1
0
mirror of https://github.com/kalmarek/PropertyT.jl.git synced 2024-11-26 00:55:27 +01:00

fix: deprecations for julia-0.7/1.0

This commit is contained in:
kalmarek 2019-01-11 06:32:09 +01:00
parent 0ee12f76a5
commit 10d319f48b
7 changed files with 87 additions and 65 deletions

View File

@ -5,11 +5,13 @@ version = "0.1.0"
[deps] [deps]
AbstractAlgebra = "c3fe647b-3220-5bb0-a1ea-a7954cac585d" AbstractAlgebra = "c3fe647b-3220-5bb0-a1ea-a7954cac585d"
Dates = "ade2ca70-3891-5945-98fb-dc099432e06a"
GroupRings = "0befed6a-bd73-11e8-1e41-a1190947c2f5" GroupRings = "0befed6a-bd73-11e8-1e41-a1190947c2f5"
Groups = "5d8bd718-bd84-11e8-3b40-ad14f4a32557" Groups = "5d8bd718-bd84-11e8-3b40-ad14f4a32557"
IntervalArithmetic = "d1acc4aa-44c8-5952-acd4-ba5d80a2a253" IntervalArithmetic = "d1acc4aa-44c8-5952-acd4-ba5d80a2a253"
JLD = "4138dd39-2aa7-5051-a626-17a0bb65d9c8" JLD = "4138dd39-2aa7-5051-a626-17a0bb65d9c8"
JuMP = "4076af6c-e467-56ae-b986-b466b2749572" JuMP = "4076af6c-e467-56ae-b986-b466b2749572"
LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
Markdown = "d6f4376e-aef5-505a-96c1-9c027394607a" Markdown = "d6f4376e-aef5-505a-96c1-9c027394607a"
MathProgBase = "fdba3010-5040-5b88-9595-932c9decdf73" MathProgBase = "fdba3010-5040-5b88-9595-932c9decdf73"
Nemo = "2edaba10-b0f1-5616-af89-8c11ac63239a" Nemo = "2edaba10-b0f1-5616-af89-8c11ac63239a"

