Closed matsen closed 6 years ago
I'm becoming less enthused about BEAST's clock assumptions, and suspect that we may get more stable results by using unrooted trees.
The ever helpful and friendly Andy Magee has supplied us with an ASR example for RevBayes.
This example is for a fixed topology. Apparently, to make it change topologies, add these two lines:
moves[++mi] = mvNNI(topology, weight=10.0)
moves[++mi] = mvSPR(topology, weight=5.0)
before this line: my_model = model(Q)
my_model = model(Q)
And remove this line: topology.clamp(true_tree)
topology.clamp(true_tree)
See #21
I'm becoming less enthused about BEAST's clock assumptions, and suspect that we may get more stable results by using unrooted trees.
The ever helpful and friendly Andy Magee has supplied us with an ASR example for RevBayes.
This example is for a fixed topology. Apparently, to make it change topologies, add these two lines:
moves[++mi] = mvNNI(topology, weight=10.0)
moves[++mi] = mvSPR(topology, weight=5.0)
before this line:
my_model = model(Q)
And remove this line:
topology.clamp(true_tree)