8:33 a.m.
Dear all,
I would like to report a possible bug in the functions invcdf() and critical().
gretl (2023b, Linux) slows down and crashes (after a while) when I run the commands
below.
<hansl>
scalar aa = invcdf(X,1,1)
</hansl>
or
<hansl>
scalar bb = critical(X,1,0)
</hansl>
Best,
Francesco
8:44 a.m.
Am 04.09.2023 um 10:33 schrieb f.valentini(a)pm.univpm.it:
Dear all,
I would like to report a possible bug in the functions invcdf() and critical().
gretl (2023b, Linux) slows down and crashes (after a while) when I run the commands
below.
<hansl>
scalar aa = invcdf(X,1,1)
</hansl>
Confirmed on Windows with 2023b (couldn't test with the latest snapshot
right now) - actually I got a freeze and didn't wait for an eventual crash.
This is potentially a bit of a nasty bug, I guess, since it's a standard
statistical calculation. I wonder when this crept in.
thanks
sven
8:53 a.m.
New subject: R: Re: Problem with "invcdf()" and "critical()"
Hi everybody,
also confirmed on Mac Os (intel) with the latest snap.
Best regards,
Luca
Da: Sven Schreiber <sven.schreiber(a)fu-berlin.de>
Data: lunedì, 4 settembre 2023 10:44
A: gretl-users(a)gretlml.univpm.it <gretl-users(a)gretlml.univpm.it>
Oggetto: [Gretl-users] Re: Problem with "invcdf()" and "critical()"
Am 04.09.2023 um 10:33 schrieb f.valentini(a)pm.univpm.it:
Dear all,
I would like to report a possible bug in the functions invcdf() and critical().
gretl (2023b, Linux) slows down and crashes (after a while) when I run the commands
below.
<hansl>
scalar aa = invcdf(X,1,1)
</hansl>
Confirmed on Windows with 2023b (couldn't test with the latest snapshot
right now) - actually I got a freeze and didn't wait for an eventual crash.
This is potentially a bit of a nasty bug, I guess, since it's a standard
statistical calculation. I wonder when this crept in.
thanks
sven
_______________________________________________
Gretl-users mailing list -- gretl-users(a)gretlml.univpm.it
To unsubscribe send an email to gretl-users-leave(a)gretlml.univpm.it
Website: https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/
3:07 p.m.
On Mon, Sep 4, 2023 at 4:44 AM Sven Schreiber
<sven.schreiber(a)fu-berlin.de> wrote:
Am 04.09.2023 um 10:33 schrieb f.valentini(a)pm.univpm.it:
> Dear all,
>
> I would like to report a possible bug in the functions invcdf() and critical().
>
> gretl (2023b, Linux) slows down and crashes (after a while) when I run the commands
below.
> <hansl>
> scalar aa = invcdf(X,1,1)
> </hansl>
Confirmed on Windows with 2023b (couldn't test with the latest snapshot
right now) - actually I got a freeze and didn't wait for an eventual crash.
This is potentially a bit of a nasty bug, I guess, since it's a standard
statistical calculation. I wonder when this crept in.
It has always been there. It seems that the underlying cephes code
goes into a fatal loop in cases such as this where the correct answer
is + or - infinity. A fix should be available before long.
Allin
3:27 p.m.
Am 04.09.2023 um 17:07 schrieb Cottrell, Allin:
On Mon, Sep 4, 2023 at 4:44 AM Sven Schreiber
<sven.schreiber(a)fu-berlin.de> wrote:
> This is potentially a bit of a nasty bug, I guess, since it's a standard
> statistical calculation. I wonder when this crept in.
It has always been there. It seems that the underlying cephes code
goes into a fatal loop in cases such as this where the correct answer
is + or - infinity. A fix should be available before long.
So an upstream bug report is needed, or is it just the way that gretl
uses it?
thanks
sven
5:29 p.m.
On Mon, Sep 4, 2023 at 11:27 AM Sven Schreiber
<sven.schreiber(a)fu-berlin.de> wrote:
Am 04.09.2023 um 17:07 schrieb Cottrell, Allin:
> On Mon, Sep 4, 2023 at 4:44 AM Sven Schreiber
> <sven.schreiber(a)fu-berlin.de> wrote:
>> This is potentially a bit of a nasty bug, I guess, since it's a standard
>> statistical calculation. I wonder when this crept in.
>
> It has always been there. It seems that the underlying cephes code
> goes into a fatal loop in cases such as this where the correct answer
> is + or - infinity. A fix should be available before long.
So an upstream bug report is needed, or is it just the way that gretl
uses it?
First, I can confirm that the cephes code entered an infinite loop in
the function igami(). However, I can't rule out the possibility that
this was due to the way libgretl uses cephes, so I'm not sure that an
upstream bug report is warranted.
Anyway, this is now fixed in git.
Allin
473
days inactive
474
days old
6 comments
4 participants
participants (4)
-
Cottrell, Allin -
f.valentini＠pm.univpm.it -
Luca Pedini -
Sven Schreiber