KeyError when solving Parameter Identification problem using NeuralPDE.jl #829
Labels
Comments
|
It is working for me: using NeuralPDE, DomainSets
@parameters x,y, λ
@variables u(..)
Dxx = Differential(x)^2
Dyy = Differential(y)^2
frequency_of_wave = 8
eq = Dxx(u(x, y)) + Dyy(u(x, y)) + (frequency_of_wave^2) * λ * u(x,y) ~ 0
bcs = [u(0, y) ~ 0.0, u(1, y) ~ 0.0, u(x, 0) ~ 0.0, u(x, 1) ~ 0.0]
domains = [x ∈ Interval(0.0, 1.0), y ∈ Interval(0.0, 1.0)]
@named pde_system = PDESystem(eq, bcs, domains, [x,y], [u(x,y)], [λ])output: julia> @named pde_system = PDESystem(eq, bcs, domains, [x,y], [u(x,y)], [λ])
PDESystem
Equations: Equation[Differential(y)(Differential(y)(u(x, y))) + Differential(x)(Differential(x)(u(x, y))) + 64u(x, y)*λ ~ 0]
Boundary Conditions: Equation[u(0, y) ~ 0.0, u(1, y) ~ 0.0, u(x, 0) ~ 0.0, u(x, 1) ~ 0.0]
Domain: Symbolics.VarDomainPairing[Symbolics.VarDomainPairing(x, 0.0 .. 1.0), Symbolics.VarDomainPairing(y, 0.0 .. 1.0)]
Dependent Variables: Num[u(x, y)]
Independent Variables: Num[x, y]
Parameters: Num[λ]
Default Parameter ValuesDict{Any, Any}()Can you post the exact stack trace? |
|
|
Try giving a default value for lambda: @named pde_system = PDESystem(eq, bcs, domains, [x,y], [u(x,y)], [λ]; defaults = Dict(λ => 1.0)) |
|
That leads to this error: |
|
What line? |
|
After I changed |
|
Try: chain = [Lux.Chain(Dense(dim, 3, Lux.σ), Dense(3, 2))]instead of chain = Lux.Chain(Dense(dim, 3, Lux.σ), Dense(3, 2))BTW, I feel the output dimension should be 1 instead of 2. |
I had some time and was looking into it. This seems to work for me. We can probably do something like returning a default value here (like |
This worked for me. Why does making it an array help? Is it because of the following: chain: a vector of Lux/Flux chains with a d-dimensional input and a 1-dimensional output corresponding to each of the dependent variables (gotten from https://docs.sciml.ai/NeuralPDE/stable/manual/pinns/). Regarding this,
should I do a pull request and change it such that it returns a default? |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hi, I am getting the following error when trying to solve the inverse problem (learning \lambda) using NeuralPDE.jl: "
KeyError: key λ not found" and it points towards the line@named pde_system = PDESystem(eq, bcs, domains, [x,y], [u(x,y)], [λ]).My code is as follows:The text was updated successfully, but these errors were encountered: