Duals and constraint names for `PowerModels` and `InfrastructureModels`

Specific Domains Optimization (Mathematical)
1

I would like to (1) use PowerModels to construct a base DCOPF model and (2) make some modifications to the model (such as allowing load shed by adding new variables and modifying cosntraints). Beginning with an instance

file = "case30.m"
nd = PowerModels.parse_file(file)
pm = instantiate_model(network_data, DCPPowerModel, PowerModels.build_opf)
# ... modify `pm` to add new variables and constraints

is there a way to retrieve constraint names for each constraint and their duals/values? I see report_duals in InfrastructureModels, but I am not sure how to obtain a DCOPF model with constraint names and duals names beginning from my nd instance. Are the arguments I can pass to instantiate_model to do this?

2

In most cases the constraints do not have names and they are not stored, so you cannot access their dual values. (There are some exceptions.)

See these issues:

In most cases, if you want specific control over the subproblem and the duals, you should use PowerModels to parse the data and code your own implementation.

1 Like
3

This is exactly what motivated us to build PGLearn.jl, we similarly wanted 1) duals and 2) the ability to easily inspect/modify the formulations.

In PGLearn.jl each formulation is contained in one file – the one corresponding to PowerModels’ DCPPowerModel is DCOPF in dcp.jl. In that file you’ll find three functions – build_opf for building the model (which has all the constraints written plainly, so you can find/fork+modify them easily), extract_primal to get the primal solution, and extract_dual for the dual solution.

Here’s what your snippet would look like:

using PGLearn, PowerModels

# use PowerModels to parse the data like usual
file = "case30.m"
data = PowerModels.make_basic_network(PowerModels.parse_file(file))

# build the model
opf = PGLearn.build_opf(
  PGLearn.DCOPF,
  PGLearn.OPFData(data),
  nothing,  # optimizer constructor, e.g. Ipopt.Optimizer
)

# solve the model
PGLearn.solve!(opf)

# extract the metadata/primal/dual solutions
results = PGLearn.extract_result(opf)

The documentation is currently quite sparse, focusing mostly on the math. But if it’s a good fit for your use case I’m happy to help you get started using it!

3 Likes
4

Oh great, thanks for the link! I think this should work for my purposes!

BTW, I’m using Julia 1.8 on a mac and had an error building PGLearn

julia> using PowerModels, PGLearn

[ Info: Precompiling PGLearn [5d2523b5-5e96-4b1c-8178-da2b93e9175f]
ERROR: LoadError: UndefVarError: CHOLMOD not defined
Stacktrace:
  [1] getproperty(x::Module, f::Symbol)
    @ Base ./Base.jl:31
  [2] top-level scope
    @ ~/.julia/packages/PGLearn/m0m1y/src/opf/ptdf.jl:8
  [3] include(mod::Module, _path::String)
    @ Base ./Base.jl:419
  [4] include(x::String)
    @ PGLearn ~/.julia/packages/PGLearn/m0m1y/src/PGLearn.jl:1
  [5] top-level scope
    @ ~/.julia/packages/PGLearn/m0m1y/src/opf/opf.jl:304
  [6] include(mod::Module, _path::String)
    @ Base ./Base.jl:419
  [7] include(x::String)
    @ PGLearn ~/.julia/packages/PGLearn/m0m1y/src/PGLearn.jl:1
  [8] top-level scope
    @ ~/.julia/packages/PGLearn/m0m1y/src/PGLearn.jl:24
  [9] include
    @ ./Base.jl:419 [inlined]
 [10] include_package_for_output(pkg::Base.PkgId, input::String, depot_path::Vector{String}, dl_load_path::Vector{String}, load_path::Vector{String}, concrete_deps::Vector{Pair{Base.PkgId, UInt64}}, source::Nothing)
    @ Base ./loading.jl:1554
 [11] top-level scope
    @ stdin:1
in expression starting at /Users/jakeroth/.julia/packages/PGLearn/m0m1y/src/opf/ptdf.jl:8
in expression starting at /Users/jakeroth/.julia/packages/PGLearn/m0m1y/src/opf/opf.jl:304
in expression starting at /Users/jakeroth/.julia/packages/PGLearn/m0m1y/src/PGLearn.jl:1
in expression starting at stdin:1
ERROR: Failed to precompile PGLearn [5d2523b5-5e96-4b1c-8178-da2b93e9175f] to /Users/jakeroth/.julia/compiled/v1.8/PGLearn/jl_8JGlEL.
Stacktrace:
 [1] error(s::String)
   @ Base ./error.jl:35
 [2] compilecache(pkg::Base.PkgId, path::String, internal_stderr::IO, internal_stdout::IO, keep_loaded_modules::Bool)
   @ Base ./loading.jl:1707
 [3] compilecache
   @ ./loading.jl:1651 [inlined]
 [4] _require(pkg::Base.PkgId)
   @ Base ./loading.jl:1337
 [5] _require_prelocked(uuidkey::Base.PkgId)
   @ Base ./loading.jl:1200
 [6] macro expansion
   @ ./loading.jl:1180 [inlined]
 [7] macro expansion
   @ ./lock.jl:223 [inlined]
 [8] require(into::Module, mod::Symbol)
   @ Base ./loading.jl:1144

with verision info:

julia> versioninfo()
Julia Version 1.8.5
Commit 17cfb8e65ea (2023-01-08 06:45 UTC)
Platform Info:
  OS: macOS (x86_64-apple-darwin21.4.0)
  CPU: 8 × Intel(R) Core(TM) i7-7820HQ CPU @ 2.90GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-13.0.1 (ORCJIT, skylake)
  Threads: 1 on 8 virtual cores

I temporarily resolved the issue by changing the LazyPTDF struct field since SparseArrays.CHOLMOD wasn’t recognized for me:


struct LazyPTDF <: AbstractPTDF
    N::Int  # number of buses
    E::Int  # number of branches
    islack::Int  # Index of slack bus

    A::SparseMatrixCSC{Float64,Int}  # incidence matrix
    b::Vector{Float64}  # branch susceptances
    BA::SparseMatrixCSC{Float64,Int}  # B*A
    AtBA::SparseMatrixCSC{Float64,Int}  # AᵀBA

    F::Factorization{Float64}#SparseArrays.CHOLMOD.Factor{Float64, Int64}  # Factorization of AᵀBA
    #=
    julia> using LinearAlgebra;
    julia> using SparseArrays;
    julia> A = sprandn(10,10,0.2);
    julia> A=A'*A+I;
    julia> cholesky(A)
    SuiteSparse.CHOLMOD.Factor{Float64}
    type:    LLt
    method:  simplicial
    maxnnz:  28
    nnz:     28
    success: true
    julia> supertypes(typeof(F))
    (SuiteSparse.CHOLMOD.Factor{Float64}, Factorization{Float64}, Any)
    =#

    # TODO: cache
end
5

Nice catch! Fixing it in Relax LazyPTDF factorization type by klamike · Pull Request #192 · AI4OPT/PGLearn.jl · GitHub

1 Like
6

@jacob-roth, I am a bit late to this thread but I wanted to remark, I think the better solution for your use case is to use one of the from scratch models in PowerModelsAnnex and then adapt to your needs, PowerModelsAnnex.jl/src/model at master · lanl-ansi/PowerModelsAnnex.jl · GitHub

Then you can use JuMP’s functionality for extracting the dual information.

2 Likes