initial kdtree script profiled for optimization.

.gitignore

@@ -102,4 +102,6 @@ ENV/
 .mypy_cache/
 
 # manages cProfiler and gprof2dot outputs
-output.*
+output.
+
+!speed_milestones/*

README.md

@@ -17,20 +17,19 @@ the network is labeled:
 |...|...|...|...|...|...|
 |(r-1)*c|(r-1)*c+1|...|...|...|(r-1)*c+c-1=r*c-1|
 
-## Todo
-- consider a way to visualize a current/resistance grid
-- Impliment Updateable objects such as FETs/Memristors
-- have update matrix U that changes/replaces G to allow for change ie switching
-- have U made/updated by querying the edge objects that make up the network
-- investigate the networkx package as a way to handle the network to matrix conversions and visualization
-
-## flow
-- make network with connections
-- add components to edges
-- make adjacency matrix from conductance of components
-    - components must have a way to calculate conductance
-- make mna matrix with additional data about sources and ground_nodes
-- solve mna
-- move mna solutions back to physical network, so that nodes have voltages,
-and edges have currents
-- update components and resolve
+
+## Optimization
+use cProfile and the [gprof2dot](https://github.com/jrfonseca/gprof2dot).py script to generate profiles of time spent for running the script.
+explicit number to run optimization on is:
+
+    kd_percolation.py 300 --pm 0.135 --length 0 --scaling 5
+
+and corresponds to 300 sticks on a 5um square area with a length distribution of 0.66pm0.44um corresponding to my experimental results. conduction is calculated in each case.
+
+which is run in sequence as:
+    python -m cProfile -o output.pstats kd_percolation.py 300
+    python gprof2dot.py -f pstats output.pstats | dot -Tpng -o output.png
+
+The total run time over 5 or 6 tries is between 14 and 15 seconds (user time) when running `time python kd_percolation.py 300` for the original brute force cluster search.
+
+Using a kdtree with a sorted length list and length dependant search radius the same run times are 3.6 to 3.8 seconds.

kd_percolation.py

@@ -16,7 +16,8 @@ class StickCollection(object):
     def __init__(self,n,l,sticks=None,pm=0,scaling=1):
         if sticks:
             self.sticks, self.intersects = self.make_clusters(sticks)
-        self.sticks, self.intersects  = self.make_clusters(self.make_sticks(n,l=l,pm=pm,scaling=scaling))
+        self.sticks, self.intersects  = self.make_clusters_kdtree(self.make_sticks(n,l=l,pm=pm,scaling=scaling))
+        self.make_cnet()
 
     def check_intersect(self, s1,s2):
         #assert that x intervals overlap
@@ -106,7 +107,7 @@ def make_clusters_kdtree(self,sticks):
                     if intersection and 0<=intersection[0]<=1 and 0<=intersection[1]<=1:
                         sticks.loc[sticks.cluster==sticks.loc[j,'cluster'],'cluster'] = sticks.loc[i,'cluster']
                         intersects.append([i,j,*intersection, sticks.iloc[i].kind+sticks.iloc[j].kind],)
-        self.percolating=sticks.loc[0,"cluster"]==sticks.loc[len(sticks)-1,"cluster"]
+        self.percolating=sticks.loc[0,"cluster"]==sticks.loc[1,"cluster"]
         intersects=pd.DataFrame(intersects, columns=["stick1",'stick2','x','y','kind'])
         intersects['cluster']=intersects['stick1'].apply(lambda x: sticks.iloc[x].cluster)
         return sticks, intersects
@@ -197,7 +198,7 @@ def show_sticks(self,sticks,intersects,ax=False):
         stick_colors=[stick_cmap[i] for i in sticks.kind]
         collection=LineCollection(sticks.endarray.values,linewidth=0.5,colors=stick_colors)
         ax.add_collection(collection)
-        isect_cmap={'ms':'g','sm':'g', 'mm':'k','ss':'k', 'vs':'k','sv':'k','vm':'k', 'sg':'k','gs':'k','mg':'k'}
+        isect_cmap={'ms':'g','sm':'g', 'mm':'k','ss':'k', 'vs':'k','sv':'k','vm':'k', 'sg':'k','gs':'k','mg':'k','gm':'k'}
         isect_colors=[isect_cmap[i] for i in self.intersects.kind]
         ax.scatter(intersects.x, intersects.y, c=isect_colors, s=20, linewidth=1, marker="x")
         ax.set_xlim((-0.02,1.02))
@@ -216,12 +217,14 @@ def show_sticks(self,sticks,intersects,ax=False):
     parser.add_argument("--scaling",type=float,default=5)
     parser.add_argument("-t", "--test", action="store_true")
     parser.add_argument("-v", "--verbose", action="store_true")
+    parser.add_argument("--show", action="store_true",default=False)
     args = parser.parse_args()
     if args.test:
         collection=StickCollection(args.number,l=0,pm=args.pm,scaling=5)
 
 
     else:
         collection=StickCollection(args.number,l=args.length,pm=args.pm,scaling=args.scaling)
-        collection.show_system()
+        if args.show:
+            collection.show_system()
         # print(len(collection.sticks.cluster.drop_duplicates()))