diff --git a/src/api/ml/z3.mli b/src/api/ml/z3.mli
index 6b797c552..252cabb7b 100644
--- a/src/api/ml/z3.mli
+++ b/src/api/ml/z3.mli
@@ -2862,6 +2862,9 @@ sig
   val get_decls : model -> FuncDecl.func_decl list
 
   (** Evaluates an expression in the current model.
+      The Boolean argument indicates whether to apply model completion.
+      When model completion is true it will assign an interpretation for
+      constants and functions that do not have an interpretation in the model.
 
       This function may fail if the argument contains quantifiers,
       is partial (MODEL_PARTIAL enabled), or if it is not well-sorted.
diff --git a/src/sat/sat_solver.cpp b/src/sat/sat_solver.cpp
index a788f7ee7..7834f0541 100644
--- a/src/sat/sat_solver.cpp
+++ b/src/sat/sat_solver.cpp
@@ -3974,7 +3974,8 @@ namespace sat {
         double multiplier = m_config.m_reward_offset * (is_sat ? m_config.m_reward_multiplier : 1.0);
         for (unsigned i = qhead; i < m_trail.size(); ++i) {
             auto v = m_trail[i].var();
-            auto reward = multiplier / (m_stats.m_conflict - m_last_conflict[v] + 1);            
+            auto d = std::max(1ull,m_stats.m_conflict - m_last_conflict[v] + 1);
+            auto reward = multiplier / d;            
             auto activity = m_activity[v];
             set_activity(v, static_cast<unsigned>(m_step_size * reward + ((1.0 - m_step_size) * activity)));
         }