Remove custom STXXL configuration from the code. From now on STXXL

should be configured externally as suggested in STXXL documentation.

Author: mdimura

examples/fastcluster_load.py

@@ -7,6 +7,8 @@
 import mdtraj as md
 import numpy as np
 import json
+import os.path
+
 
 Nk=1
 if len(sys.argv)>1:
@@ -22,25 +24,35 @@
 finishedLoad=time.time()
 print ("finished loading ({}k): {}".format(Nk,finishedLoad-start))
 
-Nframes=traj.n_frames
-D=np.empty((Nframes*(Nframes-1)/2,))
-j=0
-for i in range(0,Nframes-1):
-  rmsds=md.rmsd(traj, traj, i)[i+1:]
-  D[j:j+Nframes-i-1]=rmsds
-  j+=Nframes-1-i
+Zfile='z_fastcluster_load_{}k.json'.format(Nk)
+Z=[]
+if os.path.isfile(Zfile):
+  obj_text = open(Zfile, 'r').read()
+  Zl = json.loads(obj_text)
+  Z = np.array(Zl)
+  print('loaded the linkage matrix')
+else:
+  Nframes=traj.n_frames
+  D=np.empty((Nframes*(Nframes-1)/2,))
+  j=0
+  for i in range(0,Nframes-1):
+    rmsds=md.rmsd(traj, traj, i)[i+1:]
+    D[j:j+Nframes-i-1]=rmsds
+    j+=Nframes-1-i
+    
+  #D = np.empty((traj.n_frames, traj.n_frames))
+  #for i in range(traj.n_frames):
+      #D[i] = md.rmsd(traj, traj, i)
+  #D=squareform(D,checks=False) #rounding errors make D appear asymetric
 
-#D = np.empty((traj.n_frames, traj.n_frames))
-#for i in range(traj.n_frames):
-    #D[i] = md.rmsd(traj, traj, i)
-#D=squareform(D,checks=False) #rounding errors make D appear asymetric
-
-finishedRMSD=time.time()
-print ("finished rmsd: {}".format(finishedRMSD-finishedLoad))
-#Z-matrix contains the linkage history, see more at http://docs.scipy.org/doc/scipy/reference/generated/scipy.cluster.hierarchy.linkage.html#scipy.cluster.hierarchy.linkage
-Z=linkage(D, method='complete',preserve_input=False)
-finishedClust=time.time()
+  finishedRMSD=time.time()
+  print ("finished rmsd: {}".format(finishedRMSD-finishedLoad))
+  #Z-matrix contains the linkage history, see more at http://docs.scipy.org/doc/scipy/reference/generated/scipy.cluster.hierarchy.linkage.html#scipy.cluster.hierarchy.linkage
+  Z=linkage(D, method='complete',preserve_input=False)
+  #One would probably like to save the Z-matrix for the future, so that there is no need to redo the clustering
+  open(Zfile, 'w').write(json.dumps(Z.tolist()))
+  finishedClust=time.time()
+  print ("finished clustering: {}".format(finishedClust-finishedRMSD))
 
-#One would probably like to save the Z-matrix for the future, so that there is no need to redo the clustering
-open('z_fastcluster_load_{}k.json'.format(Nk), 'w').write(json.dumps(Z.tolist()))
-print ("finished clustering: {}".format(finishedClust-finishedRMSD))
+labels=cluster.hierarchy.fcluster(Z, 0.20, criterion='distance')
+open('frame_labels_{}k.dat'.format(Nk), 'w').write(str(labels))

src/inmatrix.cpp

@@ -7,15 +7,9 @@
 
 #include "inmatrix.h"
 
-InMatrix::InMatrix(unsigned numpoints, const string cachefile,
-		   const string cacheOptions, unsigned long ramUse):sorter(Cmp(),ramUse) {
+InMatrix::InMatrix(unsigned numpoints, unsigned long ramUse):sorter(Cmp(),ramUse) {
 
 	//set Matrix::threshold, numPoints
-	stxxl::config * cfg = stxxl::config::get_instance();
-	unsigned long diskSize=numpoints?12L*numpoints*(numpoints-1)/2:2L*1024L*1024L*1024L;
-	stxxl::disk_config disk1(cachefile, diskSize, cacheOptions);
-	//disk1.direct = stxxl::disk_config::DIRECT_OFF;
-	cfg->add_disk(disk1);
 }
 
 bool InMatrix::getNext(uint &row, uint &col, float &value) {

src/inmatrix.h

@@ -35,8 +35,7 @@ struct InMatrix: public Matrix {
 	uint pos;
 
 
-	InMatrix(unsigned numpoints=0, const std::string cachefile="sorter.tmp",
-		 const std::string cacheOptions="syscall nodirect unlink",
+	InMatrix(unsigned numpoints=0,
 		 unsigned long ramUse=1L*1024*1024*1024);
 
 	~InMatrix() {

src/main.cpp

@@ -47,7 +47,7 @@ int main(int argc, char **argv) {
 	if (!extractOptions("--linkage=%s", method, argc, argv))
 		strcat(method,"complete");
 
-	InMatrix inMat(0,"sorter.tmp","syscall nodirect unlink",64*1024*1024);
+	InMatrix inMat(0,64*1024*1024);
 
 	using std::chrono::duration_cast;
 	using std::chrono::milliseconds;