View File

@ -1,3 +1,5 @@
using Printf
############################################################################### ###############################################################################
# #
# Settings and filenames # Settings and filenames
@ -84,14 +86,15 @@ end
function computeλandP(sett::Naive, Δ::GroupRingElem; function computeλandP(sett::Naive, Δ::GroupRingElem;
solverlog=tempname()*".log") solverlog=tempname()*".log")
info("Creating SDP problem...") @info("Creating SDP problem...")
info(Base.repr(SDP_problem))
SDP_problem, varλ, varP = SOS_problem(Δ^2, Δ, upper_bound=sett.upper_bound) SDP_problem, varλ, varP = SOS_problem(Δ^2, Δ, upper_bound=sett.upper_bound)
JuMP.setsolver(SDP_problem, sett.solver) JuMP.setsolver(SDP_problem, sett.solver)
@info(Base.repr(SDP_problem))
ws = warmstart(sett) ws = warmstart(sett)
@time status, (λ, P, ws) = PropertyT.solve(solverlog, SDP_problem, varλ, varP, ws) @time status, (λ, P, ws) = PropertyT.solve(solverlog, SDP_problem, varλ, varP, ws)
@show status @info("Solver's status: $status")
save(filename(sett, :warmstart), "warmstart", ws, "P", P, "λ", λ) save(filename(sett, :warmstart), "warmstart", ws, "P", P, "λ", λ)
return λ, P return λ, P
@ -108,17 +111,17 @@ function computeλandP(sett::Symmetrized, Δ::GroupRingElem;
orbit_data = load(filename(sett, :OrbitData), "OrbitData") orbit_data = load(filename(sett, :OrbitData), "OrbitData")
orbit_data = decimate(orbit_data) orbit_data = decimate(orbit_data)
info("Creating SDP problem...") @info("Creating SDP problem...")
info(Base.repr(SDP_problem))
SDP_problem, varλ, varP = SOS_problem(Δ^2, Δ, orbit_data, upper_bound=sett.upper_bound) SDP_problem, varλ, varP = SOS_problem(Δ^2, Δ, orbit_data, upper_bound=sett.upper_bound)
JuMP.setsolver(SDP_problem, sett.solver) JuMP.setsolver(SDP_problem, sett.solver)
@info(Base.repr(SDP_problem))
ws = warmstart(sett) ws = warmstart(sett)
@time status, (λ, Ps, ws) = PropertyT.solve(solverlog, SDP_problem, varλ, varP, ws) @time status, (λ, Ps, ws) = PropertyT.solve(solverlog, SDP_problem, varλ, varP, ws)
@show status @info("Solver's status: $status")
save(filename(sett, :warmstart), "warmstart", ws, "Ps", Ps, "λ", λ) save(filename(sett, :warmstart), "warmstart", ws, "Ps", Ps, "λ", λ)
info("Reconstructing P...") info("Reconstructing P...")
@time P = reconstruct(Ps, orbit_data) @time P = reconstruct(Ps, orbit_data)
@ -146,7 +149,6 @@ function distance_to_positive_cone(Δ::GroupRingElem, λ, Q; R::Int=2)
@info("Floating point distance (to positive cone) ≈") @info("Floating point distance (to positive cone) ≈")
@info("$(@sprintf("%.10f", distance))") @info("$(@sprintf("%.10f", distance))")
@info("")
if distance 0 if distance 0
return distance return distance
@ -154,15 +156,14 @@ function distance_to_positive_cone(Δ::GroupRingElem, λ, Q; R::Int=2)
@info("------------------------------------------------------------") @info("------------------------------------------------------------")
@info("Checking in interval arithmetic...") @info("Checking in interval arithmetic...")
@info("λ ∈ ")
λ = @interval(λ) λ = @interval(λ)
@info("λ ∈ ")
eoi = Δ^2 - λ*Δ eoi = Δ^2 - λ*Δ
@info("Projecting columns of Q to the augmentation ideal...") @info("Projecting columns of Q to the augmentation ideal...")
@time Q, check = augIdproj(Interval, Q) @time Q, check = augIdproj(Interval, Q)
@info("Checking that sum of every column contains 0.0... ") @info("Checking that sum of every column contains 0.0... ")
@info((check? "They do." : "FAILED!")) @info((check ? "They do." : "FAILED!"))
check || @warn("The following numbers are meaningless!") check || @warn("The following numbers are meaningless!")
@time residual = eoi - compute_SOS(parent(eoi), Q) @time residual = eoi - compute_SOS(parent(eoi), Q)
@ -174,7 +175,7 @@ function distance_to_positive_cone(Δ::GroupRingElem, λ, Q; R::Int=2)
@info("Interval distance (to positive cone) ∈") @info("Interval distance (to positive cone) ∈")
@info("$(distance)") @info("$(distance)")
@info("") @info("------------------------------------------------------------")
return distance.lo return distance.lo
end end
@ -192,11 +193,11 @@ function interpret_results(sett::Settings, sgap::Number)
if sgap > 0 if sgap > 0
Kazhdan_κ = Kazhdan(sgap, length(sett.S)) Kazhdan_κ = Kazhdan(sgap, length(sett.S))
if Kazhdan_κ > 0 if Kazhdan_κ > 0
info("κ($(sett.name), S) ≥ $Kazhdan_κ: Group HAS property (T)!") @info("κ($(sett.name), S) ≥ $Kazhdan_κ: Group HAS property (T)!")
return true return true
end end
end end
info("λ($(sett.name), S) ≥ $sgap < 0: Tells us nothing about property (T)") @info("λ($(sett.name), S) ≥ $sgap < 0: Tells us nothing about property (T)")
return false return false
end end
@ -222,20 +223,20 @@ function check_property_T(sett::Settings)
save(filename(sett, :solution), "λ", λ, "P", P) save(filename(sett, :solution), "λ", λ, "P", P)
if λ < 0 if λ < 0
warn("Solver did not produce a valid solution!") @warn("Solver did not produce a valid solution!")
end end
end end
info("λ = ") @info("λ = ")
info("sum(P) = $(sum(P))") @info("sum(P) = $(sum(P))")
info("maximum(P) = $(maximum(P))") @info("maximum(P) = $(maximum(P))")
info("minimum(P) = $(minimum(P))") @info("minimum(P) = $(minimum(P))")
isapprox(eigvals(P), abs.(eigvals(P))) || isapprox(eigvals(P), abs.(eigvals(P))) ||
@warn("The solution matrix doesn't seem to be positive definite!") @warn("The solution matrix doesn't seem to be positive definite!")
@time Q = real(sqrtm((P+P')/2)) @time Q = real(sqrt( (P.+ P')./2 ))
sgap = distance_to_positive_cone(Δ, λ, Q, wlen=2*sett.radius) sgap = distance_to_positive_cone(Δ, λ, Q, R=2*sett.radius)
return interpret_results(sett, sgap) return interpret_results(sett, sgap)
end end

