Update reading of QPSData (#22)

update reading with QPSData

* update qps.jl and tests
* add NLPModels 0.13 compat

Author: geoffroyleconte

Project.toml

@@ -14,13 +14,14 @@ SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
 [compat]
 FastClosures = "0.2.1, 0.3.0"
 LinearOperators = "0.7.0, 1"
-NLPModels = "0.12"
+NLPModels = "0.13"
 Requires = "0.3, 0.4, 0.5"
 julia = "^1.0.0"
 
 [extras]
 Printf = "de0858da-6303-5e67-8744-51eddeeeb8d7"
+QPSReader = "10f199a5-22af-520b-b891-7ce84a7b1bd0"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
-test = ["Printf", "Test"]
+test = ["Printf", "Test", "QPSReader"]

src/QuadraticModels.jl

@@ -20,7 +20,7 @@ export QuadraticModel
 include("qpmodel.jl")
 
 function __init__()
-    @require QPSReader = "758ba83c-e923-11e8-036a-3f93d0cb3d0c" include("qps.jl")
+    @require QPSReader = "10f199a5-22af-520b-b891-7ce84a7b1bd0" include("qps.jl")
 end
 
 end # module

src/qps.jl

@@ -1,4 +1,8 @@
-function QuadraticModel(qps::QPSData)
-    QuadraticModel(qps.c, qps.Q, opHermitian(qps.Q), qps.A,
-                   qps.lcon, qps.ucon, qps.lvar, qps.uvar, c0=qps.c0)
+using .QPSReader
+
+function QuadraticModel(qps::QPSData, x0=zeros(qps.nvar))
+  QuadraticModel(qps.c, qps.qrows, qps.qcols, qps.qvals, Arows=qps.arows,
+                 Acols=qps.acols, Avals=qps.avals, lcon=qps.lcon,
+                 ucon=qps.ucon, lvar=qps.lvar, uvar=qps.uvar, c0=qps.c0,
+                 x0=x0)
 end

test/problems/bndqp.jl

@@ -6,7 +6,7 @@ function bndqp_autodiff()
   uvar = [1.0; 1.0]
   lvar = [0.0; 0.0]
 
-  return ADNLPModel(f, x0, lvar=lvar, uvar=uvar, name="bndqp_autodiff")
+  return ADNLPModel(f, x0, lvar, uvar, name="bndqp_autodiff")
 end
 
 function bndqp_QP()
@@ -18,3 +18,17 @@ function bndqp_QP()
 
   return QuadraticModel(c, H, lvar=lvar, uvar=uvar, x0=x0, name="bndqp_QP")
 end
+
+function bndqp_QPSData()
+  c    = [1.0; 1.0]
+  H    = [-2.0 0.0; 3.0 4.0]
+  uvar = [1.0; 1.0]
+  lvar = [0.0; 0.0]
+  x0   = [0.5; 0.5]
+  qps = QPSData()
+  qps.c = c
+  qps.qrows, qps.qcols, qps.qvals = findnz(sparse(H))
+  qps.lvar, qps.uvar = lvar, uvar
+  qps.nvar = length(x0)
+  return QuadraticModel(qps, x0)
+end

test/problems/eqconqp.jl

@@ -8,7 +8,7 @@ function eqconqp_autodiff()
   lcon = [0.0]
   ucon = [0.0]
 
-  return ADNLPModel(f, x0, c=c, lcon=lcon, ucon=ucon, name="eqconqp_autodiff")
+  return ADNLPModel(f, x0, c, lcon, ucon, name="eqconqp_autodiff")
 end
 
 function eqconqp_QP()
@@ -21,3 +21,22 @@ function eqconqp_QP()
 
   return QuadraticModel(c, H, A=A, lcon=lcon, ucon=ucon, name="eqconqp_QP")
 end
+
+function eqconqp_QPSData()
+  n    = 50
+  c    = zeros(n)
+  H    = spdiagm(0 => 1:n)
+  A    = ones(1, n)
+  lcon = [1.0]
+  ucon = [1.0]
+  lvar = [-Inf for i=1:n]
+  uvar = [Inf for i=1:n]
+  qps = QPSData()
+  qps.c = c
+  qps.qrows, qps.qcols, qps.qvals = findnz(H)
+  qps.arows, qps.acols, qps.avals = findnz(sparse(A))
+  qps.lcon, qps.ucon = lcon, ucon
+  qps.lvar, qps.uvar = lvar, uvar
+  qps.nvar = length(c)
+  return QuadraticModel(qps)
+end

test/problems/ineqconqp.jl

@@ -7,7 +7,7 @@ function ineqconqp_autodiff()
   lcon = [0.0; -Inf; -1.0]
   ucon = [Inf;  0.0;  1.0]
 
