Tangi Migot comments

Results 227 comments of


                                            Tangi Migot

`ERROR: LinearOperatorException("storage types cannot be promoted to a concrete type")`

I am not sure, need to investigate further. It is the product of both that fails, the product of `H` by itself or `cg_op` seems to work.

`ERROR: LinearOperatorException("storage types cannot be promoted to a concrete type")`

Oh, okay, maybe their is an error in the Hessian operator... here are the storage types: ``` julia> LinearOperators.storage_type(cg_op) CuArray{Float64, 1, CUDA.Mem.DeviceBuffer} julia> LinearOperators.storage_type(H) Vector{Float64} (alias for Array{Float64, 1}) ```

`ERROR: LinearOperatorException("storage types cannot be promoted to a concrete type")`

It is in NLPModels.jl https://github.com/JuliaSmoothOptimizers/NLPModels.jl/blob/e875390eba560e29982b6ce87b95c8a83393c26a/src/nlp/api.jl#L1221

Add support to dense Jacobian and Hessian in API

Superseded by #458

Error for quadratic function with `hessian_reverse`

Thanks @mcabbott and @ToucheSir for the feedback. From the documentation, I suspected some things might go wrong, but thought it's good to have some feedback and the example is simpler...

Combine this package with SparseMatricesCOO

Agreed, this is connected to https://github.com/JuliaSmoothOptimizers/NLPModels.jl/issues/389

`LoadError: MethodError: no method matching _sparse(::SymbolicUtils.BasicSymbolic{Any}, ::Int64)`

@aravindh-krishnamoorthy you are right, let me reformulate the example. The error is the same if the function is differentiable as follows: ``` using Symbolics f(x) = Base.ifelse( x[1]

Gradient error on simple forward/reverse comparison

Hi @wsmoses , I am not sure if it is related, but I also had a `Duplicated Returns not yet handled` on this example ( https://github.com/JuliaSmoothOptimizers/OptimizationProblems.jl/blob/main/src/ADNLPProblems/elec.jl ) that might be...

Gradient error on simple forward/reverse comparison

Oh, I see. Thanks!

Update JuMP syntax to use the new nonlinear interface

Thanks a lot @frapac for the PR. NLPModelsJuMP still doesn't support the quadratic constraints, we lagged behind probably because we did not have a big usage of it. With the...