-
读懂「Visualizing the Gilbreath expectation sequence」：先把定义、反例和适用范围钉死
One byproduct of learning how to use coding agents to create visualizati…
-
Online‑Safe‑Softmax 深度解析：从内存瓶颈到 FlashAttention 底层基石
GPU 上 Softmax 算子绝大多数场景属于内存受限算子（memory‑bound），算力单元经常处于空闲状态，性能瓶颈来自 HBM 高带宽…
-
读懂「Two more apps: visualizing the zeta process and」：先把定义、反例和适用范围钉死
I believe that the creation of visualization apps to illustrate mathemat…
-
把「A digestion of the Jacobian conjecture counterex」写成可执行的阅读路径
The notorious Jacobian conjecture can be formulated concretely over the …
-
读懂「A partial digestion of the HRT counterexample」：先把定义、反例和适用范围钉死
A function of one variable can be translated in space by a spatial shift…
-
2026 TensorRT静态与动态插件区别：底层原理+trtexec实操避坑指南
TensorRT静态插件与动态插件的核心差异不在于文件格式，二者Linux环境下均为.so共享库，本质区别是插件创建器的注册、生命周期管理方式。…
-
「A digestion of the proof of Sendov’s conjecture」：数学直觉如何落到可检验的步骤
This post concerns the following conjecture of Sendov, as well as its st…
-
「Palomar – a registry of Lean verified mathematic」笔记：问题从哪来、证明卡在哪
In recent months there has been a proliferation of AI-generated proofs o…
-
「Quantitative bounds for sets lacking polynomial」笔记：问题从哪来、证明卡在哪
Ben Krause, Hamed Mousavi, Joni Teräiväinen, and I have just uploaded to…
-
2026自动微分求解海森矩阵：原理拆解+落地步骤+算力避坑指南
PyTorch、TensorFlow等自动微分框架原生仅支持雅可比矩阵计算，海森矩阵无法通过链式法则直接求解，核心落地方法是对函数梯度再次求雅可…