EJMT Abstract

Although flick-based input methods have reduced barriers to mathematical formula entry on mobile devices, limited smartphone screen space remains a critical bottleneck for advanced STEM learning. This limitation becomes especially apparent when the interface must accommodate both formula input and either lengthy AI-generated explanations or interactive diagrams. This paper presents an extended version of the Flick-Input TeX System featuring Dynamic Interface Scaling and modular integration with Large Language Models (LLMs). The proposed system provides a flexible architecture in which the number and size of flick-input keys can be reconfigured in real time through JavaScript function calls or automatic page transitions. This capability allows the interface to switch dynamically to Compact Mode, thereby reclaiming screen space for step-by-step mathematical derivations and larger interactive diagrams without excessive scrolling. By combining human-centered flick gestures with AI-driven support in a space-efficient environment, the proposed system provides a robust platform for advanced STEM learning on mobile devices.