# Zygote error: 'try/catch is not supported'

**URL:** <https://itensor.discourse.group/t/zygote-error-try-catch-is-not-supported/1006>\
**Category:** ITensor Julia Questions\
**Created:** [July 8, 2023, 9:29am UTC](https://itensor.discourse.group/t/zygote-error-try-catch-is-not-supported/1006 "2023-07-08T09:29:58Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![JoexGibbs](https://yyz2.discourse-cdn.com/free1/user_avatar/itensor.discourse.group/joexgibbs/32/155_2.png) [@JoexGibbs](https://itensor.discourse.group/u/JoexGibbs)\
**Post date:** [July 8, 2023, 9:29am UTC](https://itensor.discourse.group/t/zygote-error-try-catch-is-not-supported/1006/1 "2023-07-08T09:29:58Z")

</div>

Hello,

I would like to use Zygote with ITensor to get gradients via autodiff. I have tried to implement a simple example where my loss function measures the inner product between two MPOs, however I get an error message `Compiling Tuple{typeof(lognorm), MPO}: try/catch is not supported.` It seems like this is due to the ‘inner’ function, but I am not sure why as in the examples folder you use ‘inner’ in the loss function without this error flagging up.

Please can you help resolve this, many thanks.

```auto
using ITensors
using OptimKit
using Random
using Zygote

X = [0 1; 1 0]
XX = kron(X, X)

n = 2

s = siteinds("Qubit", n)

function loss(θ⃗)
    
    mpo_init = MPO(s, "Id")
    mpo1 = apply(op(exp(-im*θ⃗[1]*XX), s[1], s[2]), mpo_init)
    mpo2 = apply(op(exp(-im*1*XX), s[1], s[2]), mpo_init)    
    return -abs(inner(mpo1, mpo2))^2

end

θ⃗ = 2π * rand(1)

@show loss(θ⃗)
@show loss'(θ⃗)

```