View File

@ -2,6 +2,11 @@ __precompile__()
module PropertyT module PropertyT
using AbstractAlgebra using AbstractAlgebra
using LinearAlgebra
using SparseArrays
using Markdown
using Dates
using Groups using Groups
using GroupRings using GroupRings
@ -17,5 +22,6 @@ include("RGprojections.jl")
include("orbitdata.jl") include("orbitdata.jl")
include("sos_sdps.jl") include("sos_sdps.jl")
include("checksolution.jl") include("checksolution.jl")
include("1712.07167.jl")
end # module Property(T) end # module Property(T)

View File

@ -3,6 +3,7 @@ module Projections
using AbstractAlgebra using AbstractAlgebra
using Groups using Groups
using GroupRings using GroupRings
using Markdown
export PermCharacter, DirectProdCharacter, rankOne_projections export PermCharacter, DirectProdCharacter, rankOne_projections
@ -200,13 +201,13 @@ end
# #
############################################################################## ##############################################################################
doc""" @doc doc"""
products(X::Vector{GroupElem}, Y::Vector{GroupElem}, op=*) products(X::Vector{GroupElem}, Y::Vector{GroupElem}, op=*)
> Returns a vector of all possible products (or `op(x,y)`), where $x\in X$ and > Returns a vector of all possible products (or `op(x,y)`), where $x\in X$ and
> $y\in Y$ are group elements. You may specify which operation is used when > $y\in Y$ are group elements. You may specify which operation is used when
> forming 'products' by adding `op` (which is `*` by default). > forming 'products' by adding `op` (which is `*` by default).
""" """
function products{T<:GroupElem}(X::AbstractVector{T}, Y::AbstractVector{T}, op=*) function products(X::AbstractVector{T}, Y::AbstractVector{T}, op=*) where {T<:GroupElem}
result = Vector{T}() result = Vector{T}()
seen = Set{T}() seen = Set{T}()
for x in X for x in X
@ -221,7 +222,7 @@ function products{T<:GroupElem}(X::AbstractVector{T}, Y::AbstractVector{T}, op=*
return result return result
end end
doc""" @doc doc"""
generateGroup(gens::Vector{GroupElem}, r=2, Id=parent(first(gens))(), op=*) generateGroup(gens::Vector{GroupElem}, r=2, Id=parent(first(gens))(), op=*)
> Produces all elements of a group generated by elements in `gens` in ball of > Produces all elements of a group generated by elements in `gens` in ball of
> radius `r` (word-length metric induced by `gens`). > radius `r` (word-length metric induced by `gens`).
@ -230,7 +231,7 @@ doc"""
> The identity element `Id` and binary operation function `op` can be supplied > The identity element `Id` and binary operation function `op` can be supplied
> to e.g. take advantage of additive group structure. > to e.g. take advantage of additive group structure.
""" """
function generateGroup{T<:GroupElem}(gens::Vector{T}, r=2, Id::T=parent(first(gens))(), op=*) function generateGroup(gens::Vector{T}, r=2, Id::T=parent(first(gens))(), op=*) where {T<:GroupElem}
n = 0 n = 0
R = 1 R = 1
elts = gens elts = gens

