Undefined `fill(::ComplexF64, ::Reactant.MLIR.IR.Type)` method required for differentiating complex function #238

Todorbsc · 2024-11-07T16:12:52Z

CC @mofeing
I get an error when trying to compile a gradient of a function that accepts complex parameters.

julia> using Enzyme

julia> using Reactant
AssertionError("Could not find registered platform with name: \"cuda\". Available platform names are: ")

julia> using Adapt

julia> N = 10
10

julia> params = rand(ComplexF64, N);

julia> expected = rand(ComplexF64, N);

julia> params′ = adapt(ConcreteRArray, params);

julia> expected′ = adapt(ConcreteRArray, expected);

julia> function f1(params, expected)
           return sum(abs.(expected - params))
       end
f1 (generic function with 1 method)

julia> function ∇f(params, expected)
           foo = Enzyme.gradient(ReverseWithPrimal, f1, params, Enzyme.Const(expected))
           return foo.val, foo.derivs[1]
       end
∇f (generic function with 1 method)

julia> ∇fR = Reactant.@compile ∇f(params′, expected′)
ERROR: MethodError: no method matching fill(::ComplexF64, ::Reactant.MLIR.IR.Type)

Closest candidates are:
  fill(::Type{Reactant.MLIR.IR.Attribute}, ::Any, ::Any)
   @ Reactant ~/.julia/packages/Reactant/e7PeE/src/mlir/IR/Attribute.jl:494
  fill(::Any, ::Union{Integer, AbstractUnitRange}...)
   @ Base array.jl:582
  fill(::Int8, ::Reactant.MLIR.IR.Type)
   @ Reactant ~/.julia/packages/Reactant/e7PeE/src/mlir/IR/Attribute.jl:459
  ...

Stacktrace:
  [1] broadcast_to_size(arg::ComplexF64, rsize::Tuple{Int64})
    @ Reactant ~/.julia/packages/Reactant/e7PeE/src/TracedRArray.jl:614
  [2] make_zero(::Type{Reactant.TracedRArray{ComplexF64, 1}}, seen::IdDict{Any, Any}, prev::Reactant.TracedRArray{ComplexF64, 1}, ::Val{false})
    @ Reactant ~/.julia/packages/Reactant/e7PeE/src/Reactant.jl:72
  [3] make_zero (repeats 2 times)
    @ ~/.julia/packages/EnzymeCore/Gdg5y/src/EnzymeCore.jl:524 [inlined]
  [4] macro expansion
    @ ~/.julia/packages/Enzyme/VSRgT/src/Enzyme.jl:1711 [inlined]
  [5] gradient
    @ ~/.julia/packages/Enzyme/VSRgT/src/Enzyme.jl:1661 [inlined]
  [6] ∇f
    @ ./REPL[10]:2 [inlined]
  [7] (::Tuple{})(none::Reactant.TracedRArray{ComplexF64, 1}, none::Reactant.TracedRArray{ComplexF64, 1})
    @ Base.Experimental ./<missing>:0
  [8] (::Reactant.var"#26#35"{Bool, typeof(∇f), Tuple{…}, Vector{…}, Tuple{…}})()
    @ Reactant ~/.julia/packages/Reactant/e7PeE/src/utils.jl:148
  [9] block!(f::Reactant.var"#26#35"{Bool, typeof(∇f), Tuple{…}, Vector{…}, Tuple{…}}, blk::Reactant.MLIR.IR.Block)
    @ Reactant.MLIR.IR ~/.julia/packages/Reactant/e7PeE/src/mlir/IR/Block.jl:201
 [10] make_mlir_fn(f::Function, args::Tuple{…}, kwargs::Tuple{}, name::String, concretein::Bool; toscalar::Bool, return_dialect::Symbol, no_args_in_result::Bool, construct_function_without_args::Bool)
    @ Reactant ~/.julia/packages/Reactant/e7PeE/src/utils.jl:112
 [11] make_mlir_fn
    @ ~/.julia/packages/Reactant/e7PeE/src/utils.jl:36 [inlined]
 [12] #6
    @ ~/.julia/packages/Reactant/e7PeE/src/Compiler.jl:270 [inlined]
 [13] block!(f::Reactant.Compiler.var"#6#11"{typeof(∇f), Tuple{ConcreteRArray{…}, ConcreteRArray{…}}}, blk::Reactant.MLIR.IR.Block)
    @ Reactant.MLIR.IR ~/.julia/packages/Reactant/e7PeE/src/mlir/IR/Block.jl:201
 [14] #5
    @ ~/.julia/packages/Reactant/e7PeE/src/Compiler.jl:269 [inlined]
 [15] mmodule!(f::Reactant.Compiler.var"#5#10"{Reactant.MLIR.IR.Module, typeof(∇f), Tuple{…}}, blk::Reactant.MLIR.IR.Module)
    @ Reactant.MLIR.IR ~/.julia/packages/Reactant/e7PeE/src/mlir/IR/Module.jl:93
 [16] compile_mlir!(mod::Reactant.MLIR.IR.Module, f::Function, args::Tuple{ConcreteRArray{…}, ConcreteRArray{…}}; optimize::Bool)
    @ Reactant.Compiler ~/.julia/packages/Reactant/e7PeE/src/Compiler.jl:266
 [17] compile_mlir!
    @ ~/.julia/packages/Reactant/e7PeE/src/Compiler.jl:265 [inlined]
 [18] (::Reactant.Compiler.var"#30#32"{Bool, typeof(∇f), Tuple{ConcreteRArray{ComplexF64, 1}, ConcreteRArray{ComplexF64, 1}}})()
    @ Reactant.Compiler ~/.julia/packages/Reactant/e7PeE/src/Compiler.jl:720
 [19] context!(f::Reactant.Compiler.var"#30#32"{Bool, typeof(∇f), Tuple{ConcreteRArray{…}, ConcreteRArray{…}}}, ctx::Reactant.MLIR.IR.Context)
    @ Reactant.MLIR.IR ~/.julia/packages/Reactant/e7PeE/src/mlir/IR/Context.jl:76
 [20] compile_xla(f::Function, args::Tuple{ConcreteRArray{ComplexF64, 1}, ConcreteRArray{ComplexF64, 1}}; client::Nothing, optimize::Bool)
    @ Reactant.Compiler ~/.julia/packages/Reactant/e7PeE/src/Compiler.jl:717
 [21] compile_xla
    @ ~/.julia/packages/Reactant/e7PeE/src/Compiler.jl:712 [inlined]
 [22] compile(f::Function, args::Tuple{ConcreteRArray{ComplexF64, 1}, ConcreteRArray{ComplexF64, 1}}; client::Nothing, optimize::Bool, sync::Bool)
    @ Reactant.Compiler ~/.julia/packages/Reactant/e7PeE/src/Compiler.jl:744
 [23] top-level scope
    @ ~/.julia/packages/Reactant/e7PeE/src/Compiler.jl:490
Some type information was truncated. Use `show(err)` to see complete types.

