Check for existing issues
Description
Currently, GWKokab integrates two samplers:
• FlowMC
• NumPyro-based NUTS sampler
However, both have limitations:
• FlowMC struggles with high-dimensional models.
• NUTS performs well but is relatively slow for our use cases.
💡 Proposal
We propose exploring the implementation of a new sampler inspired by the method described in this paper.
Integrating this sampler into GWKokab could potentially improve efficiency and scalability for complex hierarchical population models.
The sampler was presented at this DCC entry.
Check for existing issues
Description
Currently, GWKokab integrates two samplers:
• FlowMC
• NumPyro-based NUTS sampler
However, both have limitations:
• FlowMC struggles with high-dimensional models.
• NUTS performs well but is relatively slow for our use cases.
💡 Proposal
We propose exploring the implementation of a new sampler inspired by the method described in this paper.
Integrating this sampler into GWKokab could potentially improve efficiency and scalability for complex hierarchical population models.
The sampler was presented at this DCC entry.