|
|
Status Updates
Note: This page is not currently being maintained :-(.
Status and ToDo (7/11/99)
It's been ages since we sent out an update on things!
- Full Simulation
- it was found that the MC truth table truncation at 10
generations had a noticeable (detrimental) impact on analyses.
This bug has been fixed. The truncation limit has been removed.
- a bug was found in the code affecting the Large hadron
calorimeter. There was a limit of 100 layers in the code. The
Large detector has 120 layers. The effect of this was that the
layer numbers wrapped between the barrel and endcap (ie barrel
layers 100-119 ended up in one endcap and so on). Fixed. We are
looking to see if there might be some other oddities in the tags
(conflicting evidence).
- some benign bookkeeping errors have been fixed in the Gismo PDG
particle properties lists
- tracker hit ID's did not properly reflect N/S. Fixed. The system
bit overwrite bug has been fixed.
- we still plan to put in a geometry version tag into the output
ASCII file
- we have tracked the Linux NaN bug to bad values passed in from
the Gismo propagate code to the calorimeter digitization. We're
tracing it up the tree. Remember that this NaN bug is quite rare
on a per-event basis.
- there is a Gismo bug for loopers. They are being cut off
prematurely. We're investigating the fix.
- we have code now to implement a cryostat around the vertex
detector and it turns out that the mask geometry already
implemented will suffice for our needs.
- we were planning on a new Gismo release for the August meeting,
though an earlier release could be considered.
- the May'99 production release of Geant4 has still not arrived.
They are still fighting some bugs. So we still have no experience
with it.
- Background overlays
- a Guinea-Pig ascii file-to-StdHep interface has been written. We
will need to sort out how the many thousands of particles get into
Gismo (also needs the Gismo looper bug fix!).
- Parameters handling
- much work has been done on creating a framework to allow
parameter access (eg geometry constants, energy scales, etc) from
all phases of code use - from full simulation through to analysis.
XML is the core of this framework and we are now at the stage of
trying it out on the Gismo code. Conversion and testing is
expected to take a while yet.
- RootApps
- the Root jet and thrust finders have been debugged and tested
against the standard Fortran versions. They agree and are
available. They will be contributed back to Root-central soon.
- there seems to be a problem with our Root code on Linux RedHat
6.0. The problem is absent on 5.2!
- the FastMC has been updated to use the new helix parameters and
error matrix formulation. We still need to update the ASCII file
writing to write/read the helix information. The FastMC can still
be used interactively (as is done for the JAS version).
- note that we are not currently planning on implementing a full
reconstruction stage in Root. Event analysis and FastMC will be
maintained.
- hep.lcd package (and JAS analysis)
- Several new releases have been made during the spring and
summer, containing significant improvements to the Clustering and
Track Reconstruction. Details can be found in the release
notes for each version.
- A persistent bug affects some people trying to run event
displays and analysis on remote servers (the symptom is a
java.lang.NoClassDefFoundError: hep/lcd/io/ascii/handler/EventException).
We are trying to track down the origin of this bug, which seems
obscure.
- We have obtained the Kalman fitter code from the FNAL MCFast
package and are working on providing an interface for it.
- work is underway to characterize calorimeter clusters in terms
of min-I portion and shower. Initially we are looking for the
shower start at the end of the min-I portion and then fitting a
linear 'shower axis' to the shower hits.. We expect this to be an
iterative process with the clustering.
- The first release of JAS 2.0 will soon be available. We will
support both JAS 1.0 and 2.0 for LCD work for a few weeks, until
all the bugs are worked out of the new release, after which we
will recommend that everyone switches over to the new release.
- We will add support for performing background overlays before
the August meeting.
- Full Simulation:
- bug found in Track hit packed identifier
- it was found that the VXD system bit was overwritten
- MC Truth table problem
- Mike Ronan is assembling a JAS class to interpret the MC
truth information for event tagging
- Linux status
- Dave Gerdes has noticed a bug in the ascii file for
optimized Linux. He occasionally sees a number field writtne
(literally) as 'NaN' (unix-ese for Not a Number). This is has
been reproduced here at SLAC, but so far I am at a loss as to
how to debug it! It seems to happen in 1-2% of events.
- Root Applications
- following discussions with Rene Brun, our Event classes have
been revised for better speed. V2.0 of all the Root stuff has been
released and is on the web. This also includes bug fixes as well
as the new Event structure.
