Re: TGeant3 geant321 version number

Hi Federico,
Thanks for your response.

> we have several improvements (at least we do hope so) that we
> never ported back into the official 321 of the cern library. There
> was a synchronisation some time (2 years) ago, but then we diverged
> again. The file gversc.inc may not be meaningful any more, thanks for
> pointing this out. We might resynchronise, however when you say that
> there are differences, how can you be sure that these are due to the
> transport and not to the geometry? This is indeed the most likely cause.

I think it's important to synchronize the two because otherwise it's hard to claim that we are using Geant3 for the physics simulation.   I am still exploring the differences between CERNLIB's Geant3 and Geant3 as implemented in the geant3_vmc distribution. What I've noticed so far is that some (most) of the differences are in the direction of improvements applied to geant3_vmc but not yet imported to CERNLIB's Geant3, as expected and as you suggest. But there also appear to be some cases in which a change was made to CERNLIB's Geant3 but not geant3_vmc. For example, in the CERNLIB distribution of Geant3.21/14, the gphys/gdray.F routine has this:
  CC PEELS=SQRT(ABS((EELS+AMASS)*TELS))         PEELS=SQRT( PELS(1)**2 + PELS(2)**2 + PELS(3)**2) whereas the geant3_vmc version of the same routine has this:

        PEELS=SQRT(ABS((EELS+AMASS)*TELS)) The latter corresponds to what appeared in CERNLIB Geant3.21/11 packaged with cernlib version 98, so it appears that an improvement that was added to CERNLIB Geant3.21 has not been implemented in the geant3_vmc version.   A formal resynchronization would take care of these differences. Thanks & best regards,
-Sue

> Hello,
> we have several improvements (at least we do hope so) that we
> never ported back into the official 321 of the cern library. There
> was a synchronisation some time (2 years) ago, but then we diverged
> again. The file gversc.inc may not be meaningful any more, thanks for
> pointing this out. We might resynchronise, however when you say that
> there are differences, how can you be sure that these are due to the
> transport and not to the geometry? This is indeed the most likely cause.
>
> Federico Carminati
> CERN-PH
> 1211 Geneva 23
> Switzerland
> Tel: +41 22 76 74959
> Fax: +41 22 76 79480
> Mobile: +41 76 487 4843
>
> On 28 Oct 2007, at 06:50, Susan Kasahara wrote:
>
>> Hi vmc team,
>> We've noticed very minor differences between the use of TGeant3 and
>> using
>> geant3.21/14 from cernlib outside of the VMC framework. This has
>> raised
>> a question which is "what is the sub-version of geant3.21 used as part
>> of TGeant3?" Or should we expect that the geant321 code used in
>> TGeant3 has
>> diverged from the original cernlib source, so that this question of
>> which
>> sub-version is not meaningful.
>> I've noticed that the sub-version of geant321 in TGeant3 as
>> determined from
>> the package's geant321/gversc.inc is geant3.21/11. It appears
>> however that
>> this sub-version number hasn't changed in some time.
>> More generally, can you please describe the policy regarding the
>> updates
>> to TGeant3's geant3.21?
>> For example, are all changes to the most recent cernlib version of
>> geant3.21
>> included in TGeant3's version, but TGeant3's version may also include
>> additional updates that go beyond cernlib's standard version?
>> Thanks very much for your time.
>> -Sue
>>
>
>
Received on Mon Nov 05 2007 - 03:18:02 CET