<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Ritwik Singhai — A Mathematical Atlas</title><description>Deep dives and research notes on deep learning, probability, and the mathematics underneath.</description><link>https://ritwiksinghai.com/</link><language>en-us</language><item><title>What Attention Actually Computes</title><link>https://ritwiksinghai.com/articles/what-attention-computes/</link><guid isPermaLink="true">https://ritwiksinghai.com/articles/what-attention-computes/</guid><description>Attention is a soft, differentiable dictionary lookup—dot-product scores, a softmax, and a weighted sum of values—that gives every position a global receptive field in a single layer.</description><pubDate>Wed, 20 May 2026 00:00:00 GMT</pubDate><category>Attention</category><category>Transformers</category><category>Deep Learning</category><category>Linear Algebra</category><category>Sequence Models</category></item><item><title>Walking Downhill: What Gradient Descent Actually Does</title><link>https://ritwiksinghai.com/articles/gradient-descent/</link><guid isPermaLink="true">https://ritwiksinghai.com/articles/gradient-descent/</guid><description>Gradient descent trains almost every neural network by repeatedly taking a small step opposite the direction of steepest ascent — a procedure whose successes and pathologies both follow from that single geometric idea.</description><pubDate>Wed, 22 Apr 2026 00:00:00 GMT</pubDate><category>optimization</category><category>gradient descent</category><category>deep learning</category><category>calculus</category><category>machine learning</category></item><item><title>Why √n — Reading the Central Limit Theorem</title><link>https://ritwiksinghai.com/articles/central-limit-theorem/</link><guid isPermaLink="true">https://ritwiksinghai.com/articles/central-limit-theorem/</guid><description>The sample mean of many independent finite-variance draws becomes Gaussian, and its spread shrinks like σ/√n — a rate that follows from how variance adds, and that breaks precisely when its hypotheses do.</description><pubDate>Sat, 28 Mar 2026 00:00:00 GMT</pubDate><category>probability</category><category>central-limit-theorem</category><category>statistics</category><category>asymptotics</category><category>gaussian</category></item><item><title>Sensitive Dependence</title><link>https://ritwiksinghai.com/articles/sensitive-dependence/</link><guid isPermaLink="true">https://ritwiksinghai.com/articles/sensitive-dependence/</guid><description>Why fully deterministic systems—weather, pendulums, a tumbling die—can still defy long-range prediction, and how a positive Lyapunov exponent turns finite-precision measurement into vanishing knowledge of the future.</description><pubDate>Tue, 03 Mar 2026 00:00:00 GMT</pubDate><category>chaos</category><category>dynamical systems</category><category>Lorenz</category><category>prediction</category><category>nonlinear dynamics</category></item><item><title>Every Wave Is a Sum of Circles</title><link>https://ritwiksinghai.com/articles/sum-of-circles/</link><guid isPermaLink="true">https://ritwiksinghai.com/articles/sum-of-circles/</guid><description>Any reasonable periodic function is a sum of rotating vectors, and the recipe for the radii is a sequence of inner products.</description><pubDate>Mon, 09 Feb 2026 00:00:00 GMT</pubDate><category>Fourier</category><category>Signal Processing</category><category>Mathematics</category><category>Harmonic Analysis</category></item><item><title>The Golden Angle: Why 137.5°</title><link>https://ritwiksinghai.com/articles/the-golden-angle/</link><guid isPermaLink="true">https://ritwiksinghai.com/articles/the-golden-angle/</guid><description>A sunflower head packs its seeds by placing each one a fixed angle of about 137.5° from the last, and that exact angle falls out of the golden ratio being the hardest number to approximate by fractions.</description><pubDate>Sun, 18 Jan 2026 00:00:00 GMT</pubDate><category>phyllotaxis</category><category>golden-ratio</category><category>continued-fractions</category><category>fibonacci</category><category>geometry</category></item><item><title>The Evolution of Attention</title><link>https://ritwiksinghai.com/logs/attention-paper/</link><guid isPermaLink="true">https://ritwiksinghai.com/logs/attention-paper/</guid><description>Research log — The Evolution of Attention</description><pubDate>Mon, 15 Dec 2025 00:00:00 GMT</pubDate><category>NLP</category><category>Foundations</category><category>Deep Learning</category></item><item><title>Our Quest for AGI Might Lead Us to Rutherford...</title><link>https://ritwiksinghai.com/articles/rag-failure/</link><guid isPermaLink="true">https://ritwiksinghai.com/articles/rag-failure/</guid><description>The final abstraction layer of any intelligent system is limited to the language English. Why???</description><pubDate>Mon, 20 Oct 2025 00:00:00 GMT</pubDate><category>LLM</category><category>Language</category><category>Mathematics</category><category>Philosophy</category></item></channel></rss>