Closed ArtPoon closed 7 years ago
The current method to be used as an alternative to delta functions for setting specific parameters is to set the priors in the config file as a uniform distribution with hyperparameters 'max' and 'min' both set to the desired value.
priors:
lambda0:
dist: 'unif'
hyperparameters:
- min: 0.1
- max: 0.1
lambda1:
dist: 'unif'
hyperparameters:
- min: 0.2
- max: 0.2
mu0:
dist: 'unif'
hyperparameters:
- min: 0.03
- max: 0.03
mu1:
dist: 'unif'
hyperparameters:
- min: 0.03
- max: 0.03
q01:
dist: 'unif'
hyperparameters:
- min: 0.01
- max: 0.01
q10:
dist: 'unif'
hyperparameters:
- min: 0.01
- max: 0.01
@MathiasRenaud noted a use case where a user may want to simulate trees under specific parameter settings of a model. We discussed adding an "initial parameter values" setting to the Kaphi configuration object, but instead I think the appropriate method would be to specify Dirac delta functions as the prior distributions for the respective parameters.
Actions: