Opened 4 years ago

Closed 22 months ago

#4475 closed enhancement (fixed)

Scaling of singular Jacobian

Reported by: Christian Andersson Owned by: Christian Andersson
Priority: major Milestone: Pending
Component: FMU/JMU runtime Version: trunk
Keywords: Cc:

Description

Currently when a Jacobian is singular and a row in the matrix is identically zero, the scaling of that row will be set to the maximum. This should be limited so that once an row is detected to be identically zero, do not scale that that equation.

Change History (5)

comment:1 Changed 4 years ago by Christian Andersson

Milestone: 1.16.x1.17.x

comment:2 Changed 4 years ago by Christian Andersson

Milestone: 1.17Pending

comment:3 Changed 4 years ago by aramle

When a row is 0 in the Jacobian, use the heuristic residual scaling.

comment:4 Changed 4 years ago by aramle

r8512:
#3879 Added a check for a valid minimum norm solution in case of a singular Jacobian.
#4475 In case we have an entire row in the Jacobian equal to 0, use heuristic residual scaling instead of the maximum.

comment:5 Changed 22 months ago by Jesper Mattsson

Resolution: fixed
Status: newclosed

Done.

Note: See TracTickets for help on using tickets.