View File

@ -33,12 +33,12 @@ function Laplacian(S::Vector{E}, radius) where E<:AbstractAlgebra.GroupElem
end end
function Laplacian(S, Id, radius) function Laplacian(S, Id, radius)
info("Generating metric ball of radius $(2radius)...") @info("Generating metric ball of radius $(2radius)...")
@time E_R, sizes = Groups.generate_balls(S, Id, radius=2radius) @time E_R, sizes = Groups.generate_balls(S, Id, radius=2radius)
info("Generated balls of sizes $sizes.") @info("Generated balls of sizes $sizes.")
info("Creating product matrix...")
@time pm = GroupRings.create_pm(E_R, GroupRings.reverse_dict(E_R), sizes[radius]; twisted=true) @time pm = GroupRings.create_pm(E_R, GroupRings.reverse_dict(E_R), sizes[radius]; twisted=true)
@info("Creating product matrix...")
RG = GroupRing(parent(Id), E_R, pm) RG = GroupRing(parent(Id), E_R, pm)
Δ = spLaplacian(RG, S) Δ = spLaplacian(RG, S)
@ -52,7 +52,7 @@ end
function loadGRElem(fname::String, G::Group) function loadGRElem(fname::String, G::Group)
if isfile(fname) if isfile(fname)
info("Loading precomputed Δ...") @info("Loading precomputed Δ...")
coeffs, pm = load(fname, "coeffs", "pm") coeffs, pm = load(fname, "coeffs", "pm")
RG = GroupRing(G, pm) RG = GroupRing(G, pm)
Δ = GroupRingElem(coeffs, RG) Δ = GroupRingElem(coeffs, RG)