This does NOT happen if the parameters are real:

julia> params_real = rand(N);

julia> expected_real = rand(N);

julia> params_real′ = adapt(ConcreteRArray, params_real);

julia> expected_real′ = adapt(ConcreteRArray, expected_real);

julia> function f1(params, expected)
           return sum(abs.(expected - params))
       end
f1 (generic function with 1 method)

julia> function ∇f(params, expected)
           foo = Enzyme.gradient(ReverseWithPrimal, f1, params, Enzyme.Const(expected))
           return foo.val, foo.derivs[1]
       end
∇f (generic function with 1 method)

julia> ∇fR = Reactant.@compile ∇f(params_real′, expected_real′)
Reactant.Compiler.Thunk{Symbol("##∇f_reactant#225")}()

mofeing · 2024-11-07T20:13:58Z

I think this is related to previous bugs on deviating complex numbers. At least, it's crashing in the same line.

avik-pal · 2024-11-08T16:47:44Z

On main:

julia> @code_hlo optimize=false ∇f(params′, expected′)
Module:
module {
  func.func private @"-_broadcast_scalar"(%arg0: tensor<complex<f64>>, %arg1: tensor<complex<f64>>) -> (tensor<complex<f64>>, tensor<complex<f64>>, tensor<complex<f64>>) {
    %0 = stablehlo.transpose %arg0, dims = [] : (tensor<complex<f64>>) -> tensor<complex<f64>>
    %1 = stablehlo.transpose %arg1, dims = [] : (tensor<complex<f64>>) -> tensor<complex<f64>>
    %2 = stablehlo.subtract %0, %1 : tensor<complex<f64>>
    %3 = stablehlo.transpose %2, dims = [] : (tensor<complex<f64>>) -> tensor<complex<f64>>
    %4 = stablehlo.transpose %0, dims = [] : (tensor<complex<f64>>) -> tensor<complex<f64>>
    %5 = stablehlo.transpose %1, dims = [] : (tensor<complex<f64>>) -> tensor<complex<f64>>
    return %3, %4, %5 : tensor<complex<f64>>, tensor<complex<f64>>, tensor<complex<f64>>
  }
  func.func private @abs_broadcast_scalar(%arg0: tensor<complex<f64>>) -> (tensor<f64>, tensor<complex<f64>>) {
    %0 = stablehlo.transpose %arg0, dims = [] : (tensor<complex<f64>>) -> tensor<complex<f64>>
    %1 = stablehlo.abs %0 : (tensor<complex<f64>>) -> tensor<f64>
    %2 = stablehlo.transpose %1, dims = [] : (tensor<f64>) -> tensor<f64>
    %3 = stablehlo.transpose %0, dims = [] : (tensor<complex<f64>>) -> tensor<complex<f64>>
    return %2, %3 : tensor<f64>, tensor<complex<f64>>
  }
  func.func private @identity_broadcast_scalar(%arg0: tensor<f64>) -> tensor<f64> {
    %0 = stablehlo.transpose %arg0, dims = [] : (tensor<f64>) -> tensor<f64>
    %1 = stablehlo.transpose %0, dims = [] : (tensor<f64>) -> tensor<f64>
    return %1 : tensor<f64>
  }
  func.func private @"Const{typeof(f1)}(Main.f1)_autodiff"(%arg0: tensor<10xcomplex<f64>>, %arg1: tensor<10xcomplex<f64>>) -> (tensor<f64>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>) {
    %0 = stablehlo.transpose %arg0, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %1 = stablehlo.transpose %arg1, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %2 = stablehlo.broadcast_in_dim %1, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %3 = stablehlo.broadcast_in_dim %0, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %4:3 = enzyme.batch @"-_broadcast_scalar"(%2, %3) {batch_shape = array<i64: 10>} : (tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>) -> (tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>)
    %5 = stablehlo.broadcast_in_dim %4#0, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %6:2 = enzyme.batch @abs_broadcast_scalar(%5) {batch_shape = array<i64: 10>} : (tensor<10xcomplex<f64>>) -> (tensor<10xf64>, tensor<10xcomplex<f64>>)
    %cst = stablehlo.constant dense<0.000000e+00> : tensor<f64>
    %7 = stablehlo.broadcast_in_dim %6#0, dims = [0] : (tensor<10xf64>) -> tensor<10xf64>
    %8 = enzyme.batch @identity_broadcast_scalar(%7) {batch_shape = array<i64: 10>} : (tensor<10xf64>) -> tensor<10xf64>
    %9 = stablehlo.reduce(%8 init: %cst) applies stablehlo.add across dimensions = [0] : (tensor<10xf64>, tensor<f64>) -> tensor<f64>
    %10 = stablehlo.transpose %9, dims = [] : (tensor<f64>) -> tensor<f64>
    %11 = stablehlo.transpose %0, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %12 = stablehlo.transpose %1, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    return %10, %11, %12 : tensor<f64>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>
  }
  func.func @main(%arg0: tensor<10xcomplex<f64>>, %arg1: tensor<10xcomplex<f64>>) -> (tensor<f64>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>) {
    %0 = stablehlo.transpose %arg0, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %1 = stablehlo.transpose %arg1, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %cst = stablehlo.constant dense<(0.000000e+00,0.000000e+00)> : tensor<10xcomplex<f64>>
    %cst_0 = stablehlo.constant dense<1.000000e+00> : tensor<f64>
    %2 = stablehlo.transpose %0, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %3 = stablehlo.transpose %1, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %4 = stablehlo.transpose %cst_0, dims = [] : (tensor<f64>) -> tensor<f64>
    %5 = stablehlo.transpose %cst, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %6:4 = enzyme.autodiff @"Const{typeof(f1)}(Main.f1)_autodiff"(%2, %3, %4, %5) {activity = [#enzyme<activity enzyme_active>, #enzyme<activity enzyme_const>], ret_activity = [#enzyme<activity enzyme_active>, #enzyme<activity enzyme_active>, #enzyme<activity enzyme_const>]} : (tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>, tensor<f64>, tensor<10xcomplex<f64>>) -> (tensor<f64>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>)
    %7 = stablehlo.transpose %6#0, dims = [] : (tensor<f64>) -> tensor<f64>
    %8 = stablehlo.transpose %6#1, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %9 = stablehlo.transpose %6#2, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %10 = stablehlo.transpose %6#3, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %11 = stablehlo.transpose %7, dims = [] : (tensor<f64>) -> tensor<f64>
    %12 = stablehlo.transpose %10, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %13 = stablehlo.transpose %8, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %14 = stablehlo.transpose %9, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    return %11, %12, %13, %14 : tensor<f64>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>
  }
}

We are generating something incorrect here:

error: 'complex.add' op operand #0 must be complex type with floating-point elements, but got 'tensor<10xcomplex<f64>>'

mofeing · 2024-11-08T17:30:08Z

I believe the problem is that complex.add only works on complex types... we should change it for stablehlo.add which supports complex tensors

but where is that complex.add happening?

avik-pal · 2024-11-08T17:42:38Z

I was under the impression, stablehlo.add is being lowered to complex.add? (at least from the IR which only has stablehlo.add)

mofeing · 2024-11-08T17:44:55Z

mmm can you show it with optimize=true?

avik-pal · 2024-11-08T17:46:14Z

mmm can you show it with optimize=true?

That errors

avik-pal · 2024-11-08T17:48:49Z

actually nvm, complex.add should be raised to stablehlo https://github.com/EnzymeAD/Enzyme-JAX/blob/a1b5a6048dc358f54c5aedc2b737a3af64a2c80f/src/enzyme_ad/jax/Passes/ArithRaising.cpp#L52

avik-pal · 2024-11-11T19:38:47Z

@wsmoses any idea where #238 (comment) is originating from? Lol you replied too fast on the other issue 😅

wsmoses · 2024-11-11T19:40:07Z

lol yeah if you can spit out it and the pass pipeline we run, we can run it with enzymexlamlir-opt manually and see what’s up

avik-pal · 2024-11-11T19:42:06Z

Module:
module {
  func.func private @"-_broadcast_scalar"(%arg0: tensor<complex<f64>>, %arg1: tensor<complex<f64>>) -> (tensor<complex<f64>>, tensor<complex<f64>>, tensor<complex<f64>>) {
    %0 = stablehlo.transpose %arg0, dims = [] : (tensor<complex<f64>>) -> tensor<complex<f64>>
    %1 = stablehlo.transpose %arg1, dims = [] : (tensor<complex<f64>>) -> tensor<complex<f64>>
    %2 = stablehlo.subtract %0, %1 : tensor<complex<f64>>
    %3 = stablehlo.transpose %2, dims = [] : (tensor<complex<f64>>) -> tensor<complex<f64>>
    %4 = stablehlo.transpose %0, dims = [] : (tensor<complex<f64>>) -> tensor<complex<f64>>
    %5 = stablehlo.transpose %1, dims = [] : (tensor<complex<f64>>) -> tensor<complex<f64>>
    return %3, %4, %5 : tensor<complex<f64>>, tensor<complex<f64>>, tensor<complex<f64>>
  }
  func.func private @abs_broadcast_scalar(%arg0: tensor<complex<f64>>) -> (tensor<f64>, tensor<complex<f64>>) {
    %0 = stablehlo.transpose %arg0, dims = [] : (tensor<complex<f64>>) -> tensor<complex<f64>>
    %1 = stablehlo.abs %0 : (tensor<complex<f64>>) -> tensor<f64>
    %2 = stablehlo.transpose %1, dims = [] : (tensor<f64>) -> tensor<f64>
    %3 = stablehlo.transpose %0, dims = [] : (tensor<complex<f64>>) -> tensor<complex<f64>>
    return %2, %3 : tensor<f64>, tensor<complex<f64>>
  }
  func.func private @identity_broadcast_scalar(%arg0: tensor<f64>) -> tensor<f64> {
    %0 = stablehlo.transpose %arg0, dims = [] : (tensor<f64>) -> tensor<f64>
    %1 = stablehlo.transpose %0, dims = [] : (tensor<f64>) -> tensor<f64>
    return %1 : tensor<f64>
  }
  func.func private @"Const{typeof(f1)}(Main.f1)_autodiff"(%arg0: tensor<10xcomplex<f64>>, %arg1: tensor<10xcomplex<f64>>) -> (tensor<f64>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>) {
    %0 = stablehlo.transpose %arg0, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %1 = stablehlo.transpose %arg1, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %2 = stablehlo.broadcast_in_dim %1, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %3 = stablehlo.broadcast_in_dim %0, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %4:3 = enzyme.batch @"-_broadcast_scalar"(%2, %3) {batch_shape = array<i64: 10>} : (tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>) -> (tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>)
    %5 = stablehlo.broadcast_in_dim %4#0, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %6:2 = enzyme.batch @abs_broadcast_scalar(%5) {batch_shape = array<i64: 10>} : (tensor<10xcomplex<f64>>) -> (tensor<10xf64>, tensor<10xcomplex<f64>>)
    %cst = stablehlo.constant dense<0.000000e+00> : tensor<f64>
    %7 = stablehlo.broadcast_in_dim %6#0, dims = [0] : (tensor<10xf64>) -> tensor<10xf64>
    %8 = enzyme.batch @identity_broadcast_scalar(%7) {batch_shape = array<i64: 10>} : (tensor<10xf64>) -> tensor<10xf64>
    %9 = stablehlo.reduce(%8 init: %cst) applies stablehlo.add across dimensions = [0] : (tensor<10xf64>, tensor<f64>) -> tensor<f64>
    %10 = stablehlo.transpose %9, dims = [] : (tensor<f64>) -> tensor<f64>
    %11 = stablehlo.transpose %0, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %12 = stablehlo.transpose %1, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    return %10, %11, %12 : tensor<f64>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>
  }
  func.func @main(%arg0: tensor<10xcomplex<f64>>, %arg1: tensor<10xcomplex<f64>>) -> (tensor<f64>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>) {
    %0 = stablehlo.transpose %arg0, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %1 = stablehlo.transpose %arg1, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %cst = stablehlo.constant dense<(0.000000e+00,0.000000e+00)> : tensor<10xcomplex<f64>>
    %cst_0 = stablehlo.constant dense<1.000000e+00> : tensor<f64>
    %2 = stablehlo.transpose %0, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %3 = stablehlo.transpose %1, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %4 = stablehlo.transpose %cst_0, dims = [] : (tensor<f64>) -> tensor<f64>
    %5 = stablehlo.transpose %cst, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %6:4 = enzyme.autodiff @"Const{typeof(f1)}(Main.f1)_autodiff"(%2, %3, %4, %5) {activity = [#enzyme<activity enzyme_active>, #enzyme<activity enzyme_const>], ret_activity = [#enzyme<activity enzyme_active>, #enzyme<activity enzyme_active>, #enzyme<activity enzyme_const>]} : (tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>, tensor<f64>, tensor<10xcomplex<f64>>) -> (tensor<f64>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>)
    %7 = stablehlo.transpose %6#0, dims = [] : (tensor<f64>) -> tensor<f64>
    %8 = stablehlo.transpose %6#1, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %9 = stablehlo.transpose %6#2, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %10 = stablehlo.transpose %6#3, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %11 = stablehlo.transpose %7, dims = [] : (tensor<f64>) -> tensor<f64>
    %12 = stablehlo.transpose %10, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %13 = stablehlo.transpose %8, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    %14 = stablehlo.transpose %9, dims = [0] : (tensor<10xcomplex<f64>>) -> tensor<10xcomplex<f64>>
    return %11, %12, %13, %14 : tensor<f64>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>, tensor<10xcomplex<f64>>
  }
}

