Program

The Montana State University Algorithms Research Experience for Undergraduates (REU) is an eight-week residential summer program focused on collaborative algorithm development and problem-solving in Data Science, with topics spanning computational biology, computational geometry, computer vision, probabilistic machine learning, and cybersecurity. Participants will work alongside faculty mentors to develop their research skills and tackle novel problems in these areas.

Program Details:
  • June 07 - August 2 (8 weeks)
  • $700 stipend per week
  • Travel alowance
  • $120 per week food allowance
  • Free on-campus housing
  • Social activities

Application

Applications for the summer of 2026 are currently being accepted. Applications will be considered until May 7th, 2026, with acceptance decisions being made on a rolling basis. We strongly encourage American Indian students to apply. Please indicate in your application materials if you belong to a federally recognized tribe.

Participant Requirements:
  • NSF requires applicants to be citizens or permanent residents of the United States.
  • Applicants must be undergraduate students in good standing.
  • Applicants must have completed a course in Algorithms and programming.
Applications can be submitted online via the NSF's ETAP system by following this link: NSF ETAP. The following applications material are required:
  • Undergraduate transcript (unofficial is ok)
  • One letter of recommendation from college professor/instructor
  • Personal statement detailing:
    • Your interest in the program
    • Career aspirations
    • The (ranked) top three projects discussed below you are interested in working on.

Location

Montana State University is located in the beautiful town of Bozeman, Montana. As the epicenter of the Northern Rocky Mountains, there are endless outdoor activities on your doorstep including access to Yellowstone National Park a mere 80 miles away. Bozeman also hosts many community events in the summer that are easily accessible from campus including Shakespeare in the park, music on Main Street, and farmer's markets. Montana State University hosts many high-tech laboratory and comfortable work spaces with diverse summer dining options.

Projects

Participants in this program will partake in organized seminars and workshops to aid the development of their research skills. The cornerstone of the program is working on a project with other participant, with close mentoring from a faculty member. These projects will be centered on applied algorithms, but the application fields are quite diverse. Possible projects include:

Computational Geometry

Polyhedral terrains — three-dimensional piecewise linear functions defined over the XY-plane — model mountain landscapes and have wide-ranging applications in fields such as forest fire monitoring and geological surveying. Research on polyhedral terrains dates back to the late 1980s and remains active today, with problems like the shortest watchtower problem serving as a compelling example. This project investigates realizations of imprecise terrains with desirable properties, such as finding optimally shortest paths. Because several problems in this space are NP-hard, students will engage with the full research pipeline: proving hardness results, designing algorithms, and implementing them.

Computational Biology

Advances in DNA sequencing technology have made it possible to generate many high-quality genome assemblies for multiple individuals within the same species — a collection known as a pangenome. Because these genomes are highly similar to one another, they can be stored and analyzed more efficiently using a specialized data structure called a string graph. Beyond the raw DNA sequence, researchers can also gather information about how the genome is organized and used inside a cell. A technology called ATAC-seq reveals which regions of the DNA are physically accessible and active — these are known as accessible chromatin regions (ACRs). Nested within these regions are short, highly conserved sequences called conserved regulatory elements (CREs), which play an important role in controlling which genes get turned on or off. CREs tend to appear reliably just upstream of genes, while the DNA surrounding them varies more between individuals. This project focuses on developing computational methods to better understand and work with this type of data.

Pattern Matching problems in strings

A subsequence is a sequence of characters drawn from a string in order, but not necessarily consecutively. This project studies a family of problems centered on finding the longest subsequence of a string that repeats a certain number of times — for example, the longest subsequence that appears twice (square) or three times (cubic). Students will work on designing faster algorithms for the cubic case, investigating theoretical lower bounds on how fast any algorithm could possibly solve it, and extending these ideas to the general case where the subsequence repeats an arbitrary number of times.

Automation in Cybersecurity Reasoning

Automated discovery of vulnerabilities and understanding of how those vulnerabilities are used to exploit software systems are active areas of research with significant real-world impact. Modern software exploitation involves multiple steps---or exploit chains---which are more complex and difficult to reason about. Related, the possible effects and enabling of vulnerabilities is expansive. Automating the reasoning and assessment of software and potential emergent execution is critical to keep up with this complexity. There is existing work in using large language models (LLMs) for constructing mathematical proofs and program synthesis. This project will consider how LLMs may be used within a proof assistant to construct proofs about program behaviors and vulnerabilities.

Biometric identification

This project explores computational problems related to biometric identification and authentication. The core motivation is continuous authentication — designing a model that automatically raises an alert when a user's behavior on a device deviates from their typical patterns, potentially indicating unauthorized access. Students will develop and implement machine learning algorithms to tackle these problems, and evaluate their performance on real-world datasets.

Spatiotemporal Deep Learning for Fire Video Segmentation

Video object segmentation, i.e., automatically identifying and outlining specific objects across video frames, has important applications in surveillance, environmental monitoring, and emergency response. In fire safety, accurately detecting fires on live video can enable earlier hazard detection and rapid deployment of firefighting resources. Traditional fire detection methods often fail under conditions like variable lighting or visual clutter. This project aims to incorporate and extend Spatiotemporal Convolutional Neural Network (STCNN) architecture for domain-specific fire video segmentation, using annotated datasets of real-world and simulated fire sequences.

Contact Information

  • Dr. Adiesha Liyana Ralalage, Program Director, a.liyanaralalage@montana.edu
  • Dr. Brendan Mumey, Program Co-director, brendan.mumey@montana.edu
  • Dr. Binhai Zhu, Program Co-director, bhz@montana.edu