by

## Where communities thrive

• Join over 1.5M+ people
• Join over 100K+ communities
• Free without limits
##### Activity
• Sep 26 10:38

mforets on master

• Sep 26 00:56

mforets on gh-pages

build based on 0c17fb0 (compare)

• Sep 26 00:30

mforets on master

• Sep 26 00:29

mforets on master

• Sep 25 20:30
mforets commented #2363
• Sep 25 20:12
mforets commented #2363
• Sep 25 20:07
mforets edited #2363
• Sep 25 17:41
• Sep 25 17:40
• Sep 25 17:14
• Sep 24 18:50
• Sep 24 18:40
• Sep 24 18:39

add acc model from C2E2 (compare)

• Sep 24 18:38
• Sep 24 18:38

change max_steps for spiking ne… (compare)

• Sep 23 22:24

mforets on gh-pages

build based on 601691c (compare)

• Sep 23 22:16

github-actions[bot] on v0.11.5

• Sep 23 22:02

mforets on gh-pages

build based on 010677f (compare)

• Sep 23 21:59

mforets on gh-pages

build based on 826c114 (compare)

• Sep 23 21:34
mforets synchronize #334
Marcelo Forets
@mforets
then we can let the discrete transition map apply to dispatch on such unions
that will be useful for problems where we can take the concrete intersection between the union and the guard
Marcelo Forets
@mforets

a comment on LazySets and ModelingToolkit (cc @sebastianguadalupe) note that instead of

@variables x v

X = HPolyhedron([HalfSpace(x ≥ 0, vars),
HalfSpace(v ≤ 0, vars)])

we can write

X = HPolyhedron([x ≥ 0, v ≤ 0], vars)
it would be nice that we can also parse double inequalities such as 0 <= x <= 1. but they are not yet available in ModelingToolkit (see SciML/ModelingToolkit.jl#589 )
i think though that we may be able to parse such expression with a macro along these lines
X = @set(x ≥ 0, 0 ≤ v ≤ 1, vars)
grat!
Marcelo Forets
@mforets
Shall we register SX.jl ?
Marcelo Forets
@mforets
also, registering short non-self descriptive names such as SX.jl is discouraged
but we can change it to SXFormat.jl, for instance
or SXParser.jl
any opinions on this change?
Christian Schilling
@schillic
SpaceExParser.jl?
Marcelo Forets
@mforets
i think i prefer to change the name before trying to register because im 99% sure it will not pass
SpaceExParser sounds good to me.. it's self-descriptive :)
Christian Schilling
@schillic
the problem is that the "language" has no proper name
I had no problems using it
but I didn't use it too much either
Marcelo Forets
@mforets
@SebastianGuadalupe ok. note that the build failure in JuliaReach/ReachabilityModels.jl#31 is because it's not registered
i mean we can tweak this with the travis script, but i dont see a reason not to register the package
yes, I noticed that
Marcelo Forets
@mforets

the problem is that the "language" has no proper name

ok so the winner is SpaceExParser.jl ... i can change the name of the git repo and everything

( btw i have interest in reviving that project -- it hasnt received attention in a while)
Marcelo Forets
@mforets
just saw https://sites.uclouvain.be/adhs21/index.html @blegat you are probably aware :smile:
Benoît Legat
@blegat
Yes :)
You should come !
Marcelo Forets
@mforets
it would be great
although with 2 small kids now my traveling pass is compromised :)
Marcelo Forets
@mforets
starting in 20 mins
Marcelo Forets
@mforets
how much can be said in 3 minutes :o
Peng Yu
@yupbank
does Julia supports high dimensional sparse array/tensor with einsum supported?
Marcelo Forets
@mforets
yes, there are relevant Julia tools
which one have you checked?
Peng Yu
@yupbank
i looked a bit, which none of them mentioned sparse…
i have a 100+ dimension tensor, which is significantly sparse. with only a few million non empty elements, and i need to do a lot of contraction out of them
Peng Yu
@yupbank
E.g. $T \in \mathbb{R}^{10 \times 10 \times 10\times 10\times 10 \cdots \times 10}$, dense tensor would have $10^{100}$ elements, but i only have a few million none empty data.
Marcelo Forets
@mforets
ok, i see. i suggest that you ask on julialang's discourse to have a good answer
but in principle i would expect that packages implement the abstract array interface
in which case you can just drop in any array to work with, in particular those from SparseArrays
Peng Yu
@yupbank
:+1: thank you i’ll ask in julialang's discourse
Marcelo Forets
@mforets
FYI registration for RP 2020 is still open (and it's free) until october 15th
Christian Schilling
@schillic
:+1:
Marcelo Forets
@mforets