Repository navigation
JuliaReach Days 3
- Marcelo Forets - Universidad de la República
- Luca Ferranti - University of Vaasa
JuliaReach Days aims to foster collaboration between students, researchers and practictioners of reachability analysis and set propagation techniques. We develop the JuliaReach package ecosystem for the Julia programming language, and collaborate on [...].
In particular we invite contributors or practitioners who have not met the core development team.
The meetup is organized into two tracks: Foundations (9th December) and Applications (10th December) There is also a Hackathon that will take place during the whole JuliaReach days, until Saturday 11th December.
The tracks last for 3 hours from 9hs to 12hs Atlantic time (GMT-3).
Times in Atlantic time (GMT-3).
| Activity | When |
|---|---|
| Opening | 9h00 -- 9h30 |
| Contributed talks | 9h30 -- 11h00 |
| Round table: IntervalArithmetic.jl | 11h00 -- 12h00 |
Afternoon workshops:
| Activity | When |
|---|---|
| Introduction to JuliaIntervals ecosystem | 13h30 -- 15h30 |
Times in Atlantic time (GMT-3).
| Activity | When |
|---|---|
| Opening | 9h00 -- 9h30 |
| Contributed talks | 9h30 -- 11h00 |
| Challenge session | 11h00 -- 12h00 |
Afternoon workshops:
| Activity | When |
|---|---|
| Introduction to ReachabilityAnalysis.jl | 13h30 -- 15h30 |
- online ✔️
- 3 days?
- is December too soon???
- poll before starting to find best time zone compromise?
-
length: 4-5 hours per day?
-
Icebreaker: 10 mins
-
Introductory talk: 40 mins
-
tea break 10 mins
-
Invited talk(s): 30 - 60 mins
-
Lunch/dinner break 30 mins
-
Session: 15+5 minutes per talk (3-4 talks?)
-
tea break (10 mins)
-
Discussion panel / round table (1h)
-
social time
- Status of Juliareach (JSOC/GSOC ideas announced here at the end?)
- Invited talk 1: ???
- Session 1: reachability methods
- Discussion panel 1: Future directions of Juliareach
- Status of Juliaintervals?
- Invited talk 2: ???
- Session 2: Interval methods?
- Discussion panel 2: Future directions of Juliaintervals, what to do to get IntervalArithmetic to v1.0?
- Status of COMP-ARCH
- Invited talk 3: ???
- Session 3: ???
- Discussion panel 3: Juliaintervals + Juliareach further collaboration?
- Final hackathon
-
Introductory talk
-
Status of ARCH-COMP 2021
-
Internal talks (contribued talks)
-
Talks of people from other communities (Invited talks)
- JuMP?
- sciml?
-
Introduce project ideas for GSOC/JSOC 2022
- FYI: organisations application deadline generally in February, but don't get to have our project idea listed by then (my project idea from last year wasn't listed at all - Luca)
-
Hackathon?
-
Discussion panels
- future directions of Juliareach?
- further JuliaIntervals JuliaReach collaboration?
-
Luca Ferranti
- Option 1: Status and roadmap of IntervalLinearAlgebra.jl
- Option 2: Solving parametric interval linear systems (PILS 🍺) with IntervalLinearAlgebra.jl
-
something about affine arithmetic could be cool, see also this paper
-
Introduction to LazySets (schillic)
- Basic set representations
- Working with lazy and concrete sets
- Application to neural network verification
-
Advanced LazySets (schillic)
- Creating new set representations
- The LazySets type hierarchy
- Commonly used design patterns
-
Introduction to ReachabilityAnalysis.jl (mforets)
- The verification problem
- User interface: reach-sets, flowpipes, and solutions
- Examples for linear, nonlinear, hybrid ODEs
-
Advanced ReachabilityAnalysis.jl (mforets)
- Adding new reachability algorithms
- Exploiting problem structure: decomposition method BFFPSV18
- Parallel flowpipe construction with LGG09
-
Introduction to juliaintervals ecosystem (david and/or luca and/or luis)
- following are ideas, need not to have all in the talk (probably wouldn't have time anyway)
- General IntervalArithmetic.jl interface
- computing the range of a function with interval methods
- computing Roots with IntervalRootFinding.jl
- interval optimisation (global and/or constrained, depending on the status of ICP.jl)
-
What set-based integration has to offer for numerical solution of PDEs (mforets/jorgepz/dfcaporale)
- Traditionally, set-based integration has been addressed to safety-critical systems, where the solutions should exhaustively cover all possible behaviors of the continuous system. We relax this assumption and consider a class of time-dependent PDEs which can be solved efficiently and accurately using set-based integration. We illustrate the power of the method by showing that with a single set-based integration we cover millions of corner cases if they were computing using standard methods.
-
Formal verification of time-triggered hybrid systems (schillic/dfcaporale)
-
Status of ARCH-COMP problem set (mforets/lbenet)