Fix last code chunk in Overview vignette #19
Assignees
Comments
|
Would it be realistic to expect that the updated versions of transition() and geocorrection() will be able to handle, say, 500M cells? The last version released by Kaue, before you graciously agreed to take over, produces wrong accCost() results when either the # of columns or the # of rows of the impedance raster exceed 10,000. At 500M cells, the limit would increase to 22300+ cells, sufficiently large for a set of pressing problems that have been put on hold. |
|
This isn't really the correct issue to be discussing that, but I do plan to make sure that whatever issue is happening in #17 is resolved. |
andrewmarx
added a commit
that referenced
this issue
Jun 4, 2023
Addresses #19 Remove disconnected patches in the raster. The clump number is specific to this raster and may not be the same for other rasters. I'm not clear yet if this is actually necessary or is just a workaround for a different issue. It's possible that disconnected (unreachable) patches in the transition matrix is not possible with the matrix decomposition, but I haven't determined yet why it used to work. It's possible a bug was fixed in the Matrix package that was allowing the matrix to be incorrectly decomposed, or it's possible a new bug was introduced to the Matrix package that breaks the decomposition in certain circumstances.
|
Closing as "fixed" for now. See commit 305f1b9 for details |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
pathInc()in last code chunk of Overview vignette produces a matrix decomposition error. Need to identify source of problem and undo the workaround in commit 70c9656The text was updated successfully, but these errors were encountered: