New version of the model

1 files changed, 59 insertions, 0 deletions

diff --git a/experiments/ml.pyx b/experiments/ml.pyx
new file mode 100644
index 0000000..c6291c5
--- /dev/null
+++ b/experiments/ml.pyx
@@ -0,0 +1,59 @@
+# cython: boundscheck=False, cdivision=True
+import numpy as np
+cimport numpy as np
+from libc.math cimport log
+
+DTYPE = np.float64
+ctypedef np.float_t DTYPE_t
+
+cdef DTYPE_t weight_victim(int dist, int dt, DTYPE_t alpha,
+                           DTYPE_t delta, DTYPE_t gamma):
+    cdef DTYPE_t structural, temporal
+    structural = delta ** dist
+    temporal = (gamma - 1. / alpha) * 1. / (1. + dt / alpha) ** gamma
+    return structural * temporal
+
+
+cdef DTYPE_t weight_non_victim(int dist, int t, DTYPE_t alpha,
+                               DTYPE_t delta, DTYPE_t gamma):
+    cdef DTYPE_t structural, temporal
+    structural = delta ** dist
+    temporal = 1. - 1. / (1. + (3012. - t) / alpha) ** gamma
+    return 1. - structural * temporal
+
+
+def ml(dict root_victims, dict victims, dict non_victims,
+       DTYPE_t alpha, DTYPE_t delta, DTYPE_t gamma=1.01):
+    cdef:
+        int n_roots, n_victims, n_nodes, roots, i, dist, dt, t
+        DTYPE_t beta
+        list parents, parents_weights
+    n_roots, n_victims = len(root_victims), len(victims)
+    n_nodes = n_victims + len(non_victims)
+    cdef:
+        np.ndarray[DTYPE_t] probs = np.zeros(n_victims, dtype=DTYPE)
+        np.ndarray[DTYPE_t] probs_nv = np.zeros(len(non_victims), dtype=DTYPE)
+    for i, parents in enumerate(victims.itervalues()):
+        parents_weights = [weight_victim(dist, dt, alpha, delta, gamma)
+                           for (dist, dt) in parents]
+        probs[i] = max(parents_weights)
+    for i, parents in enumerate(non_victims.itervalues()):
+        parents_weights = [weight_non_victim(dist, t, alpha, delta, gamma)
+                           for (dist, t) in parents]
+        probs_nv[i] = max(parents_weights)
+    probs.sort()
+    probs = probs[::-1]
+    cdef:
+        np.ndarray[DTYPE_t] betas = probs / (1. + probs)
+        np.ndarray[DTYPE_t] cums = np.log(probs.cumsum())
+    for i in xrange(n_victims - 1, 0, -1):
+        roots = n_roots + n_victims - 1 - i
+        if betas[i] > roots / float(n_nodes):
+            break
+    else:
+        print "alpha: {0}, delta: {1}. Everyone is a root".format(alpha, delta)
+        roots = n_roots + n_victims
+    beta = roots / float(n_nodes)
+    return (beta, roots,
+            roots * log(beta) + (n_nodes - roots) * log(1 - beta) + cums[i]
+            + np.log(probs_nv).sum())