All courses

The Hitchhiker's Guide to Hyperproperties

Seminar Winter 2026/2027

Organized by: Bernd Finkbeiner, Mishel Carelli and N. Ege Saraç

Traditional specification languages usually describe individual execution traces. In security-critical applications, this is often insufficient: attackers may compare several executions to infer information that should remain secret. Hyperproperties address this problem by specifying permitted sets of traces.

The seminar covers logics for describing hyperproperties and algorithms for checking whether systems satisfy them, with a focus on HyperLTL and automata-based model checking. For motivation, see this article on Spectre, Meltdown, and information-flow security .

Paper preferences

Skim the abstracts and introductions of the research papers, then email your ranked list by 12:30 on . Groups and papers will be announced the same day. Organizer contact details.

Dates and materials

All times are local to Munich.

  1. 15:00

    Kick-off meeting

    Introduction and seminar organization.

  2. Preferences due by 12:30

    Paper preferences and allocation

    Email your research-paper preferences in decreasing order of interest. Groups and papers will be announced the same day.

  3. 14:00–16:00

    Group presentations

    Foundational presentations followed by discussion.

    Room 5609.EG.038 (00.09.038)

  4. 14:00–18:00

    Individual presentations

    Individual research-paper talks followed by discussion.

    Room 5609.EG.038 (00.09.038)

  5. Submit by email

    Written summaries

    Submit your approximately five-page summary, placing your paper in the context of the seminar.

Format and grading

Group phase

  • Two groups of 3-4 students read their group's readings and prepare one coherent joint presentation, using examples to explain the definitions and main ideas.
  • Present for approximately 60 minutes, then lead a 30-minute discussion using exercises and quiz questions prepared in advance.
  • Participate in both group discussions.

Individual phase

  • Read one research paper and prepare a presentation explaining its problem and contribution with an example, as well as its assumptions, evidence, and limitations.
  • Present for approximately 30 minutes, then lead a 15-minute discussion. Participate in all individual discussions.
  • Write an approximately five-page summary in your own words, connecting your paper to the foundations and other talks.
Group presentation
40%
Individual presentation
40%
Individual summary
20%

Listen actively, ask questions, contribute to discussions, give constructive feedback to your peers, and seek feedback from the organizers and your peers.

The seminar is open to Bachelor's and Master's students and has no formal prerequisites. An interest in logic is helpful; courses such as Functional Programming and Verification, Diskrete Strukturen, or Einführung in die theoretische Informatik provide relevant background, but none is required.

Use of generative AI
  • You may use AI as a study aid in ways that deepen your understanding.
  • Do the readings yourself and double-check AI's answers.
  • Prepare your presentation and summary yourself.
  • State how you used AI tools in your slides and summary.
  • Ask the organizers if you are unsure whether a use is allowed.

Further reading: TUM ProLehre: Use AI wisely

Readings and papers

The group readings establish the foundations for the individual research papers.

Group A: Hyperproperties and hyperlogics

  1. Hyperproperties

    Clarkson, Schneider · Journal of Computer Security, 2010

  2. Temporal Logics for Hyperproperties

    Clarkson, Finkbeiner, Koleini, Micinski, Rabe, Sánchez · POST 2014

  3. Meltdown, Spectre, and why hardware can be correct yet insecure

    Myers · Motivation article, 2018

Group B: HyperLTL model checking

  1. Principles of Model Checking

    Baier, Katoen · MIT Press, 2008

    Selected sections:

    • §4.3 Automata on Infinite Words
    • §4.4 Model-Checking Omega-Regular Properties
    • §5.1 Linear Temporal Logic (except §5.1.6)
    • §5.2 Automata-Based LTL Model Checking (opening material and §5.2.1)
  2. Algorithms for Model Checking HyperLTL and HyperCTL*

    Finkbeiner, Rabe, Sánchez · CAV 2015

Individual research papers

Skim each paper's abstract and introduction, then rank the papers in decreasing order of interest.

  1. AutoHyper: Explicit-State Model Checking for HyperLTL

    Beutner, Finkbeiner · TACAS 2023

  2. Prophecy Variables for Hyperproperty Verification

    Beutner, Finkbeiner · CSF 2022

  3. Finding ∀∃ Hyperbugs using Symbolic Execution

    Correnson, Nießen, Finkbeiner, Weissenbacher · OOPSLA 2024

  4. Reasoning About Quality in Hyperproperties

    Graepler, Monmege, Talbot · CSL 2026

  5. Hyperproperties of Real-Valued Signals

    Nguyen, Kapinski, Jin, Deshmukh, Johnson · MEMOCODE 2017

  6. A formalization of Heisenbugs and their causes in terms of hyperproperties

    Sallinger, Weissenbacher, Zuleger · Software and Systems Modeling, 2026

  7. Hypertesting of Programs: Theoretical Foundation and Automated Test Generation

    Pasqua, Ceccato, Tonella · ICSE 2024

Learning objectives

By the end of the seminar, you should be able to:

  • Distinguish trace properties from hyperproperties using examples.
  • Formulate and interpret simple hyperproperties in HyperLTL.
  • Explain how automata-based HyperLTL model checking handles multiple traces and quantifiers.
  • Explain a paper's contribution and its connection to the seminar's foundations.
  • Evaluate a paper's assumptions, evidence, and limitations in the context of related work.
  • Communicate technical ideas clearly in presentations, writing, and scientific discussion.

News

  • Kickoff slides available

    The slides include the readings, presentation format, assessment, and deadlines.

    Kickoff slides (PDF)
  • Pre-course meeting slides available

    The slides contain the seminar organization, preliminary paper list, and key dates.

    Pre-course slides

Contact

Last updated .

Contact Data Privacy Policy Imprint
Home People Publications
More