Array of components

[B-LIE]

Is there an example of how to use arrays of components in the documentation❓

From a presentation, I've seen the the syntax:

@components begin
    resistor[1:n] = Resistor()
end

I don't know if this is a standard notation or not. I've tried to use it, but have problems when I add equations where the instances have fields, such...

@equations begin
    resistor[1].R ~ 3 
end

would not work.

However,

@components begin
    resistor_1 = Resistor()
end
#
@equations begin
    resistor_1.R ~ 3
end

would work.

[ChrisRackauckas]

@ven-k is the array form documented?

[ven-k]

Yes it's the examples here: https://docs.sciml.ai/ModelingToolkit/stable/basics/MTKModel_Connector/#What-can-an-MTK-Model-definition-have?

The relevant bits (from ModelC example):

@mtkmodel ModelC begin
    @structural_parameters begin
        N = 2
    end
:
    @components begin
        model_a = ModelA(; k_array)
        model_array_a = [ModelA(; k = i) for i in 1:N]
        model_array_b = for i in 1:N
            k = i^2
            ModelA(; k)
        end
    end
:
end

For @named model_c = ModelC() ,the subcomponents can be then accessed as model_c.model_array_a_1, model_c.model_array_a_2 and so on.

[B-LIE]

Thanks for answer! Useful, but it only addresses part of my question, and misses the most important part (marked with red): Creation of vectors: The example you refer to illustrates how a vector of models, `model_array_a` and `model_array_b` can be created. I missed that. Yingbo has used a third way, e.g., `model_array_c[1:N] = ModelA()` in his presentations. Perhaps this third way is only suitable when all model instantiations are equal... Reference to vector elements: You write: The relevant bits (from ModelC example): @mtkmodel ModelC begin @structural_parameters begin N = 2 end : @components begin model_a = ModelA(; k_array) model_array_a = [ModelA(; k = i) for i in 1:N] model_array_b = for i in 1:N k = i^2 ModelA(; k) end end : end For @nAmed model_c = ModelC() ,the subcomponents can be then accessed as model_c.model_array_a_1, model_c.model_array_a_2 and so on. This is useful. However, it doesn't answer my question, which was: how do I address the elements of, say, model_a within the ModelC definition. In other words, suppose I add: ... @components begin model_a = ModelA(; k_array) model_array_a = [ModelA(; k = i) for i in 1:N] model_array_b = for i in 1:N k = i^2 ModelA(; k) end end @equations begin for i in 1:N model_array_a[i].k ~ 3 end end : end OK – not a perfect example since `k` is a parameter; in my case it would be a variable. What I have inserted in red above does not work. Should it be `model_array_a_i` instead? Or something else? BR, Bernt

…

________________________________ From: Christopher Rackauckas ***@***.***> Sent: Monday, April 29, 2024 8:04 AM To: SciML/ModelingToolkit.jl ***@***.***> Cc: Bernt Lie ***@***.***>; Author ***@***.***> Subject: Re: [SciML/ModelingToolkit.jl] Array of components (Issue #2674) Closed #2674<#2674> as completed. — Reply to this email directly, view it on GitHub<#2674 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AM5VM72TFKURHHYMTHC6PTLY7XPFLAVCNFSM6AAAAABG4EHJVWVHI2DSMVQWIX3LMV45UABCJFZXG5LFIV3GK3TUJZXXI2LGNFRWC5DJN5XDWMJSGYZTOOJSGQZTCMQ>. You are receiving this because you authored the thread. T

[B-LIE]

To make it simpler, consider my slightly modified code from the documentation, where my changes are highlighted in *red*: using ModelingToolkit using ModelingToolkit: t @mtkmodel ModelA begin @parameters begin k k_array[1:2] end end @mtkmodel ModelB begin @parameters begin p1 = 1.0, [description = "Parameter of ModelB"] p2 = 1.0, [description = "Parameter of ModelB"] end end @mtkmodel ModelC begin # @ICON "https://github.com/SciML/SciMLDocs/blob/main/docs/src/assets/logo.png" @Constants begin c::Int = 1, [description = "Example constant."] end @structural_parameters begin f = sin N = 2 end begin v_var = 1.0 end @variables begin v(t) = v_var v_array(t)[1:2, 1:3] v_for_defaults(t) end @extend ModelB(; p1) @components begin model_a = ModelA(; k_array = [1, 2]) model_array_a = [ModelA(; k = i) for i in 1:N] model_array_b = for i in 1:N k = i^2 ModelA(; k) end end @equations begin model_a.k ~ f(v) model_array_a[1:N].k ~ f(v) end @defaults begin v_for_defaults => 2.0 end end ; The addition of model_array_a[1:N].k ~ f(v) causes the error message below [same error message if I write .~ ]:: @mtkbuild modelc = ModelC() type Array has no field k Stacktrace: [1] getproperty @ .\Base.jl:37<https://untitled+.vscode-resource.vscode-cdn.net/Base.jl:37> [inlined] [2] #__ModelC__#74 @ C:\Users\Bernt\.julia\packages\UnPack\EkESO\src\UnPack.jl:1396 [inlined] [3] __ModelC__(; name::Symbol, v_array::Nothing, p1::Nothing, N::Int64, model_array_a__k::Nothing, f::typeof(sin), v::Nothing, model_a__k_array::Nothing, v_for_defaults::Nothing, model_array_b__k::Nothing) @ Main .\none:0<https://untitled+.vscode-resource.vscode-cdn.net/none:0> [4] (::ModelingToolkit.Model{typeof(__ModelC__), Dict{Symbol, Any}})(; ***@***.***{name::Symbol}) @ ModelingToolkit C:\Users\Bernt\.julia\packages\ModelingToolkit\QP8aH\src\systems\model_parsing.jl:25 [5] top-level scope @ C:\Users\Bernt\.julia\packages\ModelingToolkit\QP8aH\src\systems\abstractsystem.jl:1396

…

________________________________ From: Bernt Lie ***@***.***> Sent: Monday, April 29, 2024 10:25 AM To: SciML/ModelingToolkit.jl ***@***.***>; SciML/ModelingToolkit.jl ***@***.***> Cc: Author ***@***.***> Subject: Re: [SciML/ModelingToolkit.jl] Array of components (Issue #2674) Thanks for answer! Useful, but it only addresses part of my question, and misses the most important part (marked with red): Creation of vectors: The example you refer to illustrates how a vector of models, `model_array_a` and `model_array_b` can be created. I missed that. Yingbo has used a third way, e.g., `model_array_c[1:N] = ModelA()` in his presentations. Perhaps this third way is only suitable when all model instantiations are equal... Reference to vector elements: You write: The relevant bits (from ModelC example): @mtkmodel ModelC begin @structural_parameters begin N = 2 end : @components begin model_a = ModelA(; k_array) model_array_a = [ModelA(; k = i) for i in 1:N] model_array_b = for i in 1:N k = i^2 ModelA(; k) end end : end For @nAmed model_c = ModelC() ,the subcomponents can be then accessed as model_c.model_array_a_1, model_c.model_array_a_2 and so on. This is useful. However, it doesn't answer my question, which was: how do I address the elements of, say, model_a within the ModelC definition. In other words, suppose I add: ... @components begin model_a = ModelA(; k_array) model_array_a = [ModelA(; k = i) for i in 1:N] model_array_b = for i in 1:N k = i^2 ModelA(; k) end end @equations begin for i in 1:N model_array_a[i].k ~ 3 end end : end OK – not a perfect example since `k` is a parameter; in my case it would be a variable. What I have inserted in red above does not work. Should it be `model_array_a_i` instead? Or something else? BR, Bernt

________________________________ From: Christopher Rackauckas ***@***.***> Sent: Monday, April 29, 2024 8:04 AM To: SciML/ModelingToolkit.jl ***@***.***> Cc: Bernt Lie ***@***.***>; Author ***@***.***> Subject: Re: [SciML/ModelingToolkit.jl] Array of components (Issue #2674) Closed #2674<#2674> as completed. — Reply to this email directly, view it on GitHub<#2674 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AM5VM72TFKURHHYMTHC6PTLY7XPFLAVCNFSM6AAAAABG4EHJVWVHI2DSMVQWIX3LMV45UABCJFZXG5LFIV3GK3TUJZXXI2LGNFRWC5DJN5XDWMJSGYZTOOJSGQZTCMQ>. You are receiving this because you authored the thread. T

[B-LIE]

(My color coding disappeared when transferring my e-mail to the GitHub system...)

[cstjean]

For @named model_c = ModelC() ,the subcomponents can be then accessed as model_c.model_array_a_1, model_c.model_array_a_2 and so on.

How can I iterate over that array of components? Eg. I can write

connect(model_c.model_array_a_1, other_component)
connect(model_c.model_array_a_2, other_component)
connect(model_c.model_array_a_3, other_component)
connect(model_c.model_array_a_4, other_component)

but that's terrible. Can I somehow write a for loop?

Context

[ChrisRackauckas]

[connect(model_c.model_array_a[i], other_component) for i in 1:n]

[cstjean]

This does not work for me. MWE:

using ModelingToolkit    # v9.48.0
using ModelingToolkit: t_nounits as t

@mtkmodel MyComponent begin
   @variables begin
      x(t)
   end
   @equations begin
      x ~ 10
   end
end

@mtkmodel Moo begin
   @components begin
      some_comp = MyComponent()
      comp_array = [MyComponent() for i in 1:4]
   end
end

@named moo = Moo()

moo.some_comp      # works
moo.comp_array_1   # works
moo.comp_array     # ArgumentError: System moo: variable comp_array does not exist

This is the same problem highlighted in @B-LIE's comment above (#2674 (comment))

EDIT: first version accidentally used @connector, but the result is the same.

[ven-k]

The component-array is splat before adding them as subsystems. So moo.comp_array won't work. They should be queried along with the number as moo.comp_array_i.

Note that getproperty can be used to query the subsystems via a loop as [getproperty(moo, Symbol(:comp_array_, i)) for i in 1:4]

[cstjean]

Sounds good! Is there any reason why we couldn't have comp_array = [comp_array_1, comp_array_2, ...] automatically defined by the macro? Then the above would just work, instead of having to go through getproperty and Symbol.

[ven-k]

ODESystem constructor doesn't accept Vector{ODESystem} (for subsystem) as input.

Also, [getproperty(moo, Symbol(:comp_array_, i)) for i in 1:4] is meant for accessing the subsystems outside the macro.
The connect statement above, is placed inside macro and macro handles that.

 @mtkmodel Moo begin
          @components begin
             some_comp = MyComponent()
             comp_array = [MyComponent() for i in 1:4]
          end
           @equations begin
               [connect(comp_array[i], some_comp) for i in 1:4]
           end
   end

comp_array = [getproperty(moo, Symbol(:comp_array_, i)) for i in 1:4]

Maybe there can be a helper function to access it. But I fear it will have many non-obvious failing edge cases. If I can come up with a reliable one, I'll add.

[cstjean]

ODESystem constructor doesn't accept Vector{ODESystem} (for subsystem) as input.

That looks like a complicated change indeed.

For context, I wanted to have a component with N connectors. In that case, the connect equation is outside of the @mtkmodel, and it does not work:

using ModelingToolkit
using ModelingToolkit: t_nounits as t

@mtkmodel MyConnector begin
    @variables begin
        x(t)
    end
end

@mtkmodel MilkTank begin
    @components begin
        connections = [MyConnector() for i in 1:4]
    end
end

@mtkmodel Farm begin
    @components begin
        m1 = MilkTank()
	m2 = MilkTank()
    end
    @equations begin
        # (sorry for the meaningless example)
        [connect(co1, co2) for (co1, co2) in zip(m1.connections, m2.connections)]...
    end
end

@named farm = Farm()  # System m1: variable connections does not exist

But on further investigation, stream connectors look like a better solution to my problem, so I don't need multiple connectors. Thank you for the input!