- we see an overall factor of 2 improvement in speed. It also
allows independent access to the various subsystems. If one
just wants tracking, it goes 20x faster! Most of the data is
in the calorimetry, so those folks are stuck with the factor
of 2. The Root folks have just released a new version of the
I/O which they think will give us another factor of 2.
- ports of the JAS thrust and jet finders to Root are coded and
undergoing test now. We're planning to have them ready for the
Root Users' Meeting on 3/23.
- Full Recon is almost ready to go in dev. The Cluster Cheater is
complete and the track smear code now uses hits to determine
acceptance. There is a problem with access to Root files that
needs to be solved before this goes out the door.
- Java Framework (hep.lcd package)
- We have initiated regular weekly meetings to discuss progress on
the Java reconstruction/fast-simulation package. Users not at SLAC
are welcome to join the meeting by phone conference (contact tony_johnson@slac.stanford.edu
for detatils). Meetings are at 1:30pm (PST) on Tuesdays.
- Progress is proceeding on a number of fronts:
- Nic Sinev is generalizing Mike Ronan track finder to work
with Small and Large detector and to take its parameters and
geometry info directly from the detector.ini file
- Joanne Bogart is working on reviving her Java Cluster
Finder.
- Gary Bower is working on a framework for track-cluster
association
- Mike Ronan has developed some tools for categorizing events
based on their MC truth info.
- The fast MC now uses Bruce Schumm's parameterizations for
track smearing.
- A page summarizing the status
of the Java Framework is here.
- Generator
- Pandora has been successfully interfaced to Pythia. Tim reports
that for the existing processes that Pandora handles, it is more
efficient than PYBMS and functionally better.
- MC Farms
- production of tt(bar) at Michigan is long completed
- SLAC continues on various backgrounds
- Colorado is producing selectrons
- the full simulation code is being installed on a Linux PC
cluster at FNAL. We're still iterating on getting all the
necessary packages installed and found by us. The Gismo code is
all there now.
- the farm system is also being installed at Vanderbilt
- we are about to start testing Don Benton's auto-ftp script to
push farm files to Penn's repository. Once the files are there,
they will be concatenated at converted to JAS and Root object
files. Then other farms can install the Perl scripts to push their
data over. We need to publicize the directory structure!
- Getting Ready for FNAL Simulations meeting
- a fair amount of effort at SLAC went into getting ready for the
upcoming meeting! It resulted in uncovering a few more bugs.
- Full Simulation:
- Small EC Disk #3 outer radius wrong
- Keith Riles noticed this today. The radius will probably end
up being 44.2 cm; previously 38.3.
- MC Truth table problem
- we upped the cut from 5 to 10 generations in Gismo.cxx (see
last week's report above).
- Linux status
- I've run off a fair number of events now and saw only rare
cases where Gismo got into trouble. It is useable.
- FastMC
- problem with Gismo ascii->Root conversion
- only the last smeared cluster was being written to the Root
file - nCluster times. Fixed.
- Homer Neal and David Wagner ran into some troubles setting up a
Makefile on DEC unix for the RootApps stuff. David has now
succeeded and will spread the word.
- Fallout from Root consulting (cont'd)
- the new Event classes & asc2root are done. Some tweaking is
still going on, but the code is in dev now.
- Generator
- a diagnostic generator that outputs StdHep is ready to go.
Should be in the next RootApps distribution.
- MC Farms
- Dave Gerdes continues production of tt(bar) at Michigan
- SLAC continues on udscb background
- Don Benton at Penn has established a repository (50 GB for
starters) for all the MC farms. He has written scripts to allow
pushing the data from the farm machines to Penn. We haven't had a
chance to install it at SLAC yet.
- TODO
- analysis tools (CLHEP, thrust/jet finders) for Root (GRB, RS)
- Pasha Murat at FNAL is producing vector classes etc to
'replace' CLHEP. It is due out in a week or two.
- FYI, there will be a meeting
of the US Root Users group at FNAL from 3/23-25.
- finish Event shuffle (RS)
- design facility for dealing with parameters (Joanne)
- allow all stages of processing from simulation to analysis
to access needed parameters in a uniform way. Not easy!
- Vacation time!
- I (Richard) will be away until March 1, being incommunicado from
2/17-26 inclusive.
- Full Simulation: bugs, bugs, bugs!
- Bad EGS/Gheisha materials file found
- the Cu.mat file was found yesterday to have the density to
2.702 (aluminum's!). This invalidates any full simulation
studies of the Small detector design
- there is no Cu defined in the Large, so it ducked this
problem.
