Add lin_nnzj and nln_nnzj and docstrings for functions

src/nlp/api.jl

@@ -1344,6 +1344,30 @@ function hess_op!(
   return LinearOperator{T}(nlp.meta.nvar, nlp.meta.nvar, true, true, prod!, prod!, prod!)
 end
 
+"""
+    varscale(model::AbstractNLPModel)
+
+Return a vector containing the scaling factors for each variable in the model.
+This is typically used to normalize variables for numerical stability in solvers.
+
+By default, the scaling is model-dependent. If not overridden by the model, a vector of ones 
+is returned. Inspired by the AMPL scaling conventions.
+"""
 function varscale end
+
+"""
+    lagscale(model::AbstractNLPModel)
+
+Return a vector of scaling factors for the Lagrange multipliers associated with constraints.
+This can be used to improve numerical stability or condition number when solving KKT systems.
+"""
 function lagscale end
+
+"""
+    conscale(model::AbstractNLPModel)
+
+Return a vector of constraint scaling factors for the model.
+These are typically used to normalize constraints to have similar magnitudes and improve 
+convergence behavior in nonlinear solvers.
+"""
 function conscale end

src/nlp/show.jl

@@ -69,37 +69,47 @@ end
 Describe `meta` for the `show` function.
 """
 function lines_of_description(m::AbstractNLPModelMeta)
-  V = [
-    length(m.ifree),
-    length(m.ilow),
-    length(m.iupp),
-    length(m.irng),
-    length(m.ifix),
-    length(m.iinf),
-  ]
-  V = [sum(V); V]
-  S = ["All variables", "free", "lower", "upper", "low/upp", "fixed", "infeas"]
-  varlines = lines_of_hist(S, V)
-  push!(varlines, sparsityline("nnzh", m.nnzh, m.nvar * (m.nvar + 1) / 2))
-
-  V = [
-    length(m.jfree),
-    length(m.jlow),
-    length(m.jupp),
-    length(m.jrng),
-    length(m.jfix),
-    length(m.jinf),
-  ]
-  V = [sum(V); V]
-  S = ["All constraints", "free", "lower", "upper", "low/upp", "fixed", "infeas"]
-  conlines = lines_of_hist(S, V)
-  push!(conlines, histline("linear", m.nlin, m.ncon), histline("nonlinear", m.nnln, m.ncon))
-  push!(conlines, sparsityline("nnzj", m.nnzj, m.nvar * m.ncon))
-
-  append!(varlines, repeat([" "^length(varlines[1])], length(conlines) - length(varlines)))
-  lines = varlines .* conlines
-
-  return lines
+    V = [
+        length(m.ifree),
+        length(m.ilow),
+        length(m.iupp),
+        length(m.irng),
+        length(m.ifix),
+        length(m.iinf),
+    ]
+    V = [sum(V); V]
+    S = ["All variables", "free", "lower", "upper", "low/upp", "fixed", "infeas"]
+    varlines = lines_of_hist(S, V)
+    push!(varlines, sparsityline("nnzh", m.nnzh, m.nvar * (m.nvar + 1) ÷ 2))
+
+    V = [
+        length(m.jfree),
+        length(m.jlow),
+        length(m.jupp),
+        length(m.jrng),
+        length(m.jfix),
+        length(m.jinf),
+    ]
+    V = [sum(V); V]
+    S = ["All constraints", "free", "lower", "upper", "low/upp", "fixed", "infeas"]
+    conlines = lines_of_hist(S, V)
+
+    # Extract lin_nnzj and nln_nnzj if available
+    lin_nnzj = try getfield(m, :lin_nnzj) catch; 0 end
+    nln_nnzj = try getfield(m, :nln_nnzj) catch; 0 end
+
+    append!(conlines, [
+        histline("linear", m.nlin, m.ncon),
+        histline("nonlinear", m.nnln, m.ncon),
+        sparsityline("nnzj", m.nnzj, m.nvar * m.ncon),
+        sparsityline("lin_nnzj", lin_nnzj, m.nlin * m.nvar),
+        sparsityline("nln_nnzj", nln_nnzj, m.nnln * m.nvar),
+    ])
+
+    append!(varlines, repeat([" "^length(varlines[1])], length(conlines) - length(varlines)))
+    lines = varlines .* conlines
+
+    return lines
 end
 
 function Base.show(io::IO, m::AbstractNLPModelMeta)