Update app.py

app.py

@@ -1338,17 +1338,24 @@ def _run_problem(axl_def:AXLProblemDef):
     t0=time.time()
     try:
         base_prob=build_base_problem(axl_def)
-        binding,ce,rounds=DEBATE_ENGINE.debate(axl_def,base_prob)
+        binding, domain_ce, rounds=DEBATE_ENGINE.debate(axl_def,base_prob)
         elapsed=round(time.time()-t0,2)
+        
         survived=[t for t in DEBATE_ENGINE.all_traces if t.survived]
         failed=[t for t in DEBATE_ENGINE.all_traces if not t.survived]
         best_trace=min((t for t in survived),key=lambda t:t.ce_after,default=None)
+        
+        # FIX: Calculate actual CE against the original base problem
+        actual_base_ce = base_prob.constraint_energy(binding)
+        is_truly_solved = len(survived) > 0
+
         record={
             "problem":axl_def.name,
             "description":axl_def.description,
             "axioms":sorted(axl_def.axioms),
-            "ce":round(ce,6),
-            "solved":ce<SOLVE_THRESHOLD,
+            "ce":round(actual_base_ce, 6),        # Use REAL energy
+            "domain_ce":round(domain_ce, 6),      # Keep domain CE for reference
+            "solved":is_truly_solved,             # ONLY true if verified
             "elapsed":elapsed,
             "binding":{k:round(v,5) for k,v in binding.items()},
             "traces":[{
@@ -1412,11 +1419,17 @@ def _render_collapse(record:Dict) -> str:
     b=record["binding"]
     traces=record["traces"]
     survived=[t for t in traces if t["survived"]]
+    
+    # FIX: If nothing survived, pick the best attempt to show WHY it failed
     best=min(survived,key=lambda t:t["ce"],default=None)
+    if not best and traces:
+        best=min(traces,key=lambda t:t.get("g1_ce",999.0))
+
     obs_vars=set()
     for axl in AXL_PROBLEMS:
         if axl.name==record["problem"]: obs_vars=set(axl.observations.keys()); break
 
+    # FIX: Ensure it correctly marks it as UNSOLVED when CE is high or unverified
     ce_color="#4CAF50" if record["solved"] else ("#FF9800" if record["ce"]<1.0 else "#EF5350")
     status="✓ SOLVED + VERIFIED" if record["solved"] else ("~ PARTIAL" if record["ce"]<1.0 else "✗ UNSOLVED")
 
@@ -1440,12 +1453,14 @@ def _render_collapse(record:Dict) -> str:
         for t in traces
     )
 
+    dom_ce_str = f" | DomCE={record.get('domain_ce', 0.0):.6f}" if 'domain_ce' in record else ""
+
     return f"""
 <div style='border:1px solid #2a2a2a;border-radius:8px;margin-bottom:20px;overflow:hidden'>
   <div style='background:#111;padding:10px 14px;border-bottom:1px solid #222'>
     <span style='color:{ce_color};font-weight:700;font-size:1.1em'>{status}</span>
     <span style='color:#555;margin-left:16px;font-size:0.85em'>{record['problem']}</span>
-    <span style='color:#333;margin-left:12px;font-size:0.8em'>CE={record['ce']:.6f}</span>
+    <span style='color:#333;margin-left:12px;font-size:0.8em'>BaseCE={record['ce']:.6f}{dom_ce_str}</span>
     <span style='color:#555;margin-left:12px;font-size:0.8em'>{record['elapsed']}s</span>
   </div>
   <div style='display:flex;gap:0'>
@@ -1525,7 +1540,7 @@ async def dashboard():
 <div style='margin-bottom:18px'>
   <span class='badge' style='color:#aaa'>Runs: {runs}</span>
   <span class='badge' style='color:#4CAF50'>Solved: {solved}</span>
-  <span class='badge' style='color:#26C6DA'>Avg CE: {avg_ce}</span>
+  <span class='badge' style='color:#26C6DA'>Avg BaseCE: {avg_ce}</span>
   <span class='badge' style='color:#FF9800'>Problems: {len(AXL_PROBLEMS)}</span>
   <span class='badge' style='color:#5C6BC0'>Library: {len(DEBATE_ENGINE.library)}</span>
 </div>