View File

@ -12,25 +12,25 @@ struct OrbitData{T<:AbstractArray{Float64, 2}, GEl<:GroupElem, P<:perm}
end end
function OrbitData(RG::GroupRing, autS::Group, verbose=true) function OrbitData(RG::GroupRing, autS::Group, verbose=true)
verbose && info("Decomposing basis of RG into orbits of $(autS)") verbose && @info("Decomposing basis of RG into orbits of $(autS)")
@time orbs = orbit_decomposition(autS, RG.basis, RG.basis_dict) @time orbs = orbit_decomposition(autS, RG.basis, RG.basis_dict)
@assert sum(length(o) for o in orbs) == length(RG.basis) @assert sum(length(o) for o in orbs) == length(RG.basis)
verbose && info("The action has $(length(orbs)) orbits") verbose && @info("The action has $(length(orbs)) orbits")
verbose && info("Projections in the Group Ring of AutS")
@time autS_mps = Projections.rankOne_projections(GroupRing(autS)) @time autS_mps = Projections.rankOne_projections(GroupRing(autS))
verbose && @info("Projections in the Group Ring of AutS = $autS")
verbose && info("AutS-action matrix representatives") verbose && @info("AutS-action matrix representatives")
@time preps = perm_reps(autS, RG.basis[1:size(RG.pm,1)], RG.basis_dict) @time preps = perm_reps(autS, RG.basis[1:size(RG.pm,1)], RG.basis_dict)
@time mreps = matrix_reps(preps) @time mreps = matrix_reps(preps)
verbose && info("Projection matrices Uπs") verbose && @info("Projection matrices Uπs")
@time Uπs = [orthSVD(matrix_repr(p, mreps)) for p in autS_mps] @time Uπs = [orthSVD(matrix_repr(p, mreps)) for p in autS_mps]
multiplicities = size.(Uπs,2) multiplicities = size.(Uπs,2)
verbose && info("multiplicities = $multiplicities") verbose && @info("multiplicities = $multiplicities")
dimensions = [Int(p[autS()]*Int(order(autS))) for p in autS_mps] dimensions = [Int(p[autS()]*Int(order(autS))) for p in autS_mps]
verbose && info("dimensions = $dimensions") verbose && @info("dimensions = $dimensions")
@assert dot(multiplicities, dimensions) == size(RG.pm,1) @assert dot(multiplicities, dimensions) == size(RG.pm,1)
return OrbitData(orbs, preps, Uπs, dimensions) return OrbitData(orbs, preps, Uπs, dimensions)
@ -43,32 +43,32 @@ function decimate(od::OrbitData)
#dimensions of the corresponding πs: #dimensions of the corresponding πs:
dims = od.dims[nzros] dims = od.dims[nzros]
return OrbitData(od.orbits, od.preps, full.(Us), dims); return OrbitData(od.orbits, od.preps, Array{Float64}.(Us), dims);
end end
function orthSVD(M::AbstractMatrix{T}) where {T<:AbstractFloat} function orthSVD(M::AbstractMatrix{T}) where {T<:AbstractFloat}
M = full(M) M = Matrix(M)
fact = svdfact(M) fact = svd(M)
M_rank = sum(fact[:S] .> maximum(size(M))*eps(T)) M_rank = sum(fact.S .> maximum(size(M))*eps(T))
return fact[:U][:,1:M_rank] return fact.U[:,1:M_rank]
end end
function orbit_decomposition(G::Group, E::Vector, rdict=GroupRings.reverse_dict(E)) function orbit_decomposition(G::Group, E::Vector, rdict=GroupRings.reverse_dict(E))
elts = collect(elements(G)) elts = collect(elements(G))
tovisit = trues(E); tovisit = trues(size(E));
orbits = Vector{Vector{Int}}() orbits = Vector{Vector{Int}}()
orbit = zeros(Int, length(elts)) orbit = zeros(Int, length(elts))
for i in 1:endof(E) for i in eachindex(E)
if tovisit[i] if tovisit[i]
g = E[i] g = E[i]
Threads.@threads for j in 1:length(elts) Threads.@threads for j in eachindex(elts)
orbit[j] = rdict[elts[j](g)] orbit[j] = rdict[elts[j](g)]
end end
tovisit[orbit] = false tovisit[orbit] .= false
push!(orbits, unique(orbit)) push!(orbits, unique(orbit))
end end
end end
@ -82,9 +82,9 @@ end
############################################################################### ###############################################################################
