patch_condn: Patch condition number extraction

In xpose.xtras: Extra Functionality for the 'xpose' Package

Patch condition number extraction

Description

Bugfix for xpose:::sum_condn, the internal function xpose uses to populate the 'condn' (condition number) entry of xpdb$summary.

For NONMEM runs with more than one estimation method (e.g. SAEM followed by importance sampling), the .lst file contains more than one ⁠EIGENVALUES OF COR MATRIX OF ESTIMATE⁠ block. xpose always uses the first block found, which is not necessarily from the final estimation method, so the reported condition number can be wrong. This patch instead uses the last block, matching the value reported by NONMEM-adjacent tools such as PsN's sumo.

xpose:::sum_condn() itself (not this function) is also where a multi- method run with more than one ⁠EIGENVALUES OF COR⁠ block raises "numerical expression has ... elements: only the first used" – it runs automatically inside xpose::xpose_data(), before patch_condn() gets a chance to run, so that warning is expected and cannot be suppressed from here; this function only fixes the resulting 'condn' value afterward.

Usage

patch_condn(xpdb)

Arguments

xpdb	An xpose_data or xp_xtras object.

Value

The xpdb object, with a corrected 'condn' entry in xpdb$summary (unchanged if xpdb is not from nonmem, or if no eigenvalues could be found).

Examples

xpdb_ex_pk <- patch_condn(xpose::xpdb_ex_pk)

xpose.xtras documentation built on Sept. 1, 2026, 5:08 p.m.