Skip to content

Potential bug in periodic TDMA #301

@pbartholomew08

Description

@pbartholomew08

In the TDMA solvers the solution du is declared as intent(out). In the periodic solver the initial boundary condition is implemented as (using CUDA kernel as an example) where jm1=n when j=1

  jm1 = modulo(j - 2, n) + 1
  du(i, j, b) = temp_du - du(i, jm1, b)*thom_s(j)

As du is intent(out) the value of du(i,n,b) is undefined at this point.

N.B. we might be getting away with this if thom_s(1)=0

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