dens(M::SparseMatrixCSC) = nnz(M)/length(M) dens(M::SparseMatrixCSC) = nnz(M)/length(M)
dens(M::AbstractArray) = countnz(M)/length(M) dens(M::AbstractArray) = count(!iszero, M)/length(M)
function sparsify!{Tv,Ti}(M::SparseMatrixCSC{Tv,Ti}, eps=eps(Tv); verbose=false) function sparsify!(M::SparseMatrixCSC{Tv,Ti}, eps=eps(Tv); verbose=false) where {Tv,Ti}
densM = dens(M) densM = dens(M)
for i in eachindex(M.nzval) for i in eachindex(M.nzval)
@ -95,16 +95,16 @@ function sparsify!{Tv,Ti}(M::SparseMatrixCSC{Tv,Ti}, eps=eps(Tv); verbose=false)
dropzeros!(M) dropzeros!(M)
if verbose if verbose
info("Sparsified density:", rpad(densM, 20), "", rpad(dens(M), 20), " ($(nnz(M)) non-zeros)") @info("Sparsified density:", rpad(densM, 20), "", rpad(dens(M), 20), " ($(nnz(M)) non-zeros)")
end end
return M return M
end end
function sparsify!{T}(M::AbstractArray{T}, eps=eps(T); verbose=false) function sparsify!(M::AbstractArray{T}, eps=eps(T); verbose=false) where T
densM = dens(M) densM = dens(M)
if verbose if verbose
info("Sparsifying $(size(M))-matrix... ") @info("Sparsifying $(size(M))-matrix... ")
end end
for n in eachindex(M) for n in eachindex(M)
@ -114,13 +114,15 @@ function sparsify!{T}(M::AbstractArray{T}, eps=eps(T); verbose=false)
end end
if verbose if verbose
info("$(rpad(densM, 20))$(rpad(dens(M),20))), ($(countnz(M)) non-zeros)") @info("$(rpad(densM, 20))$(rpad(dens(M),20))), ($(count(!iszero, M)) non-zeros)")
end end
return sparse(M) return sparse(M)
end end
sparsify{T}(U::AbstractArray{T}, tol=eps(T); verbose=false) = sparsify!(deepcopy(U), tol, verbose=verbose) function sparsify(U::AbstractArray{T}, tol=eps(T); verbose=false) where T
return sparsify!(deepcopy(U), tol, verbose=verbose)
end
############################################################################### ###############################################################################
# #
@ -129,7 +131,7 @@ sparsify{T}(U::AbstractArray{T}, tol=eps(T); verbose=false) = sparsify!(deepcopy
############################################################################### ###############################################################################
function perm_repr(g::GroupElem, E::Vector, E_dict) function perm_repr(g::GroupElem, E::Vector, E_dict)
p = Vector{Int}(length(E)) p = Vector{Int}(undef, length(E))
for (i,elt) in enumerate(E) for (i,elt) in enumerate(E)
p[i] = E_dict[g(elt)] p[i] = E_dict[g(elt)]
end end
@ -139,7 +141,7 @@ end
function perm_reps(G::Group, E::Vector, E_rdict=GroupRings.reverse_dict(E)) function perm_reps(G::Group, E::Vector, E_rdict=GroupRings.reverse_dict(E))
elts = collect(elements(G)) elts = collect(elements(G))
l = length(elts) l = length(elts)
preps = Vector{perm}(l) preps = Vector{perm}(undef, l)
permG = PermutationGroup(length(E)) permG = PermutationGroup(length(E))
@ -151,13 +153,13 @@ function perm_reps(G::Group, E::Vector, E_rdict=GroupRings.reverse_dict(E))
end end
function matrix_repr(x::GroupRingElem, mreps::Dict) function matrix_repr(x::GroupRingElem, mreps::Dict)
nzeros = findn(x.coeffs) nzeros = findall(!iszero, x.coeffs)
return sum(x[i].*mreps[parent(x).basis[i]] for i in nzeros) return sum(x[i].*mreps[parent(x).basis[i]] for i in nzeros)
end end
function matrix_reps(preps::Dict{T,perm{I}}) where {T<:GroupElem, I<:Integer} function matrix_reps(preps::Dict{T,perm{I}}) where {T<:GroupElem, I<:Integer}
kk = collect(keys(preps)) kk = collect(keys(preps))
mreps = Vector{SparseMatrixCSC{Float64, Int}}(length(kk)) mreps = Vector{SparseMatrixCSC{Float64, Int}}(undef, length(kk))
Threads.@threads for i in 1:length(kk) Threads.@threads for i in 1:length(kk)
mreps[i] = AbstractAlgebra.matrix_repr(preps[kk[i]]) mreps[i] = AbstractAlgebra.matrix_repr(preps[kk[i]])
end end

