I'm Aoyang Qin (秦傲洋), a fourth-year Ph.D. student at Tsinghua University, advised by Dr. Sirui Xie and Prof. Song-Chun Zhu. I'm widely interested in fundamental problems in AI and Machine Learning, including Generative Modeling, Representation Learning and Decision-Making. I'm also excited to build various products and co-founded PicWiz. I'm currently focusing on building large-scale video generative model and honored to have an internship for training Vidu at Shengshu. I learn and write here.
We now know one electron's magnetism is relativistic. But why do billions of them in an iron bar line up — when the force aligning them is not magnetic at all? The finale: exchange interaction, the Ising model, and the symmetry breaking that links a fridge magnet to the origin of mass.
Part 2 conjured spin by switching on a field. But spin is intrinsic — it should be written into how the electron transforms when we rotate space itself. Part 3: the group theory of SU(2), SO(3), and the Lorentz group, and why the electron has no choice but to be a spinor.
Demand a wave equation that is first order in time and respects relativity, and the electron is forced to grow four components, an internal axis, and a magnetic moment out of thin air. Part 2: how spin — and magnetism — fall out of Dirac's equation.
A magnet pulls a nail across the table — surely the simplest demonstration in physics. Yet the magnetic force does no work, and classical physics turns out to forbid magnetism outright. Part 1 of a series: setting up the paradox.
Invariant tori, the KAM theorem, and the chaos that grows in the cracks; then the leap from a single trajectory to a thermodynamic law. The finale: how the whole journey resolves into one geometric picture of physics.
Strip mechanics down to its bones and what remains is a closed, non-degenerate 2-form on phase space. Part 3 is the geometry of that form — Poisson brackets as a Lie algebra, canonical transformations, Liouville and Poincaré, and the Hamilton–Jacobi equation that almost became quantum mechanics.
Every continuous symmetry hides a conserved quantity, and — turned around — symmetry alone can dictate the action. Part 2 follows that idea from Noether's theorem through relativistic invariance to the electromagnetic coupling and its quantum echo.
Newton hands us forces; we will hand them back as a single number attached to a path. Part 1 builds the foundations — potential and force, differential forms and Stokes, and the least-action principle that quietly reorganizes all of mechanics.
How do you pull a classical answer out of a quantum state? Measurement is the subtlest piece of the whole story — and the moment we understand it, superdense coding and teleportation fall right out. Part 2, the finale.
Before qubits and entanglement, there is a quieter question: why must computation be reversible at all? Part 1 of a two-part series builds the foundations — gates, qubits, dynamics, and the first taste of entanglement — ending with real Qiskit code.
Run the whole argument backwards: write an energy model, sum a single function, and let entropy fall out of it. Part 2 follows that reversal into quantum statistics — to particles that lose their names, Fermi seas, and Bose–Einstein condensates, and the deepest poem in physics.
Concede that you cannot track $10^{23}$ particles, agree only that Nature plays no favorites, and watch entropy, temperature, pressure, and free energy fall out one after another like dominoes. Part 1 of a two-part series: from counting to the architecture of thermodynamics.
Classical chaos lives and dies by the butterfly effect. But the Schrödinger equation is linear, so two close quantum states stay close forever. So where does chaos hide in the quantum world?