[Therion] Bug with uncertainty on fixed station on windows 10?

[Xavier Pennec]

Hi Bruce,

With the help of the latest snapshot solving the crash issue, I found 
the source of these large distortions. The loop closure that I am using 
is The are due to the limited numerical accuracy (float rather than 
double somewhere?) in the weighted least-squares for error compensation 
when we put very large standard deviations (in my case "sd length 400 
m)". I had indeed shots for far away for triangulation, and since the 
length is estimated a posteriori from the map, I wanted to put a very 
high confidence interval on it. Getting down to a stddev of 50 m solves 
the problem. Thus this problem is different from the one we experienced 
in 2013.

Xavier



Le 10/09/2020 à 07:33, Bruce Mutton a écrit :
> Xavier
> I did have a project that distorted part of a loop 400 m vertically too far downwards.  Changing from Therion loop closure to Survex loop closure resolved the problem, and so I have retained that setting ever since.  That may have been in 2013, or maybe earlier.
> On a few occasions since I have checked to see if that problem still manifested, but it did not.  I never figured out why it did it for only a window in time/Therion versions.
> Bruce
>
> -----Original Message-----
> From: Therion <therion-bounces at speleo.sk> On Behalf Of Xavier Pennec
> Sent: Thursday, 10 September 2020 09:05
> To: therion at speleo.sk
> Subject: Re: [Therion] Bug with uncertainty on fixed station on windows 10?
>
> Well, on my side, my projects with multiple fixed points with uncertainty that were compiling last years are now generally crashing at compilation. However, since I changed computer, therion and survex version, it is difficult to single out the reason.
>
> With no loop and several fixed points with uncertainty, it seems to crash systematically like in the example I posted. When it does not crash with multiple loops and multiple fixed points with uncertainty, it can either gives a .3d output that seems reasonable in aven, or it can displace one of the fixed points kilometres away, either in horizontal or vertical direction! Just fixing a point at 950 altitude can output a 3d file where the altitude is at -4000m....  A new way to explore abysses! I have just looked at the .3d result so far since I have not drawn all the maps yet in the cases where it compiles.
>
> As soon as I manage to set up a mock-up example, I will post it to document this amazing behaviour.
>
> Best
>
> Xavier
>
> Le 09/09/2020 à 22:34, Bruce Mutton a écrit :
>> I forgot to point out that most of my projects probably have multiple fixed stations with uncertainty parameters set.  I don't think I have experienced any problems, at least in recent times and with recent versions of Therion.
>> So your example is interesting.
>> Bruce
>
> _______________________________________________
> Therion mailing list
> Therion at speleo.sk
> https://mailman.speleo.sk/listinfo/therion

-- 
> -------------------------
> Xavier Pennec
> Senior Research Scientist / Directeur de recherche
> Asclepios project-team, INRIA Sophia-Antipolis
> 2004 Route des Lucioles, BP93
> F-06902 Sophia-Antipolis Cedex, France
> +33 4 92 38 76 64
> +33 6 78 35 16 90
> https://team.inria.fr/epione/en/
> -------------------------------