-  return ADNLPModel(f, x0, c=c, lcon=lcon, ucon=ucon, name="ineqconqp_autodiff")
+  return ADNLPModel(f, x0, c, lcon, ucon, name="ineqconqp_autodiff")
 end
 
 function ineqconqp_QP()
@@ -25,3 +25,28 @@ function ineqconqp_QP()
 
   return QuadraticModel(c, Hrows, Hcols, Hvals, Arows=Arows, Acols=Acols, Avals=Avals, lcon=lcon, ucon=ucon, c0=c0, x0=x0, name="ineqconqp_QP")
 end
+
+function ineqconqp_QPSData()
+  c     = -ones(2)
+  Hrows = [1, 2]
+  Hcols = [1, 2]
+  Hvals = ones(2)
+  Arows = [1, 1, 2, 2, 3, 3]
+  Acols = [1, 2, 1, 2, 1, 2]
+  Avals = [1.0; -1.0; -1.0; 1.0; 1.0; 1.0]
+  c0    = 1.0
+  lcon  = [0.0; -Inf; -1.0]
+  ucon  = [Inf;  0.0;  1.0]
+  x0    = ones(2)
+  lvar = [-Inf; -Inf]
+  uvar = [Inf; Inf]
+  qps = QPSData()
+  qps.c = c
+  qps.c0 = c0
+  qps.qrows, qps.qcols, qps.qvals = Hrows, Hcols, Hvals
+  qps.arows, qps.acols, qps.avals = Arows, Acols, Avals
+  qps.lcon, qps.ucon = lcon, ucon
+  qps.lvar, qps.uvar = lvar, uvar
+  qps.nvar = length(x0)
+  return QuadraticModel(qps, x0)
+end

test/problems/uncqp.jl

@@ -17,3 +17,21 @@ function uncqp_QP()
 
   return QuadraticModel(c, Hrows, Hcols, Hvals, c0=c0, x0=x0, name="uncqp_QP")
 end
+
+function uncqp_QPSData()
+  c     = [3.0; -2.0]
+  Hrows = [1, 2, 2]
+  Hcols = [1, 1, 2]
+  Hvals = [8.0; -1.0; 10.0]
+  lvar = [-Inf; -Inf]
+  uvar = [Inf; Inf]
+  c0    = 1.0
+  x0    = [-1.2; 1.0]
+  qps = QPSData()
+  qps.c = c
+  qps.qrows, qps.qcols, qps.qvals = Hrows, Hcols, Hvals
+  qps.c0 = c0
+  qps.lvar, qps.uvar = lvar, uvar
+  qps.nvar = length(x0)
+  return QuadraticModel(qps, x0)
+end

test/runtests.jl

@@ -2,7 +2,7 @@
 using Printf, SparseArrays, Test
 
 # our packages
-using LinearAlgebra, LinearOperators, NLPModels, QuadraticModels
+using LinearAlgebra, LinearOperators, NLPModels, QPSReader, QuadraticModels
 
 nlpmodels_path = joinpath(dirname(pathof(NLPModels)), "..", "test")
 nlpmodels_problems_path = joinpath(nlpmodels_path, "problems")

test/test_consistency.jl

@@ -34,7 +34,8 @@ for problem in qp_problems
   @info "Checking consistency of problem $problem"
   nlp_ad = eval(Symbol(problem * "_autodiff"))()
   nlp_qm = eval(Symbol(problem * "_QP"))()
-  nlps = [nlp_ad, nlp_qm]
+  nlp_qps = eval(Symbol(problem * "_QPSData"))()
+  nlps = [nlp_ad, nlp_qm, nlp_qps]
   consistent_nlps(nlps)
   @info "  Consistency checks ✓"
 end
@@ -50,11 +51,11 @@ for problem in [:brownden, :hs5, :hs6, :hs10, :hs11, :hs14, :lincon]
   nlp_ad = if nlp.meta.ncon > 0
     cx, Ax = cons(nlp, x), jac(nlp, x)
     ADNLPModel(s -> fx + dot(gx, s) + dot(s, Hx * s) / 2, zeros(nlp.meta.nvar),
-               lvar=nlp.meta.lvar - x, uvar=nlp.meta.uvar - x,
-               c=s -> Ax * s, lcon=nlp.meta.lcon - cx, ucon=nlp.meta.ucon - cx)
+               nlp.meta.lvar - x, nlp.meta.uvar - x,
+               s -> Ax * s, nlp.meta.lcon - cx, nlp.meta.ucon - cx)
   else
     ADNLPModel(s -> fx + dot(gx, s) + dot(s, Hx * s) / 2, zeros(nlp.meta.nvar),
-               lvar=nlp.meta.lvar - x, uvar=nlp.meta.uvar - x)
+               nlp.meta.lvar - x, nlp.meta.uvar - x)
   end
   nlp_qm = QuadraticModel(nlp, x)
   nlps = [nlp_ad, nlp_qm]