# DiffEqFlux.initial\_params not defined

**URL:** <https://discourse.julialang.org/t/diffeqflux-initial-params-not-defined/100156>\
**Category:** General Usage\
**Created:** [June 11, 2023, 2:42am UTC](https://discourse.julialang.org/t/diffeqflux-initial-params-not-defined/100156 "2023-06-11T02:42:26Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![Reza\_AO](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/reza_ao/32/51137_2.png) [@Reza\_AO](https://discourse.julialang.org/u/Reza_AO)\
**Post date:** [June 11, 2023, 2:42am UTC](https://discourse.julialang.org/t/diffeqflux-initial-params-not-defined/100156/1 "2023-06-11T02:42:27Z")

</div>

Hi  
I want to run a PINN code on my GPU. I used to use the code below to run the code on my GPU and I also want to use QuasiRanodmTraining.

```julia
initθ = DiffEqFlux.initial_params(chain) |> gpu

```

I get the error below:

> ERROR: UndefVarError: `initial_params` not defined

I had never had this problem before I reinstalled my Julia and its packages.  
Any Idea how to solve this?  
I don’t want to use GridTraining and Quadrature training since they take longer.

---

<div class="post-metadata">

**Author:** ![ChrisRackauckas](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/chrisrackauckas/32/77_2.png) [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)\
**Post date:** [June 11, 2023, 3:43am UTC](https://discourse.julialang.org/t/diffeqflux-initial-params-not-defined/100156/2 "2023-06-11T03:43:26Z")

</div>

What tutorial are you looking at? That function hasn’t existed for a long time. Are you looking at the documentation here?

[https://docs.sciml.ai/NeuralPDE/dev/](https://docs.sciml.ai/NeuralPDE/dev/)

---

<div class="post-metadata">

**Author:** ![Reza\_AO](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/reza_ao/32/51137_2.png) [@Reza\_AO](https://discourse.julialang.org/u/Reza_AO)\
**Post date:** [June 12, 2023, 10:37pm UTC](https://discourse.julialang.org/t/diffeqflux-initial-params-not-defined/100156/3 "2023-06-12T22:37:05Z")

</div>

Yes, It was there way back then. What is its substitute\>? I want to define a system of PDE in multi neural networks and I don’t want to use QuadratureTraining since it doesn’t run on GPU.

---

<div class="post-metadata">

**Author:** ![ChrisRackauckas](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/chrisrackauckas/32/77_2.png) [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)\
**Post date:** [June 13, 2023, 2:08am UTC](https://discourse.julialang.org/t/diffeqflux-initial-params-not-defined/100156/4 "2023-06-13T02:08:20Z")

</div>

DiffEqFlux’s FastChain was replaced by the library Lux.jl which fully fledged out its explicit neural network form.

[https://lux.csail.mit.edu/stable/](https://lux.csail.mit.edu/stable/)

If you look at the NeuralPDE docs you’ll see all tutorials use Lux.

[https://docs.sciml.ai/NeuralPDE/dev/tutorials/pdesystem/](https://docs.sciml.ai/NeuralPDE/dev/tutorials/pdesystem/)