2025-03-04: The program is full. There was more interest than we anticipated in this REU, and we had more qualified applicants than we could admit. Unfortunately we can not admit any more students.
Program Description
This REU is a fully funded opportunity for undergraduates students and focuses on mathematical aspects of problems encountered when sampling from probability distributions in high dimensions. Such problems arise often in many applications ranging from molecular dynamics to machine learning. This is also used in image generation algorithms, such as the one used to generate the picture of the cat in a red shirt on a skateboard at the top of this page. The goal of this REU is to to study, analyze and develop algorithms that are useful when sampling from high dimensional distributions.
This program is designed for mathematical/analytically inclined students, and preference will be given to students with a strong background probability. We aim to admit about 6 students, who will be divided into two teams and work with the faculty member Gautam Iyer, and a graduate student. Each team will produce a report, and a presentation at the end of the program. If suitable, the report will be converted into a paper for submission to a peer-reviewed journal, and selected participants will be given an opportunity to present their work in the Joint Mathematics Meetings.
Eligibility
To be eligible for this program, candidates must be:
- US citizens or permanent residents
- Current undergraduate students (current seniors are not eligible)
- Majoring in mathematics
How to Apply
The application packet should include:
- A personal statement, indicating the student’s interests and career goals
- A CV / resume
- An academic transcript
- Two letters of recommendation
All applications should be submitted electronically on MathPrograms by February 15th.
Logistical Arrangements
Admitted students will be paid a stipend and housed on CMU campus.
Contact
For questions, please contact Gautam Iyer
Support Acknowledgement
This REU is supported by the NSF award DMS 2342349