- a correct Cu.mat has been made available in the web zips
area and has been copied to the SLAC central repositories on
unix and NT.
- MC Truth table problem
- it turns out that Gismo cuts off the truth table at 5
generations (for display purposes). This also truncates the MC
truth table! It seems that the PYMBS is more verbose than
Isajet and the truth table has been truncated too much.
- we will up the cut. There must be one, the way Gismo is
done.
- Linux bug
- this bug went away when our SLAC Linux box was set up with
egcs 1.1. The bug was present in egcs 1.03. Gismo has gone
through a set of 100 ZZ 500 GeV events and a couple thousand
single electrons and pions. Usually these are stringent tests.
- all is not quite rosy, in that I've seen for the 20 GeV
pions, cases where Gismo gets stuck in a WARNING loop
complaining about slow electrons (below the EGS cutoff). I've
been having troubles reproducing it. It seems rare (enough).
- FastMC
- shareables have been made in dev. This will allow running the
FastMC interactively in Root.
- Fallout from Root consulting
- we met with Rene Brun, top Root guru last week.
- the main outcome was advice on our Root event structures. The
bottom line is that we will dump all our TMaps and convert to
TObjArrays. Also, converter pointers to McParts to indices into
its TObjArray. This will allow us to assign each subsystem's data
to separate Root I/O branches for selective reading. Removing the
TMaps should vastly speed up the I/O (our main perceived problem).
- these mods are underway in dev.
- this will have a ripple effect on asc2root, the fastMC and
fullRecon, of course!
- Generator
- a diagnostic generator should be about ready to go, pending the
Event shuffles above
- MC Farms
- Dave Gerdes continues production of tt(bar) at Michigan
- SLAC fired up udscb background last weekend
- Colorado is has solved their startup problems with StdHep and
are poised to begin.
- Vanderbilt is installing the latest code version and will try it
out on one node for starters
- Code development
- Tony Waite has written up his code development package. As
things stand we expect to use it as an interim tool while
switching to SNiFF+. He is still working out trying to get
compatible Makefiles between NT and unix.
- TODO
- Short term
- publish timing test results and recommend numbers of events
to generate
- analysis tools (CLHEP, thrust/jet finders) for Root (GRB,
RS)
- finish Event shuffle (RS)
- really squish Linux Gismo bug (RXD)
- Full Simulation
- Detector geometry files have been redefined a few times
and are on the web.
- There was some kafuffle about the LUM/mask layout. In the
end, a vestigial mask was put in and the LUM remains active
over most of the remaining acceptance.
- a typo was corrected in the VXD radii for layer 2 in the
Large detector. The web version has been updated to include
this fix.
- Nothing has been done yet to debug the crash on Linux.
- fastMC
- it has been released to the SLAC public unix area
- it is now capable of working on NT. The FNAL folks (Philippe
Canal, specifically) were able to debug the problem, which turned
out to be not setting up the link properly.
- this bring to the surface the hitch that we need the Fortran
stdhep libraries to get the stdchg utility. So we have to make
the Fortran libraries on NT. We'll have to get Fortran
PowerStation for it.
- a bug was found in the helical extrapolation and fixed. A new
distribution was made up as well as a patch tar file.
- shared libraries are in the process of gettiing made so the
fastMC can be run interactively in Root.
- Root consulting
- we will be meeting tomorrow with Rene Brun, top Root guru, to
get advice on our event structure design, plus anything else we
can think to ask him.
- Generator
- we ran into a problem on DEC unix with PAW/COMIS for Tim's code;
he came up with a work-around.
- Events from each of the physics
processes of interest were run to ring out the system and do
timing tests to then determine how many events to ask the sites to
produce. Full results are still to be posted, but essentially all
the 'hadronic' final states had about the same timing and the
two-photon backgrounds were a factor of 6 faster.
- a diagnostic generator is just about completed that will allow
firing single particles into the fastMC (or Gismo)
- MC Farms
- Dave Gerdes has started production of tt(bar) at Michigan
- SLAC is about to fire up udscb background here at SLAC. We are
discussing cuts to apply.
- Vanderbilt is installing the latest code version and will try it
out on one node for starters
- Colorado is having some startup problems getting the binaries
working.
