On 16/10/2024 14:36, Allin Cottrell wrote:
On Wed, 16 Oct 2024, Riccardo (Jack) Lucchetti wrote:
> On 11/10/2024 19:31, Cottrell, Allin wrote:
>
>
> More importantly: we found a problem in a third-party library that (on
> Windows only) made the invpd() function for inverting positive
> definite matrices quite slow for medium-sized matrices. The problem
> itself is not fixed, since it doesn't depend on us, but we managed to
> find a workaround that appears to be effective.
That's true, but the invpd() fix is in all of the current Windows
snapshots, not just the "new" UCRT/GTK3 one.
Ah, you're right of course.
I just wanted to tell people about the invpd() thing so they can watch
out and tell us before release if anything catastrophic happens
(unliklely, but still).
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------