View File

@ -75,7 +75,7 @@ function SOS_problem(X::GroupRingElem, orderunit::GroupRingElem, data::OrbitData
Ns = size.(data.Uπs, 2) Ns = size.(data.Uπs, 2)
m = JuMP.Model(); m = JuMP.Model();
P = Vector{Matrix{JuMP.Variable}}(length(Ns)) P = Vector{Matrix{JuMP.Variable}}(undef, length(Ns))
for (k,s) in enumerate(Ns) for (k,s) in enumerate(Ns)
P[k] = JuMP.@variable(m, [i=1:s, j=1:s]) P[k] = JuMP.@variable(m, [i=1:s, j=1:s])
@ -87,7 +87,7 @@ function SOS_problem(X::GroupRingElem, orderunit::GroupRingElem, data::OrbitData
JuMP.@constraint(m, λ <= upper_bound) JuMP.@constraint(m, λ <= upper_bound)
end end
info("Adding $(length(data.orbits)) constraints... ") @info("Adding $(length(data.orbits)) constraints... ")
@time addconstraints!(m,P,λ,X,orderunit, data) @time addconstraints!(m,P,λ,X,orderunit, data)
@ -96,7 +96,7 @@ function SOS_problem(X::GroupRingElem, orderunit::GroupRingElem, data::OrbitData
end end
function constraintLHS!(M, cnstr, Us, Ust, dims, eps=1000*eps(eltype(first(M)))) function constraintLHS!(M, cnstr, Us, Ust, dims, eps=1000*eps(eltype(first(M))))
for π in 1:endof(Us) for π in eachindex(Us)
M[π] = PropertyT.sparsify!(dims[π].*Ust[π]*cnstr*Us[π], eps) M[π] = PropertyT.sparsify!(dims[π].*Ust[π]*cnstr*Us[π], eps)
end end
end end
@ -112,13 +112,13 @@ function addconstraints!(m::JuMP.Model,
cnstrs = constraints(parent(X).pm) cnstrs = constraints(parent(X).pm)
orb_cnstr = spzeros(Float64, size(parent(X).pm)...) orb_cnstr = spzeros(Float64, size(parent(X).pm)...)
M = [Array{Float64}(n,n) for n in size.(UπsT,1)] M = [Array{Float64}(undef, n,n) for n in size.(UπsT,1)]
for (t, orbit) in enumerate(data.orbits) for (t, orbit) in enumerate(data.orbits)
orbit_constraint!(orb_cnstr, cnstrs, orbit) orbit_constraint!(orb_cnstr, cnstrs, orbit)
constraintLHS!(M, orb_cnstr, data.Uπs, UπsT, data.dims) constraintLHS!(M, orb_cnstr, data.Uπs, UπsT, data.dims)
lhs = @expression(m, sum(vecdot(M[π], P[π]) for π in 1:endof(data.Uπs))) lhs = @expression(m, sum(dot(M[π], P[π]) for π in eachindex(data.Uπs)))
x, u = X_orb[t], orderunit_orb[t] x, u = X_orb[t], orderunit_orb[t]
JuMP.@constraint(m, lhs == x - λ*u) JuMP.@constraint(m, lhs == x - λ*u)
end end
@ -198,8 +198,8 @@ function solve(solverlog::String, model::JuMP.Model, varλ::JuMP.Variable, varP,
isdir(dirname(solverlog)) || mkpath(dirname(solverlog)) isdir(dirname(solverlog)) || mkpath(dirname(solverlog))
Base.flush(Base.stdout)
status, (λ, P, warmstart) = open(solverlog, "a+") do logfile status, (λ, P, warmstart) = open(solverlog, "a+") do logfile
Base.Libc.flush_cstdio()
redirect_stdout(logfile) do redirect_stdout(logfile) do
status, (λ, P, warmstart) = PropertyT.solve(model, varλ, varP, warmstart) status, (λ, P, warmstart) = PropertyT.solve(model, varλ, varP, warmstart)
Base.Libc.flush_cstdio() Base.Libc.flush_cstdio()
@ -224,7 +224,7 @@ function fillfrominternal!(m::JuMP.Model, traits)
m.objBound = NaN m.objBound = NaN
m.objVal = NaN m.objVal = NaN
m.colVal = fill(NaN, numCols) m.colVal = fill(NaN, numCols)
m.linconstrDuals = Array{Float64}(0) m.linconstrDuals = Array{Float64}(undef, 0)
discrete = (traits.int || traits.sos) discrete = (traits.int || traits.sos)
@ -262,7 +262,7 @@ function fillfrominternal!(m::JuMP.Model, traits)
@assert length(infray) == numRows @assert length(infray) == numRows
infray infray
catch catch
suppress_warnings || warn("Infeasibility ray (Farkas proof) not available") @warn("Infeasibility ray (Farkas proof) not available")
fill(NaN, numRows) fill(NaN, numRows)
end end
elseif stat == :Unbounded elseif stat == :Unbounded
@ -271,7 +271,7 @@ function fillfrominternal!(m::JuMP.Model, traits)
@assert length(unbdray) == numCols @assert length(unbdray) == numCols
unbdray unbdray
catch catch
suppress_warnings || warn("Unbounded ray not available") @warn("Unbounded ray not available")
fill(NaN, numCols) fill(NaN, numCols)
end end
end end
@ -292,7 +292,10 @@ function fillfrominternal!(m::JuMP.Model, traits)
# Do a separate try since getobjval could work while getobjbound does not and vice versa # Do a separate try since getobjval could work while getobjbound does not and vice versa
objBound = MathProgBase.getobjbound(m.internalModel) + m.obj.aff.constant objBound = MathProgBase.getobjbound(m.internalModel) + m.obj.aff.constant
m.objBound = objBound m.objBound = objBound
catch
@warn("objBound could not be obtained")
end end
try try
objVal = MathProgBase.getobjval(m.internalModel) + m.obj.aff.constant objVal = MathProgBase.getobjval(m.internalModel) + m.obj.aff.constant
colVal = MathProgBase.getsolution(m.internalModel)[1:numCols] colVal = MathProgBase.getsolution(m.internalModel)[1:numCols]
@ -304,8 +307,15 @@ function fillfrominternal!(m::JuMP.Model, traits)
# Don't corrupt the answers if one of the above two calls fails # Don't corrupt the answers if one of the above two calls fails
m.objVal = objVal m.objVal = objVal
m.colVal = colVal m.colVal = colVal
catch
@warn("objVal/colVal could not be obtained")
end end
end end
if traits.conic && m.objSense == :Max
m.objBound = -1 * (m.objBound - m.obj.aff.constant) + m.obj.aff.constant
m.objVal = -1 * (m.objVal - m.obj.aff.constant) + m.obj.aff.constant
end
return stat return stat
end end