Fix DualObjectiveValue

src/MOI_wrapper.jl

@@ -2142,10 +2142,10 @@ function MOI.get(model::Optimizer, attr::MOI.DualObjectiveValue)
     for (li, i, ui) in zip(lower, set, upper)
         di = model.solution.coldual[i+1]
         if model.solution.has_dual_ray
-            dual_objective_value += sense * ifelse(di <= 0, ui, li) * di
+            dual_objective_value += sense * _active_bound(li, NaN, ui, di) * di
         else
             xi = model.solution.colvalue[i+1]
-            dual_objective_value += _dual_objective_contribution(li, xi, ui, di)
+            dual_objective_value += _active_bound(li, xi, ui, di) * di
         end
     end
     # Row components of the dual objective value
@@ -2171,24 +2171,29 @@ function MOI.get(model::Optimizer, attr::MOI.DualObjectiveValue)
     for (li, i, ui) in zip(lower, set, upper)
         di = model.solution.rowdual[i+1]
         if model.solution.has_dual_ray
-            dual_objective_value += sense * ifelse(di <= 0, ui, li) * di
+            dual_objective_value += sense * _active_bound(li, NaN, ui, di) * di
         else
             ri = model.solution.rowvalue[i+1]
-            dual_objective_value += _dual_objective_contribution(li, ri, ui, di)
+            dual_objective_value += _active_bound(li, ri, ui, di) * di
         end
     end
     return dual_objective_value
 end
 
-function _dual_objective_contribution(l, x, u, d)
+function _active_bound(l, x, u, d)
     if isfinite(l) && isfinite(u)
-        # Pick the bound that is closest to the primal value
-        return ifelse(abs(x - l) < abs(x - u), l, u) * d
+        if isfinite(x)
+            # Pick the bound that is closest to the primal value
+            return ifelse(abs(x - l) < abs(x - u), l, u)
+        else
+            # Pick the bound depending on the sign of the dual value
+            return ifelse(d >= 0, l, u)
+        end
     elseif isfinite(l)
-        return l * d
+        return l
     else
         @assert isfinite(u)
-        return u * d
+        return u
     end
 end
 

test/MOI_wrapper.jl

@@ -974,6 +974,32 @@ function test_callback_interrupt()
     return
 end
 
+function test_active_bound()
+    for ((l, x, u, d), result) in [
+        # Primal exists. Pick closest bound lower
+        (0.0, 0.0, 1.0, 2.0) => 0.0,
+        (0.0, 0.4, 1.0, 2.0) => 0.0,
+        (0.0, 0.4, 1.0, -2.0) => 0.0,  # incorrect d but doesn't matter
+        # Primal exists. Pick closest bound upper
+        (0.0, 1.0, 1.0, -2.0) => 1.0,
+        (0.0, 0.6, 1.0, -2.0) => 1.0,
+        (0.0, 0.6, 1.0, 2.0) => 1.0,  # incorrect d but doesn't matter
+        # It's a ray. Choose based on sign
+        (0.0, NaN, 1.0, 2.0) => 0.0,
+        (0.0, NaN, 1.0, 1e-10) => 0.0,
+        (0.0, NaN, 1.0, -2.0) => 1.0,
+        (0.0, NaN, 1.0, -1e-10) => 1.0,
+        # It's a one-sided ray
+        (0.0, NaN, Inf, 2.0) => 0.0,
+        (0.0, NaN, Inf, -1e-10) => 0.0,
+        (-Inf, NaN, 1.0, 1e-10) => 1.0,
+        (-Inf, NaN, 1.0, -2.0) => 1.0,
+    ]
+        @test HiGHS._active_bound(l, x, u, d) == result
+    end
+    return
+end
+
 end  # module
 
 TestMOIHighs.runtests()