Why can't FeynGame directly convert drawn diagrams into symbolic integrals like FeynCalc does?

Your question touches on a fundamental difference in what these tools are designed to do—and honestly, the preprints provided don't directly address this architectural distinction. Let me explain what I can infer from the archive:

FeynGame is a diagram drawing tool [AG-2024.01-1123, AG-2025.01-1133]. Its job is to help you create publication-quality Feynman diagrams and, more recently, to visualize diagrams generated by other programs like QGRAF [AG-2025.01-1133]. It's a front-end for human communication and learning.

FeynCalc, by contrast, is a symbolic computation engine that takes Feynman diagrams (in a machine-readable format) and performs the actual mathematical work—reducing them to master integrals, evaluating loop momentum integrals, handling tensor structures, and so on [AG-2025.12-1817].

The gap between them is one of representation and automation. A drawn diagram is visual and intuitive for humans but ambiguous for a computer: it doesn't automatically encode which momentum flows where, what the loop structure is, or how to apply symmetries. FeynCalc needs an unambiguous symbolic specification—typically generated by dedicated programs like QGRAF that *parse* physics rules—not a picture.

Interestingly, newer work is bridging this: an agentic framework called Diagrammatica can convert diagram specifications into FeynCalc code automatically [AG-2026.03-1733], but it works with compact, convention-fixing specifications rather than freehand drawings.

The provided preprints don't directly explain why direct drawing-to-integral conversion is hard; a detailed answer would require papers on the semantic parsing of diagram topology and momentum routing.