add ability to run GC phase or not, and pick beta, through CLI

app/Main.hs

@@ -46,12 +46,16 @@ setupEnv (Right n) = getUnlabGraph (Random n)
 
 main :: IO ()
 main = do
-  [fpin, fpout] <- getArgs >>= \xs ->
-    if length xs /= 2
-    then Prelude.error "Usage: gargantext-graph-exe input_graph.csv output_file.txt"
+  [fpin, fpout, gcstr, betastr] <- getArgs >>= \xs ->
+    if length xs /= 4
+    then Prelude.error "Usage: gargantext-graph-exe input_graph.csv output_file.txt <'gc' or 'nogc'> <beta>"
     else return xs
+  let gc = if gcstr == "gc" then True else False
+      beta = case readMaybe betastr of
+        Just d -> d
+        _      -> Prelude.error "beta must be a Double"
   setupEnv (Left fpin) >>= \(dico, ~g) -> do
-    let (clusts, score) = withG g (\fg -> clusteringOptim 3 Conf fg beta)
+    let (clusts, score) = withG g (\fg -> clusteringOptim 3 Conf fg beta gc)
         clusts' = Prelude.map (sort . Prelude.map (lkp dico) . IntSet.toList)
                 $ sortBy (\a b -> flipOrd $ comparing IntSet.size a b)
                 $ Prelude.map (\(n, ns) -> IntSet.insert n ns)
@@ -65,8 +69,7 @@ main = do
           "len=" ++ show (length clust) ++
           " [" ++ intercalate ", " [ "'" ++ Text.unpack w ++ "'" | [w] <- clust ] ++ "]\n"
 
-  where beta = 0.0
-        flipOrd LT = GT
+  where flipOrd LT = GT
         flipOrd GT = LT
         flipOrd EQ = EQ
         lkp dico i = fromMaybe (Prelude.error "Node not in dictionary?!") $

src/Graph/BAC/ProxemyOptim.hs

@@ -73,8 +73,9 @@ clusteringOptim :: forall n a b. KnownNat n
                 -> Similarity
                 -> FiniteGraph n a b
                 -> Double -- beta
+                -> Bool   -- True = run GC, False = don't
                 -> (Dict IntSet, Double)
-clusteringOptim l s fg@(FiniteGraph g) beta = runClustering beta adj' prox sorted_edges
+clusteringOptim l s fg@(FiniteGraph g) beta gc = runClustering gc beta adj' prox sorted_edges
     where
       !adj' = symAdjacent fg True
       !tra' = symTransition adj'
@@ -474,15 +475,24 @@ data Clust = Clust
 
 runClustering
   :: forall (n :: Nat). KnownNat n
-  => Double -- ^ beta
+  => Bool -- ^ do we run the 'garbage collector'?
+  -> Double -- ^ beta
   -> SMatrix.Matrix n n Double -- ^ adjacency
   -> SMatrix.Matrix n n Double -- ^ proxemie
   -> SortedEdges
   -> (Dict IntSet, Double)
-runClustering beta adj prox se = trace "clustering starts" $ runST $ do
+runClustering gc beta adj prox se = trace "clustering starts" $ runST $ do
   mclust <- newMClustering n
   forM_ se $ \(x, y, _) -> clusteringStep beta adj prox mclust (x, y)
-  trace "basic clusters done, running collector now" $ clusteringCollector beta adj prox mclust
+  if gc
+    then trace "basic clusters done, running collector now" $ clusteringCollector beta adj prox mclust
+    else do cps <- V.unsafeFreeze (mparts mclust)
+            let cps' = Dict.fromList
+                  [ (n, xs)
+                  | (n, Just (PartData xs _)) <- zip [0..] (V.toList cps)
+                  ]
+            sc <- MVU.unsafeRead (mscore mclust) 0
+            return (cps', sc)
 
   where n = fromIntegral $ natVal (Proxy :: Proxy n)
         sestr = intercalate "\n"