^this works? Or do you need additional info.

wsmoses · 2024-11-11T19:43:05Z

I think the critical thing here is more so the optimization pass pipeline itself.

avik-pal · 2024-11-11T19:44:52Z

I do see an arith raise pass

Reactant.jl/src/Compiler.jl

Line 291 in f2a91bf

"arith-raise{stablehlo=true}",

avik-pal · 2024-11-11T19:50:41Z

This is the full pipeline

inline{default-pipeline=canonicalize max-iterations=4},canonicalize,cse,canonicalize,enzyme-hlo-generate-td{patterns=compare_op_canon<16>;transpose_transpose<16>;broadcast_in_dim_op_canon<16>;convert_op_canon<16>;dynamic_broadcast_in_dim_op_not_actually_dynamic<16>;chained_dynamic_broadcast_in_dim_canonicalization<16>;dynamic_broadcast_in_dim_all_dims_non_expanding<16>;noop_reduce_op_canon<16>;empty_reduce_op_canon<16>;dynamic_reshape_op_canon<16>;get_tuple_element_op_canon<16>;real_op_canon<16>;imag_op_canon<16>;get_dimension_size_op_canon<16>;gather_op_canon<16>;reshape_op_canon<16>;merge_consecutive_reshapes<16>;transpose_is_reshape<16>;zero_extent_tensor_canon<16>;reorder_elementwise_and_shape_op<16>;cse_broadcast_in_dim<16>;cse_slice<16>;cse_transpose<16>;cse_convert<16>;cse_pad<16>;cse_dot_general<16>;cse_reshape<16>;cse_mul<16>;cse_div<16>;cse_add<16>;cse_subtract<16>;cse_min<16>;cse_max<16>;cse_neg<16>;cse_concatenate<16>;concatenate_op_canon<16>(1024);select_op_canon<16>(1024);add_simplify<16>;sub_simplify<16>;and_simplify<16>;max_simplify<16>;min_simplify<16>;or_simplify<16>;negate_simplify<16>;mul_simplify<16>;div_simplify<16>;rem_simplify<16>;pow_simplify<16>;sqrt_simplify<16>;cos_simplify<16>;sin_simplify<16>;noop_slice<16>;const_prop_through_barrier<16>;slice_slice<16>;shift_right_logical_simplify<16>;pad_simplify<16>;negative_pad_to_slice<16>;tanh_simplify<16>;exp_simplify<16>;slice_simplify<16>;convert_simplify<16>;dynamic_slice_to_static<16>;dynamic_update_slice_elim<16>;concat_to_broadcast<16>;reduce_to_reshape<16>;broadcast_to_reshape<16>;gather_simplify<16>;iota_simplify<16>(1024);broadcast_in_dim_simplify<16>(1024);convert_concat<1>;dynamic_update_to_concat<1>;slice_of_dynamic_update<1>;slice_elementwise<1>;slice_pad<1>;dot_reshape_dot<1>;concat_const_prop<1>;concat_fuse<1>;pad_reshape_pad<1>;pad_pad<1>;concat_push_binop_add<1>;concat_push_binop_mul<1>;scatter_to_dynamic_update_slice<1>;reduce_concat<1>;slice_concat<1>;bin_broadcast_splat_add<1>;bin_broadcast_splat_subtract<1>;bin_broadcast_splat_div<1>;bin_broadcast_splat_mul<1>;reshape_iota<16>;slice_reshape_slice<1>;dot_general_simplify<16>;transpose_simplify<16>;reshape_empty_broadcast<1>;add_pad_pad_to_concat<1>;broadcast_reshape<1>;slice_reshape_concat<1>;slice_reshape_elementwise<1>;slice_reshape_transpose<1>;slice_reshape_dot_general<1>;concat_pad<1>;reduce_pad<1>;broadcast_pad<1>;zero_product_reshape_pad<1>;mul_zero_pad<1>;div_zero_pad<1>;binop_const_reshape_pad<1>;binop_const_pad_add<1>;binop_const_pad_subtract<1>;binop_const_pad_mul<1>;binop_const_pad_div<1>;slice_reshape_pad<1>;binop_binop_pad_pad_add<1>;binop_binop_pad_pad_mul<1>;binop_pad_pad_add<1>;binop_pad_pad_subtract<1>;binop_pad_pad_mul<1>;binop_pad_pad_div<1>;binop_pad_pad_min<1>;binop_pad_pad_max<1>;unary_pad_push_convert<1>;unary_pad_push_tanh<1>;unary_pad_push_exp<1>;transpose_pad<1>;transpose_dot_reorder<1>;dot_transpose<1>;convert_convert_float<1>;concat_to_pad<1>;concat_appending_reshape<1>;reshape_iota<1>;broadcast_reduce<1>;slice_dot_general<1>;dot_reshape_pad<1>;pad_dot_general<1>(0);dot_reshape_pad<1>;pad_dot_general<1>(1)},transform-interpreter,enzyme-hlo-remove-transform,enzyme-batch,inline{default-pipeline=canonicalize max-iterations=4},canonicalize,cse,canonicalize,enzyme-hlo-generate-td{patterns=compare_op_canon<16>;transpose_transpose<16>;broadcast_in_dim_op_canon<16>;convert_op_canon<16>;dynamic_broadcast_in_dim_op_not_actually_dynamic<16>;chained_dynamic_broadcast_in_dim_canonicalization<16>;dynamic_broadcast_in_dim_all_dims_non_expanding<16>;noop_reduce_op_canon<16>;empty_reduce_op_canon<16>;dynamic_reshape_op_canon<16>;get_tuple_element_op_canon<16>;real_op_canon<16>;imag_op_canon<16>;get_dimension_size_op_canon<16>;gather_op_canon<16>;reshape_op_canon<16>;merge_consecutive_reshapes<16>;transpose_is_reshape<16>;zero_extent_tensor_canon<16>;reorder_elementwise_and_shape_op<16>;cse_broadcast_in_dim<16>;cse_slice<16>;cse_transpose<16>;cse_convert<16>;cse_pad<16>;cse_dot_general<16>;cse_reshape<16>;cse_mul<16>;cse_div<16>;cse_add<16>;cse_subtract<16>;cse_min<16>;cse_max<16>;cse_neg<16>;cse_concatenate<16>;concatenate_op_canon<16>(1024);select_op_canon<16>(1024);add_simplify<16>;sub_simplify<16>;and_simplify<16>;max_simplify<16>;min_simplify<16>;or_simplify<16>;negate_simplify<16>;mul_simplify<16>;div_simplify<16>;rem_simplify<16>;pow_simplify<16>;sqrt_simplify<16>;cos_simplify<16>;sin_simplify<16>;noop_slice<16>;const_prop_through_barrier<16>;slice_slice<16>;shift_right_logical_simplify<16>;pad_simplify<16>;negative_pad_to_slice<16>;tanh_simplify<16>;exp_simplify<16>;slice_simplify<16>;convert_simplify<16>;dynamic_slice_to_static<16>;dynamic_update_slice_elim<16>;concat_to_broadcast<16>;reduce_to_reshape<16>;broadcast_to_reshape<16>;gather_simplify<16>;iota_simplify<16>(1024);broadcast_in_dim_simplify<16>(1024);convert_concat<1>;dynamic_update_to_concat<1>;slice_of_dynamic_update<1>;slice_elementwise<1>;slice_pad<1>;dot_reshape_dot<1>;concat_const_prop<1>;concat_fuse<1>;pad_reshape_pad<1>;pad_pad<1>;concat_push_binop_add<1>;concat_push_binop_mul<1>;scatter_to_dynamic_update_slice<1>;reduce_concat<1>;slice_concat<1>;bin_broadcast_splat_add<1>;bin_broadcast_splat_subtract<1>;bin_broadcast_splat_div<1>;bin_broadcast_splat_mul<1>;reshape_iota<16>;slice_reshape_slice<1>;dot_general_simplify<16>;transpose_simplify<16>;reshape_empty_broadcast<1>;add_pad_pad_to_concat<1>;broadcast_reshape<1>;slice_reshape_concat<1>;slice_reshape_elementwise<1>;slice_reshape_transpose<1>;slice_reshape_dot_general<1>;concat_pad<1>;reduce_pad<1>;broadcast_pad<1>;zero_product_reshape_pad<1>;mul_zero_pad<1>;div_zero_pad<1>;binop_const_reshape_pad<1>;binop_const_pad_add<1>;binop_const_pad_subtract<1>;binop_const_pad_mul<1>;binop_const_pad_div<1>;slice_reshape_pad<1>;binop_binop_pad_pad_add<1>;binop_binop_pad_pad_mul<1>;binop_pad_pad_add<1>;binop_pad_pad_subtract<1>;binop_pad_pad_mul<1>;binop_pad_pad_div<1>;binop_pad_pad_min<1>;binop_pad_pad_max<1>;unary_pad_push_convert<1>;unary_pad_push_tanh<1>;unary_pad_push_exp<1>;transpose_pad<1>;transpose_dot_reorder<1>;dot_transpose<1>;convert_convert_float<1>;concat_to_pad<1>;concat_appending_reshape<1>;reshape_iota<1>;broadcast_reduce<1>;slice_dot_general<1>;dot_reshape_pad<1>;pad_dot_general<1>(0);dot_reshape_pad<1>;pad_dot_general<1>(1)},transform-interpreter,enzyme-hlo-remove-transform,enzyme,arith-raise{stablehlo=true},canonicalize,remove-unnecessary-enzyme-ops,enzyme-simplify-math,inline{default-pipeline=canonicalize max-iterations=4},canonicalize,cse,canonicalize,enzyme-hlo-generate-td{patterns=compare_op_canon<16>;transpose_transpose<16>;broadcast_in_dim_op_canon<16>;convert_op_canon<16>;dynamic_broadcast_in_dim_op_not_actually_dynamic<16>;chained_dynamic_broadcast_in_dim_canonicalization<16>;dynamic_broadcast_in_dim_all_dims_non_expanding<16>;noop_reduce_op_canon<16>;empty_reduce_op_canon<16>;dynamic_reshape_op_canon<16>;get_tuple_element_op_canon<16>;real_op_canon<16>;imag_op_canon<16>;get_dimension_size_op_canon<16>;gather_op_canon<16>;reshape_op_canon<16>;merge_consecutive_reshapes<16>;transpose_is_reshape<16>;zero_extent_tensor_canon<16>;reorder_elementwise_and_shape_op<16>;cse_broadcast_in_dim<16>;cse_slice<16>;cse_transpose<16>;cse_convert<16>;cse_pad<16>;cse_dot_general<16>;cse_reshape<16>;cse_mul<16>;cse_div<16>;cse_add<16>;cse_subtract<16>;cse_min<16>;cse_max<16>;cse_neg<16>;cse_concatenate<16>;concatenate_op_canon<16>(1024);select_op_canon<16>(1024);add_simplify<16>;sub_simplify<16>;and_simplify<16>;max_simplify<16>;min_simplify<16>;or_simplify<16>;negate_simplify<16>;mul_simplify<16>;div_simplify<16>;rem_simplify<16>;pow_simplify<16>;sqrt_simplify<16>;cos_simplify<16>;sin_simplify<16>;noop_slice<16>;const_prop_through_barrier<16>;slice_slice<16>;shift_right_logical_simplify<16>;pad_simplify<16>;negative_pad_to_slice<16>;tanh_simplify<16>;exp_simplify<16>;slice_simplify<16>;convert_simplify<16>;dynamic_slice_to_static<16>;dynamic_update_slice_elim<16>;concat_to_broadcast<16>;reduce_to_reshape<16>;broadcast_to_reshape<16>;gather_simplify<16>;iota_simplify<16>(1024);broadcast_in_dim_simplify<16>(1024);convert_concat<1>;dynamic_update_to_concat<1>;slice_of_dynamic_update<1>;slice_elementwise<1>;slice_pad<1>;dot_reshape_dot<1>;concat_const_prop<1>;concat_fuse<1>;pad_reshape_pad<1>;pad_pad<1>;concat_push_binop_add<1>;concat_push_binop_mul<1>;scatter_to_dynamic_update_slice<1>;reduce_concat<1>;slice_concat<1>;bin_broadcast_splat_add<1>;bin_broadcast_splat_subtract<1>;bin_broadcast_splat_div<1>;bin_broadcast_splat_mul<1>;reshape_iota<16>;slice_reshape_slice<1>;dot_general_simplify<16>;transpose_simplify<16>;reshape_empty_broadcast<1>;add_pad_pad_to_concat<1>;broadcast_reshape<1>;slice_reshape_concat<1>;slice_reshape_elementwise<1>;slice_reshape_transpose<1>;slice_reshape_dot_general<1>;concat_pad<1>;reduce_pad<1>;broadcast_pad<1>;zero_product_reshape_pad<1>;mul_zero_pad<1>;div_zero_pad<1>;binop_const_reshape_pad<1>;binop_const_pad_add<1>;binop_const_pad_subtract<1>;binop_const_pad_mul<1>;binop_const_pad_div<1>;slice_reshape_pad<1>;binop_binop_pad_pad_add<1>;binop_binop_pad_pad_mul<1>;binop_pad_pad_add<1>;binop_pad_pad_subtract<1>;binop_pad_pad_mul<1>;binop_pad_pad_div<1>;binop_pad_pad_min<1>;binop_pad_pad_max<1>;unary_pad_push_convert<1>;unary_pad_push_tanh<1>;unary_pad_push_exp<1>;transpose_pad<1>;transpose_dot_reorder<1>;dot_transpose<1>;convert_convert_float<1>;concat_to_pad<1>;concat_appending_reshape<1>;reshape_iota<1>;broadcast_reduce<1>;slice_dot_general<1>;dot_reshape_pad<1>;pad_dot_general<1>(0);dot_reshape_pad<1>;pad_dot_general<1>(1)},transform-interpreter,enzyme-hlo-remove-transform

