Skip to content

user specification of sparse matrices in models #11

@paciorek

Description

@paciorek

For discussion perhaps sooner than later since we are thinking about models...

How will a user specify a matrix is sparse? Some options that occur to me:

  1. via the type of the matrix in inits, data, constants, but some of these are not required when a model is built
  2. via sparse(Q[1:5,1:5]) ~ dwish(....)
  3. via Q[1:5,1:5,'sparse'] ~ dwish(...)
  4. via some new argument to nimbleModel (akin to how a user can specify dimensions via an argument instead of implicit in model code), perhaps in conjunction with (1).

I presume we could do a combination, such as (1) and (2) such that we use the types in whatever inits, etc. are provided, and otherwise default to dense unless sparse() is used.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions