def trace(self):
debug("Trace called")
data = self.data_source.get_counts()
if sum(data) == 0.0:
return
mn = self.param_list[0].min_val
mx = self.param_list[0].max_val
stride = (mx - mn)/self.num_points_to_plot
score_lists = []
inverted_maxima = []
overall_max = []
for i, parameter in enumerate(self.param_list):
tree = self.trees[i]
to_optimize = TreeLikelihoodFunc(tree, data, parameter)
prev = parameter.value
curr = mn
sc = []
max_pt = (mn, float('-inf'))
while curr < mx:
val = to_optimize(curr)
pt = (curr, val)
if val > max_pt[1]:
max_pt = pt
sc.append(pt)
curr += stride
# call the function to reset the variable to its original value
to_optimize(prev)
inverted_maxima.append((max_pt[1], max_pt[0], sc.index(max_pt), i))
score_lists.append(sc)
inverted_maxima.sort(reverse=True)
maxima = [(i[1], i[0]) for i in inverted_maxima]
self.do_plot(score_lists, maxima, x_range=(mn, mx) )
def opt_model_param(self, param, curr_step=0.04):
debug('opt_model')
return self.call_opt(param, curr_step=curr_step)
def opt_D(self,curr_step=0.04):
debug('opt_D')
return self.call_opt(self.free_parameters[BranchEnum.D], curr_step=curr_step)
def opt_Internal(self,curr_step=0.04):
debug('opt_Internal')
return self.call_opt(self.free_parameters[BranchEnum.INTERNAL], curr_step=curr_step)
def prob_vec_updated(self, prob_vec_index):
debug("prob_vec_updated(%d)" % prob_vec_index)
self.repaint()
