Samuel Teuber
  • About Me
  • Research
  • Teaching
  • Blog
  • Talks
    • Heterogeneous Dynamic Logic: Provability Modulo Program Theories
    • Of Good Demons and Bad Angels: Guaranteeing Safe Control under Finite Precision
    • Heterogeneous Dynamic Logic: Provability Modulo Program Theories
    • Guaranteed Safe AI Seminars
    • Replication of the Landscape for LLM-generated Formal Specifications for C Programs
    • Formally Verified LLM Program Synthesis for Solidity Smart Contracts
    • Replication of the Landscape for LLM-generated Formal Specifications for C Programs
    • Of Good Demons and Bad Angels: Guaranteeing Safe Neural Control under Finite Precision
    • Of Good Demons and Bad Angels: Guaranteeing Safe Control under Finite Precision
    • Of Good Demons and Bad Angels: Guaranteeing Safe Neural Control under Finite Precision
    • Dynamic Logic à la Locale
    • Provable Guarantees for Neural Networks
    • Verification of Autonomous Neural Car Control with KeYmaera X
    • Revisiting Differential Verification: Equivalence Verification with Confidence
    • Next Steps in LLM-Supported Java Verification
    • Towards LLM-support for Deductive Verification of Java Programs
    • Provably Safe Neural Network Controllers via Differential Dynamic Logic
    • Heterogeneous Dynamic Logic
    • Provably Safe Neural Network Controllers via Differential Dynamic Logic
    • An Information-Flow Perspective on Algorithmic Fairness
  • Publications
    • Compositional Neural-Cyber-Physical System Verification in the Interactive Theorem Prover of Your Choice
    • Heterogeneous Dynamic Logic: Provability Modulo Program Theories
    • Of Good Demons and Bad Angels: Guaranteeing Safe Control under Finite Precision
    • Verification of Autonomous Neural Car Control with KeYmaera X
    • Revisiting Differential Verification: Equivalence Verification with Confidence
    • Next Steps in LLM-Supported Java Verification
    • Provably Safe Neural Network Controllers via Differential Dynamic Logic
    • Towards Combining the Cognitive Abilities of Large Language Models with the Rigor of Deductive Progam Verification
    • An Information-Flow Perspective on Algorithmic Fairness
    • Quantifying Software Correctness by Combining Architecture Modeling and Formal Program Analysis
    • Formally Verified Algorithmic Fairness using Information-Flow Tools (Extended Abstract)
    • Geometric Path Enumeration for Equivalence Verification of Neural Networks
    • Quantifying Software Reliability via Model-Counting
    • An Incremental Abstraction Scheme for Solving Hard SMT-Instances over Bit-Vectors
    • Efficient unpacking of required software from CERNVM-FS
  • Posts
    • Refreshingly structured
    • Uni und zyklische Abhängigkeiten
    • Mein erster Tag in Pittsburgh
    • Meine ersten Tage in den USA
    • Über das Facebook-Monopol und die Telekom der 1990er Jahre
    • Klausurenphase: Da kommt Adventsstimmung auf!
    • Fortsetzung folgt...
    • Studentisches Wohnen in Grenoble
    • Über die (Un)Bequemlichkeit zu wählen
    • Verbleibende Tage bis zur ersten Vorlesung: 0
    • How to make the UN effective — a modest proposal
    • Week 24 - Good Bye Sligo!
    • Woche 21 - Examen
    • Woche 15 - Praktikum
    • Woche *3 - deutsche Firmen
    • Woche 11 - Wochenenden
    • Woche 10 - Halloween
    • Woche 8 - Killary
    • Woche 6 und 7 - Sport
    • Woche 5 - irisches Essen
    • Woche 4 - Alles pendelt sich so langsam ein
    • Woche 3 - Ein Vergleich
    • Woche 2 - Fächer
    • Woche 1
    • Endlich da!
    • Noch eine Woche und 2 Tage
    • Jamboree Log 8
    • Jamboree Log 7
    • Jamboree Log 6
    • Jamboree Log 5
    • Jamboree Log 4
    • Jamboree Log 3
    • Jamboree Log 2
    • Jamboree Log 1
  • Projects
    • Solving difficult SMT instances using abstractions and incremental SMT solving
    • Efficient unpacking of required software from CVMFS
    • Developing a Deca subset compiler
    • Studyplan
  • Blog
  • Corrigenda
    • Corrigendum to "An Information-Flow Perspective on Algorithmic Fairness"
  • Programmierparadigmen
  • Research
  • Teaching

Of Good Demons and Bad Angels: Guaranteeing Safe Control under Finite Precision

24. Jul 2026·
Samuel Teuber
Samuel Teuber
· 0 min read
Paper
Last updated on 24. Jul 2026
Samuel Teuber
Authors
Samuel Teuber
PhD Candidate
Researching formal verification of AI-based heterogeneous systems, with applications in neural network control and algorithmic fairness.
← Heterogeneous Dynamic Logic: Provability Modulo Program Theories 25. Jul 2026
Heterogeneous Dynamic Logic: Provability Modulo Program Theories 19. Jun 2026 →

© 2026 Samuel Teuber

Published with Hugo Blox Builder.