What question did this study set out to answer?

The paper aims to present a model for reducing vulnerabilities in unsafe programming languages by focusing on geometric security principles.

February 19, 2026Open Access

Partial Geometric Security

Key Points

The paper aims to present a model for reducing vulnerabilities in unsafe programming languages by focusing on geometric security principles.
Introduces a practical security model called Partial Geometric Security.
Focuses on enforcing geometric invariants at system ingress.
Maps untrusted inputs into fixed-shape memory structures or rejects them at I/O boundaries.
Significantly reduces the attack surface in unsafe languages like C/C++.
Eliminates classes of exploits such as buffer overflows and injection attacks.
Proven to be inexpensive to implement and compatible with existing software stacks.

Abstract

Partial Geometric Security Structural Hardening of Unsafe Systems via Shape‑Validated Ingress This paper introduces **Partial Geometric Security**, a practical security model that significantly reduces attack surface in unsafe languages (e.g. C/C++) by enforcing **geometric invariants at system ingress**. Unlike full Geometric Security systems, which seal both interpretation and execution, Partial Geometric Security focuses on **structural closure at I/O boundaries**. By mapping all untrusted input into fixed‑shape memory structures—or rejecting it outright—this approach eliminates entire classes of exploits (buffer overflows, parser confusion, injection) while remaining compatible with existing software stacks. Partial Geometric Security is inexpensive to implement, deployable as a shim, and represents a decisive improvement over policy‑based defenses. Package Contents:* `manuscript.md`: Paper* `README.md`: Overview **Status:** Architectural Blueprint for Independent Implementation

Demander à l'IA

Bookmark

View Full Paper