Ian Philip Eglin

Multivariate Variational Mode Decomposition (MVMD) can be used to decompose multichannel fMRI BOLD signals into a set of intrinsic frequency modes that are aligned across brain regions, enabling frequency-resolved functional connectivity analyses within and between large-scale networks. This allows investigation of whether anhedonia-related alterations emerge preferentially in specific slow-frequency bands (e.g., slow-4, slow-5) rather than in the full-band signal. Complementarily, the Continuous Wavelet Transform (CWT) provides a time–frequency representation of BOLD signals, enabling the study of transient and scale-dependent connectivity patterns across time. By combining MVMD (for mode-aligned band separation) with CWT (for fine-grained time–frequency characterization), the thesis can assess how frequency-specific neural dynamics and cross-network interactions differ as a function of anhedonia severity across task-based and resting-state paradigms.

Skrytelister is a digital platform launched for Helingball that allows students to exchange anonymous compliments and tokens of appreciation. To participate as a recipient, users must register, as you will only appear in the searchable directory once your account is active. The service is designed to build anticipation leading up to the ball; all messages remain hidden in a locked inbox and are only revealed to the recipients once the designated writing window has closed.

Kefli is a Rust crate implementing the Raft consensus protocol from scratch. The project prioritises correctness over performance, with a clean public API and comprehensive test coverage for leader election, log replication, and membership changes. Designed as an educational reference and drop-in building block for distributed Rust applications.

The Elevator Project focuses on designing and implementing a robust, fault-tolerant control system for n elevators operating across m floors in a distributed environment. The system must guarantee that all hall and cab calls are serviced, that button lights represent a service guarantee, and that no calls are lost — even under failure conditions such as network disconnection, packet loss, software crashes, or temporary power loss. A key requirement is resilience: elevators must continue operating independently when disconnected from the network, retain active calls across crashes or power interruptions, and recover automatically without manual reinitialization. The system must ensure consistent behavior of hall button lights across workspaces under normal conditions, while maintaining correct operation even during unreliable networking. Cab calls are local to each elevator, whereas hall calls must be coordinated across the distributed system. In addition to reliability, elevators must behave sensibly and efficiently. They should avoid unnecessary stops, respect travel direction semantics when clearing hall calls, and manage door timing and obstruction logic correctly. Once the core guarantees are satisfied, the system should distribute calls among elevators to minimize service time and optimize overall efficiency. The project emphasizes distributed systems design, state synchronization, failure recovery, real-time control logic, and scalable configuration (supporting variable numbers of elevators and floors without hard-coding). It represents a practical exercise in building a resilient, concurrent control system with strong service guarantees under realistic operational constraints.