mofeing · 2024-11-12T16:47:58Z

we have the same problem but with complex.conj not being raised to chlo.conj

Pangoraw · 2024-11-12T17:15:34Z

We probably want to disable the verifier for the pass-manager like in the complex lit tests.

#269.

Todorbsc · 2024-11-12T19:48:58Z

We probably want to disable the verifier for the pass-manager like in the complex lit tests.

#269.

I tested the PR #269 and it solves the complex.add and complex.conj issues.
However, the execution of the MWE crashes with this error now:

julia> @code_hlo ∇f(params′, expected′)
julia: external/llvm-project/llvm/include/llvm/Support/Casting.h:566: decltype(auto) llvm::cast(const From&) [with To = mlir::FloatType; From = mlir::Type]: La declaración `isa<To>(Val) && "cast<Ty>() argument of incompatible type!"' no se cumple.

[12417] signal (6.-6): Abortado
in expression starting at REPL[18]:1
pthread_kill at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
raise at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
abort at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
unknown function (ip: 0x719b8862871a)
__assert_fail at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
cast<mlir::FloatType, mlir::Type> at /proc/self/cwd/external/llvm-project/llvm/include/llvm/Support/Casting.h:566
cast<mlir::FloatType> at /proc/self/cwd/external/llvm-project/mlir/include/mlir/IR/Types.h:352
getConstantAttr at /proc/self/cwd/external/enzyme/Enzyme/MLIR/Implementations/CoreDialectsAutoDiffImplementations.cpp:30
createReverseModeAdjoint at /proc/self/cwd/bazel-out/k8-dbg/bin/external/enzyme_ad/src/enzyme_ad/jax/Implementations/StableHLODerivatives.inc:3182
createReverseModeAdjoint at /proc/self/cwd/bazel-out/k8-dbg/bin/external/enzyme/Enzyme/MLIR/Interfaces/AutoDiffOpInterface.h.inc:381
createReverseModeAdjoint at /proc/self/cwd/bazel-out/k8-dbg/bin/external/enzyme/Enzyme/MLIR/Interfaces/AutoDiffOpInterface.cpp.inc:50
visitChild at /proc/self/cwd/external/enzyme/Enzyme/MLIR/Interfaces/EnzymeLogicReverse.cpp:51
visitChildren at /proc/self/cwd/external/enzyme/Enzyme/MLIR/Interfaces/EnzymeLogicReverse.cpp:66
differentiate at /proc/self/cwd/external/enzyme/Enzyme/MLIR/Interfaces/EnzymeLogicReverse.cpp:174
CreateReverseDiff at /proc/self/cwd/external/enzyme/Enzyme/MLIR/Interfaces/EnzymeLogicReverse.cpp:244
HandleAutoDiffReverse<mlir::enzyme::AutoDiffOp> at /proc/self/cwd/external/enzyme/Enzyme/MLIR/Passes/EnzymeMLIRPass.cpp:271
lowerEnzymeCalls at /proc/self/cwd/external/enzyme/Enzyme/MLIR/Passes/EnzymeMLIRPass.cpp:325
operator() at /proc/self/cwd/external/enzyme/Enzyme/MLIR/Passes/EnzymeMLIRPass.cpp:352
operator() at /proc/self/cwd/external/llvm-project/mlir/include/mlir/IR/Visitors.h:338
callback_fn<mlir::detail::walk<>(mlir::Operation*, (anonymous namespace)::DifferentiatePass::runOnOperation()::<lambda(mlir::FunctionOpInterface)>&&)::<lambda(mlir::Operation*)> > at /proc/self/cwd/external/llvm-project/llvm/include/llvm/ADT/STLFunctionalExtras.h:45
operator() at /proc/self/cwd/external/llvm-project/llvm/include/llvm/ADT/STLFunctionalExtras.h:68
walk<mlir::ForwardIterator> at /proc/self/cwd/external/llvm-project/mlir/include/mlir/IR/Visitors.h:186
walk<mlir::ForwardIterator> at /proc/self/cwd/external/llvm-project/mlir/include/mlir/IR/Visitors.h:181
walk<> at /proc/self/cwd/external/llvm-project/mlir/include/mlir/IR/Visitors.h:340
walk<> at /proc/self/cwd/external/llvm-project/mlir/include/mlir/IR/Operation.h:794
runOnOperation at /proc/self/cwd/external/enzyme/Enzyme/MLIR/Passes/EnzymeMLIRPass.cpp:351
operator() at /proc/self/cwd/external/llvm-project/mlir/lib/Pass/Pass.cpp:526
callback_fn<mlir::detail::OpToOpPassAdaptor::run(mlir::Pass*, mlir::Operation*, mlir::AnalysisManager, bool, unsigned int)::<lambda()> > at /proc/self/cwd/external/llvm-project/llvm/include/llvm/ADT/STLFunctionalExtras.h:45
operator() at /proc/self/cwd/external/llvm-project/llvm/include/llvm/ADT/STLFunctionalExtras.h:68
executeAction<mlir::PassExecutionAction, mlir::Pass&> at /proc/self/cwd/external/llvm-project/mlir/include/mlir/IR/MLIRContext.h:280
run at /proc/self/cwd/external/llvm-project/mlir/lib/Pass/Pass.cpp:520
runPipeline at /proc/self/cwd/external/llvm-project/mlir/lib/Pass/Pass.cpp:592
runPasses at /proc/self/cwd/external/llvm-project/mlir/lib/Pass/Pass.cpp:905
run at /proc/self/cwd/external/llvm-project/mlir/lib/Pass/Pass.cpp:885
mlirPassManagerRunOnOp at /proc/self/cwd/external/llvm-project/mlir/lib/CAPI/IR/Pass.cpp:44
mlirPassManagerRunOnOp at /home/tkrasimi/.julia/dev/Reactant/src/mlir/libMLIR_h.jl:5853 [inlined]
