编码器与16单元传输网络构成极简数字接口,总参数量仅2560。在训练好的去噪U-Net激活数据上测试,系统解码余弦相似度达0.9906(理论上限1.0000),同时保持约10⁷倍于GPU推理的理论能耗优势。此为首个实现训练权重、生产规模热力学扩散推理的成果。

原文摘要 · Abstract (English)

Diffusion-model inference and overdamped Langevin dynamics are formally identical. A physical substrate that encodes the score function therefore equilibrates to the correct output by thermodynamics alone, requiring no digital arithmetic during inference and potentially achieving a $10{,}000\times$ reduction in energy relative to a GPU. Two fundamental barriers have until now prevented this equivalence from being realized at production scale: non-local skip connections, which locally coupled analog substrates cannot represent, and input conditioning, in which the coupling constants carry roughly $2{,}600\times$ too little signal to anchor the system to a specific input. We resolve both obstacles. \emph{Hierarchical bilinear coupling} encodes U-Net skip connections as rank-$k$ inter-module interactions derived directly from the singular structure of the encoder and decoder Gram matrices, requiring only $O(Dk)$ physical connections instead of $O(D^2)$. A \emph{minimal digital interface} -- a 4-dimensional bottleneck encoder together with a 16-unit transfer network, totalling \textbf{2,560 parameters} -- overcomes the conditioning barrier. When evaluated on activations drawn from a trained denoising U-Net, the complete system attains a decoder cosine similarity of \textbf{0.9906} against an oracle upper bound of 1.0000, while preserving theoretical net energy savings of approximately $10^7\times$ over GPU inference. These results constitute the first demonstration of trained-weight, production-scale thermodynamic diffusion inference.

扩散模型热力学推理低功耗计算硬件加速

Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品;中文卡片由大模型生成,请以原文为准。