- TODO
- Short term
- publish timing test results and recommend numbers of events
to generate
- analysis tools (CLHEP, thrust/jet finders) for Root (GRB,
RS)
- finish cluster cheater for full recon (RS)
- squish Linux Gismo bug (RXD)
- Full Simulation
- Detector geometry files have been defined and are on the web
- Root parser has been updated for the new formats
- Nothing has been done yet to debug the problem on Linux.
- fastMC
- it has been released to the SLAC public unix area (it does not
work on NT; we have reported it to FNAL for a suspected conflict
between Root and StdHep), $NLDROOT/RootApps/.
- the code is ready for export from our web
- Generator
- there is now a batch farm interface to Tim's pybms generator
machinery.
- we ran into a problem with long filenames for Tim's code and
have found a fix to that.
- the plan is to run off 100 events from each of the physics
processes of interest to do timing tests to then determine how
many events to ask the sites to produce.
- Full recon
- a first pass at the full recon is done, some gussying up and
checking remains. It could be available next week. It will contain
the fastMC track smear (but uses hit multiplicity for
acceptance) and the cluster-cheater calorimeter clustering
algorithm
- Documentation
- prior to code release, we are trying to generate a fair bit of documentation
on the various aspects of the packages.
- TODO
- Short term
- wait for things to hit the fan after release of fastMC
- analysis tools (CLHEP, thrust/jet finders) for Root (GRB,
RS)
- finish cluster cheater for full recon (RS)
- squish Linux Gismo bug (RXD)
- Full Simulation
- As per the December simulations, the requests to allow spec of
individual layers of the tracker and VXD to allow
- tapered length and irregular radial spacing
- endplates on the TPC
- disk 'endcap' trackers
has been completed in the Gismo code. The code version is in
place on unix now and, pending a few more tests, will be
released imminently.
- Nothing has been done yet to debug the problem that has surfaced
with the full simulation code on Linux. We now have a Linux box
here at SLAC to work with.
- the public NT version of the code has migrated to the SLAC
'group' V: disk under LCD\. The current version is called lcd-working,
just as on unix.
- fastMC
- there has been some mission creep here as the working groups
wanted more/different functionality than offered in December
- Calorimeter
- version 1 ready to release
- includes helical/linear extrapolation of particles to
calorimeter and follows them in to determine interaction
point
- Tracking
- the power law description of the resolutions was dumped
in favour of lookup tables per parameter. This required a
complete rewrite of the tracking smear. The form and
number of the tables is still under discussion, so
completion of this part will probably take the rest of the
week.
- Generator
- there is now a batch farm interface to Tim's pybms generator
machinery.
- it appears there will be some mix of isajet and pybms for the
various farm operations for the first round as we are still not
set up with a generator that will do it all: SUSY, beammstrahlung
and polarization.
- Analysis tools
- the JAS environment has new analysis tools: a CLHEP clone plus
thrust and jet finders. Documentation for these is upcoming. They
will also be ported to the root environment.
- TODO
- Short term
- finish cluster cheater for full recon (RS)
- add clusters to JAS, root parsers (RS, ASJ)
- package asc2root (root ascii file parser), fastMC &
fullRecon for export (JRB)
- package asc2root, fastMC & fullRecon for NT (RXD)
- package pybms and isajet interfaces for export (GRB)
Status & ToDo (12/14/98)
- The workshop was attended by about 35 people. Minutes summarizing
the meeting will be following. The last afternoon was devoted to
tweaking the Strawman detector designs. The new versions will also be
out shortly. There were some consequences on the full simulation
code, with the intention of having a revised code version ready for
startup in early January.
- allow spec of individual layers of the tracker and VXD to allow
tapered length and irregular radial spacing
- allow endplates on the TPC
- allow disk 'endcap' trackers
- It was also deemed a priority that we get the fastMC out the door
and to debug the problem that has surfaced with the full simulation
code on Linux.
- We have prepared short- and longer-term TODO lists. Short-term
addresses getting full MC going early in January and making the fastMC
available.
- TODO
- Short term
- allow individual spec for VXD & tracking layers and
endplate for TPC (RS - done)
- allow for endcap disks for tracking (RS)
- debug Linux problem (RXD)
- find out if weird Small detector pion behaviour is a bug or
feature (RXD, see workshop talk)
- finish cluster cheater (RS)
- finish calSmear (JRB)
- add clusters to JAS, root parsers (RS, ASJ)
- redo parameter access in fastMC (JRB)
- package asc2root (root ascii file parser), fastMC &
fullRecon for export (JRB)
- complete batch wrapper for Tim Barklow's generators (GRB)
- run test cases for GISMO (GRB)
- Longer term ToDo
- root event display (RS)
- GEANT4 (APW?)
