Remove the topology-only clock

[hyanwong]

At the moment in the forwards_algorithm, the section labelled # Topology-only clock assigns into the vv variable, but then nothing is done with that variable. It looks like there's a mistake there to me.

[awohns]

This is pretty old code I added to check the accuracy of the likelihood calculations vs a naive topology-only metric that basically just counts the number of edges beneath a node. I can’t think of a time where we’d actually want to use this anymore.

…

On Dec 6, 2019, at 9:03 PM, Yan Wong ***@***.***> wrote: At the moment in the forwards, the section labelled # Topology-only clock assigns into the vv variable, but then nothing is done with that variable. It looks like there's a mistake there to me. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[hyanwong]

I guess you meant to do the same as the previous lines, and assign val *= vv and probably g_val *= vv too in that branch of the if statement?

[awohns]

Yes, another bit of code that used to be in there but got lost because we never used/tested the topology-only clock!

…

On Dec 6, 2019, at 9:59 PM, Yan Wong ***@***.***> wrote: I guess you meant to do the same as the previous lines, and assign val *= vv and probably g_val *= vv too in that branch of the if statement? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#73>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACRLT2BSWSTVL2Z7XYXJCSDQXLDKTANCNFSM4JXBIXLQ>.

[hyanwong]

Tests have show that basing node times on a naive expectation under the coalescent based on each marginal tree (what we call the "topology-only" clock) does not give good results (see e.g. #292), so I think we should simply remove the ability to do this, and raise a NotImplementedError