Implementing tension statistics

README.rst

@@ -2,7 +2,7 @@
 anesthetic: nested sampling post-processing
 ===========================================
 :Authors: Will Handley and Lukas Hergt
-:Version: 2.3.0
+:Version: 2.4.0
 :Homepage: https://github.com/handley-lab/anesthetic
 :Documentation: http://anesthetic.readthedocs.io/
 

anesthetic/_version.py

@@ -1 +1 @@
-__version__ = '2.3.0'
+__version__ = '2.4.0'

tension_calculator.py

@@ -0,0 +1,48 @@
+class tension_calculator(dict):
+     """Class for computing tension metrics
+
+     Parameters
+     ----------
+     samples : dict
+         Dictionary of NestedSamples
+     nsamples : int
+         Number of samples to use for computing statistics
+     """
+     def __init__(self, samples, nsamples=1000):
+         for k, v in tqdm.tqdm(samples.items()):
+             self[k] = v.stats(nsamples)
+
+     def union(self, A, B):
+         A_ = set(A.split('_'))
+         B_ = set(B.split('_'))
+         return  '_'.join(sorted(list(A_.union(B_))))
+
+     def intersection(self, A, B):
+         A_ = set(A.split('_'))
+         B_ = set(B.split('_'))
+         return  '_'.join(sorted(list(A_.intersection(B_))))
+
+     def stat(self, A, B, stat):
+         union = self.union(A, B)
+         intersection = self.intersection(A, B)
+         return self[union][stat] - self[A][stat] - self[B][stat] + self[intersection][stat]
+
+     def logR(self, A, B):
+         return self.stat(A, B, 'logZ')
+
+     def logS(self, A, B):
+         return self.stat(A, B, 'logL_P')
+
+     def d(self, A, B):
+         return self.stat(A, B, 'd_G')
+
+     def D_KL(self, A, B):
+         return self.stat(A, B, 'D_KL')
+
+     def p(self, A, B, d=None):
+         logS = self.logS(A, B)
+         if d is None:
+             d = self.d(A, B)
+         else:
+             d = np.ones_like(logS)*d
+         return scipy.stats.chi2.sf(d[d>0]-2*logS[d>0], d[d>0])