Merge pull request #388 from SebastianM-C/interp

Interpolation with the model macro

Author: ChrisRackauckas

docs/src/tutorials/input_component.md

@@ -35,7 +35,7 @@ using Plots
 
 function MassSpringDamper(; name)
     @named input = RealInput()
-    @variables f(t)=0 x(t)=0 dx(t)=0 ddx(t)=0
+    @variables f(t) x(t)=0 dx(t)=0 ddx(t)
     @parameters m=10 k=1000 d=1
 
     eqs = [f ~ input.u
@@ -74,6 +74,35 @@ sol = solve(prob)
 plot(sol)
 ```
 
+Note that in the case of the `Interpolation` block, the `data` and the `time` act like
+structural parameters.
+
+As such, we can also build the interpolation object outside of the model
+
+```@example interpolation_block
+my_interpolation = LinearInterpolation(df.data, df.time)
+
+@mtkmodel MassSpringDamperSystem2 begin
+    @components begin
+        src = Interpolation(itp=my_interpolation)
+        clk = ContinuousClock()
+        model = MassSpringDamper()
+    end
+    @equations begin
+        connect(src.input, clk.output)
+        connect(src.output, model.input)
+    end
+end;
+@mtkbuild sys = MassSpringDamperSystem2()
+
+prob = ODEProblem(sys, [], (0, df.time[end]))
+sol = solve(prob, Tsit5())
+plot(sol)
+```
+
+Note that the interpolation is constructed outside of the model, so we cannot use `remake` to change the
+data. For that usecase, see the `ParametrizedInterpolation`.
+
 ## `ParametrizedInterpolation` Block
 
 The `ModelingToolkitStandardLibrary.Blocks.ParametrizedInterpolation` component is similar to `Interpolation`, but as the name suggests, it is parametrized by the data, allowing one to change the underlying data without rebuilding the model as the data is represented via vector parameters.
@@ -145,7 +174,7 @@ plot(sol2)
 ```
 
 !!! note
-    
+
     Note that when changing the data, the length of the new data must be the same as the length of the original data.
 
 ## Custom Component with External Data

src/Blocks/sources.jl

@@ -755,10 +755,12 @@ such as `LinearInterpolation`, `ConstantInterpolation` or `CubicSpline`.
 """
 function Interpolation(interp_type, u, x, args...; name)
     itp = interp_type(u, x, args...)
-    Interpolation(itp; name)
+    Interpolation(; itp, name)
 end
 
-function Interpolation(itp; name)
+@deprecate Interpolation(itp; name) Interpolation(; itp, name)
+
+function Interpolation(; itp, name)
     @parameters (interpolator::typeof(itp))(..) = itp
     @named input = RealInput()
     @named output = RealOutput()
@@ -868,3 +870,7 @@ function ParametrizedInterpolation(
         systems = [input, output],
         name)
 end
+
+function ParametrizedInterpolation(; interp_type, u::AbstractVector, x::AbstractVector, name)
+    ParametrizedInterpolation(interp_type, u, x; name)
+end

test/Blocks/sources.jl

@@ -500,6 +500,44 @@ end
     @test SciMLBase.successful_retcode(sol)
 end
 
+@testset "Interpolation in model macro" begin
+
+    function MassSpringDamper(; name)
+        @named input = RealInput()
+        @variables f(t) x(t)=0 dx(t)=0 ddx(t)
+        @parameters m=10 k=1000 d=1
+
+        eqs = [f ~ input.u
+               ddx * 10 ~ k * x + d * dx + f
+               D(x) ~ dx
+               D(dx) ~ ddx]
+
+        ODESystem(eqs, t; name, systems = [input])
+    end
+
+    table_data = [1.0, 2.0, 3.0]
+    table_bkp = [0.0, 0.5, 1.0]
+    itp = LinearInterpolation(table_data, table_bkp)
+
+    @mtkmodel model_with_lut begin
+        @components begin
+            src = Interpolation(itp)
+            clk = ContinuousClock()
+            model = MassSpringDamper()
+        end
+        @equations begin
+            connect(src.input, clk.output)
+            connect(src.output, model.input)
+        end
+    end;
+    @mtkbuild sys = model_with_lut()
+
+    prob = ODEProblem(sys, [], (0.0, 1))
+    sol = solve(prob, Tsit5())
+
+    @test SciMLBase.successful_retcode(sol)
+end
+
 @testset "ParametrizedInterpolation" begin
     @variables y(t) = 0
     u = rand(15)