run! at /home/tkrasimi/.julia/dev/Reactant/src/mlir/IR/Pass.jl:74 [inlined]
#run_pass_pipeline!#1 at /home/tkrasimi/.julia/dev/Reactant/src/Compiler.jl:252
run_pass_pipeline! at /home/tkrasimi/.julia/dev/Reactant/src/Compiler.jl:247 [inlined]
#compile_mlir!#8 at /home/tkrasimi/.julia/dev/Reactant/src/Compiler.jl:298
compile_mlir! at /home/tkrasimi/.julia/dev/Reactant/src/Compiler.jl:280 [inlined]
#6 at /home/tkrasimi/.julia/dev/Reactant/src/Compiler.jl:275 [inlined]
context! at /home/tkrasimi/.julia/dev/Reactant/src/mlir/IR/Context.jl:76
unknown function (ip: 0x719b87fdd109)
_jl_invoke at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/gf.c:2894 [inlined]
ijl_apply_generic at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/gf.c:3076
#compile_mlir#5 at /home/tkrasimi/.julia/dev/Reactant/src/Compiler.jl:273
compile_mlir at /home/tkrasimi/.julia/dev/Reactant/src/Compiler.jl:270
unknown function (ip: 0x719b87fdb540)
_jl_invoke at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/gf.c:2894 [inlined]
ijl_apply_generic at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/gf.c:3076
jl_apply at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/julia.h:1982 [inlined]
do_call at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/interpreter.c:126
eval_value at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/interpreter.c:223
eval_stmt_value at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/interpreter.c:174 [inlined]
eval_body at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/interpreter.c:617
jl_interpret_toplevel_thunk at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/interpreter.c:775
jl_toplevel_eval_flex at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/toplevel.c:934
jl_toplevel_eval_flex at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/toplevel.c:877
ijl_toplevel_eval_in at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/toplevel.c:985
eval at ./boot.jl:385 [inlined]
eval_user_input at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/usr/share/julia/stdlib/v1.10/REPL/src/REPL.jl:150
repl_backend_loop at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/usr/share/julia/stdlib/v1.10/REPL/src/REPL.jl:246
#start_repl_backend#46 at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/usr/share/julia/stdlib/v1.10/REPL/src/REPL.jl:231
start_repl_backend at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/usr/share/julia/stdlib/v1.10/REPL/src/REPL.jl:228
_jl_invoke at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/gf.c:2894 [inlined]
ijl_apply_generic at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/gf.c:3076
#run_repl#59 at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/usr/share/julia/stdlib/v1.10/REPL/src/REPL.jl:389
run_repl at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/usr/share/julia/stdlib/v1.10/REPL/src/REPL.jl:375
jfptr_run_repl_91689.1 at /opt/julia-1.10.0/lib/julia/sys.so (unknown line)
_jl_invoke at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/gf.c:2894 [inlined]
ijl_apply_generic at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/gf.c:3076
#1013 at ./client.jl:432
jfptr_YY.1013_82677.1 at /opt/julia-1.10.0/lib/julia/sys.so (unknown line)
_jl_invoke at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/gf.c:2894 [inlined]
ijl_apply_generic at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/gf.c:3076
jl_apply at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/julia.h:1982 [inlined]
jl_f__call_latest at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/builtins.c:812
#invokelatest#2 at ./essentials.jl:887 [inlined]
invokelatest at ./essentials.jl:884 [inlined]
run_main_repl at ./client.jl:416
exec_options at ./client.jl:333
_start at ./client.jl:552
jfptr__start_82703.1 at /opt/julia-1.10.0/lib/julia/sys.so (unknown line)
_jl_invoke at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/gf.c:2894 [inlined]
ijl_apply_generic at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/gf.c:3076
jl_apply at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/julia.h:1982 [inlined]
true_main at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/jlapi.c:582
jl_repl_entrypoint at /cache/build/builder-amdci4-6/julialang/julia-release-1-dot-10/src/jlapi.c:731
main at julia (unknown line)
unknown function (ip: 0x719b88629d8f)
__libc_start_main at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
unknown function (ip: 0x4010b8)
Allocations: 38032866 (Pool: 37988934; Big: 43932); GC: 51
Abortado (`core' generado)

