Theory of Statistics II
(STATS 300B, Winter 2024)
- Instructor: Tselil Schramm
- Time: Mondays & Wednesdays, 11:30–12:50 Pacific
- Location: 60-109
- TAs: John Cherian and Tim Sudijono
- Office hours: listed on Canvas
- Contact: email us, including "STATS 300B" in the subject line.
Resources:
- Our course texts (all available online free or with Stanford ID access) will be:
-
The course scribe notes on Overleaf (read-only); editing link available on Canvas.
- You may also find useful the lecture notes and slides from a previous iteration of this course, taught by John Duchi.
Prerequisites:
- Real Analysis (MATH 171 or equivalent).
- Probability Theory (at the level of STATS 310A, or equivalent).
- Linear Algebra (chapter 4.1 of Vershynin's book should feel comfortable).
- Finite-sample hypothesis testing theory (e.g. STATS 300A) is helpful, but not required.
If you are missing this background, I suspect you will be able to easily catch up by reading some of Lehmann and/or Lehmann-Romano.
Course Policies: A detailed overview of course policies (including grading and assignments) can be found in the course syllabus.
Course description
This course is the second in the "Theory of Statistics" sequence for Stanford Statistics PhD students.
We will first cover classical results from asymptotic statistics, where the primary goal will be to understand (taking a hard-line mathematical perspective) what makes a good estimator in the large sample limit.
Then, we will shift our focus to non-asymptotic high-dimensional statistics, where we will build tools for establishing concentration (a.k.a. confidence intervals) of estimators.
Finally, we will briefly discuss algorithmic concerns in the high-dimensional regime.
Lectures and Reading
Below is a preliminary schedule (subject to change), including the readings relevant to each lecture.