feat: add BatchedInterface
#65
Conversation
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #65 +/- ##
===========================================
+ Coverage 0.00% 26.68% +26.68%
===========================================
Files 8 10 +2
Lines 372 476 +104
===========================================
+ Hits 0 127 +127
+ Misses 372 349 -23 ☔ View full report in Codecov by Sentry. |
| @test associated_systems(bi) == [1, 1, 1, 1, 2, 2, 3, 3] | ||
|
|
||
| getter = getu(bi) | ||
| @test (@inferred getter(probs...)) == [1.0, 3.0, 0.2, 0.3, 5.0, 0.6, 0.7, 0.8] |
There was a problem hiding this comment.
This doesn't seem quite right, you'd want an array for each to make remaking easy?
src/batched_interface.jl
Outdated
| if !is_variable(sys, sym) && !is_parameter(sys, sym) | ||
| error("Only variables and parameters allowed in BatchedInterface.") | ||
| end | ||
| if !in(sym, symbol_order) |
There was a problem hiding this comment.
| if !in(sym, symbol_order) | |
| if isnothing(findfirst(isequal(sym), symbol_order)) |
I think this should be changed as in least to using symbolic variables in a Boolean context.
AayushSabharwal
force-pushed
the
as/batched-getu
branch
from
59b6700 to
0e8e037
Compare
AayushSabharwal
force-pushed
the
as/batched-getu
branch
from
9ac7f73 to
1bfde6e
Compare
AayushSabharwal
force-pushed
the
as/batched-getu
branch
from
1bfde6e to
5aad4fa
Compare
| error("Only symbolic variables allowed in BatchedInterface.") | ||
| end | ||
| if symbolic_type(sym) === ArraySymbolic() | ||
| append!(allsyms, collect(sym)) |
There was a problem hiding this comment.
This removes the need for ArrayPartitions?
| @variables x[1:2] y z | ||
|
|
||
| syss = [ | ||
| SymbolCache([x..., y]), |
There was a problem hiding this comment.
Is splatting required? Would [x, y] work?
There was a problem hiding this comment.
Splatting is required. SymbolCache treats the variables as-is, so it would think x is at index 1 and y at 2
There was a problem hiding this comment.
Ah, I see. But for array parameters in MTK, that would be different?
There was a problem hiding this comment.
Yes. This is specifically only needed for SymbolCache, MTK handles it completely differently. I didn't want to pull in MTK for this one test so I used Symbolics and the rest of the SII infrastructure
|
SciMLBase downstream failing here is my fault in MTK 😅 It's fixed in SciML/ModelingToolkit.jl#2630 |
|
Can we get a follow up tutorial on the batched interface? That's one that takes a bit more depth than the other pieces. |
Checklist
contributor guidelines, in particular the SciML Style Guide and
COLPRAC.
Additional context
Add any other context about the problem here.