mofeing changed the title ~~Error in Reactant.@compile for complex types parameters~~ Undefined fill(::ComplexF64, ::Reactant.MLIR.IR.Type) method required for differentiating complex function Nov 7, 2024

mofeing added the good first issue Good for newcomers label Nov 7, 2024

avik-pal mentioned this issue Nov 11, 2024

feat: compile neural operators using Reactant SciML/NeuralOperators.jl#52

Open

Pangoraw mentioned this issue Nov 12, 2024

Disable verifier in main pass manager pipeline #269

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Undefined `fill(::ComplexF64, ::Reactant.MLIR.IR.Type)` method required for differentiating complex function #238

Undefined `fill(::ComplexF64, ::Reactant.MLIR.IR.Type)` method required for differentiating complex function #238

Todorbsc commented Nov 7, 2024 •

edited

Loading

mofeing commented Nov 7, 2024

avik-pal commented Nov 8, 2024

mofeing commented Nov 8, 2024 •

edited

Loading

avik-pal commented Nov 8, 2024

mofeing commented Nov 8, 2024

avik-pal commented Nov 8, 2024

avik-pal commented Nov 8, 2024

avik-pal commented Nov 11, 2024 •

edited

Loading

wsmoses commented Nov 11, 2024

avik-pal commented Nov 11, 2024

wsmoses commented Nov 11, 2024

avik-pal commented Nov 11, 2024

avik-pal commented Nov 11, 2024

mofeing commented Nov 12, 2024

Pangoraw commented Nov 12, 2024

Todorbsc commented Nov 12, 2024 •

edited

Loading

Undefined fill(::ComplexF64, ::Reactant.MLIR.IR.Type) method required for differentiating complex function #238

Undefined fill(::ComplexF64, ::Reactant.MLIR.IR.Type) method required for differentiating complex function #238

Comments

Todorbsc commented Nov 7, 2024 • edited Loading

mofeing commented Nov 7, 2024

avik-pal commented Nov 8, 2024

mofeing commented Nov 8, 2024 • edited Loading

avik-pal commented Nov 8, 2024

mofeing commented Nov 8, 2024

avik-pal commented Nov 8, 2024

avik-pal commented Nov 8, 2024

avik-pal commented Nov 11, 2024 • edited Loading

wsmoses commented Nov 11, 2024

avik-pal commented Nov 11, 2024

wsmoses commented Nov 11, 2024

avik-pal commented Nov 11, 2024

avik-pal commented Nov 11, 2024

mofeing commented Nov 12, 2024

Pangoraw commented Nov 12, 2024

Todorbsc commented Nov 12, 2024 • edited Loading

Undefined `fill(::ComplexF64, ::Reactant.MLIR.IR.Type)` method required for differentiating complex function #238

Undefined `fill(::ComplexF64, ::Reactant.MLIR.IR.Type)` method required for differentiating complex function #238

Todorbsc commented Nov 7, 2024 •

edited

Loading

mofeing commented Nov 8, 2024 •

edited

Loading

avik-pal commented Nov 11, 2024 •

edited

Loading

Todorbsc commented Nov 12, 2024 •

edited

Loading