- cal clustering (JRB)
- background overlay (GRB)
- analysis tools for root, Java (RS, GRB)
- CLHEP for root (RXD, RS)
- fix Gismo shower parentage problem (RXD)
- fix endcap Gismo draw bug (GRB)
- backfill NT project info into unix distribution (APW?)
- port Event classes, fastMC, fullRecon to NT and allow export
(RXD)
- CVS et seq (APW)
- better web organization and code doc (all)
- create some standard test suite for Gismo (GRB)
- ponder detector parameters following datasets around (ASJ)
- new NBEAM more like SLD MCDIAG (ie allow angular, momentum
ranges for particles) that outputs to StdHep (RS)
- 'write up' tutorials (ASJ, RXD)
- Gismo code porting status
- there appears to be a compiler bug on DEC unix in egcs 1.1 in
which Gismo's Messenger.cxx will not compile. Recall there was a
compiler bug (we think) in 1.03 on DEC unix in which Menu.cxx
wouldn't compile with debug symbols.
- David Wagner has submitted the bug to egcs and in the meantime
has picked up the 1.03 binaries
from David Gerdes. The code is running fine at Colorado now.
- Don Belton found that the Gismo code runs on Linux with
optimization turned off, but fails just as it did for Dave Gerdes
with in on (fails while reading the Si materials file in EGS).
This needs to be addressed.
- SLAC Central NT group disk space
- we have acquired a quota of 500 MB on the (groups) disk V:\LCD
on the central SLAC NT fileservers.
Status & ToDo (12/06/98)
SLAC Simulation workshop tomorrow!
- Full Simulation
- code has been released and is on the web as are .ini files for
the two Strawman detector designs
- generated 1000 Small and 500 Large SM events
- generated 1-5k events for20 GeV electrons and pions for the
Snowmass, Large and Small detectors.
- the root & Java parsers are operational. PAW ntupler is
being debugged.
- the machinery for the full reconstruction has been cloned from
the fastMC
- the cluster-cheater code has been written and is ready for
debugging.
- found a bug in the calorimeter shower bookkeeping while
preparing for workshop
- I encountered an McPart with all properties set to zero. It
appears as a contributor to a tower ID, but never get filled
with real properties.
- analysis kludge is to avoid McParts with status code set to
zero.
- FastMC
- the standard parsers can now read the fastMC output as well as
fullSim.
- the calorimeter smear is coded and about to be debugged. Ray
Frey contributed a helical extrapolator.
- generated 10k Standard Model events with the Small detector
parameters.
- Generators
- For farming, we still need the wrapper that passes in I/O specs,
random number init etc
- Miscellany
- Dave Gerdes installed the new fullSim code at Michigan. It
workson his DEC unix machine. Still fails on Linux.
Status & ToDo (11/27/98)
- Full Simulation
- we have continued to make last minute mods to the full
simulation code. The last one broke the code before our hoped-for
Thanksgiving-eve release. It is apparently now fixed. We have run
a few hundred Standard Model events through the simulation with no
problems.
- the Java parser has been updated and the JAS event display shows
the events to look reasonable to the eye.
- the root parser has been updated and is being debugged.
- the cluster-cheater has taken a back seat to the code release.
We'll get back to it when the code goes out the door. We can use
the FastMC machinery (see below) to jumpstart the reconstruction
package.
- we now allow optional inner/outer skins and endplates on all the
cylindrical detector types (most needed for the central tracker as
TPC). That's the code that broke at the last minute!
- FastMC
- the machinery is now set up and the track smear is done. We can
now read StdHep files, do the smear and write out ascii files.
Parameters can be input from an ascii file.
- the calorimeter smear is underway. Ray Frey is contributing a
helical extrapolator.
- the formats for the MC and track smear are identical to the full
simulation versions, so the parsers should be able to eat the new
files directly. They will need to be modified for the (new)
calorimeter cluster format.
- Generators
- Tim Barklow has interfaced his PYBMS machinery to StdHep and we
have successfully read files he output.
- For farming, we still need the wrapper that passes in I/O specs,
random number init etc
- Miscellany
- Dave Gerdes did a 'quick and dirty' port to his new Linux box of
the Gismo code and met with a code crash. This occurred in malloc
while trying to read in a materials file in EGS. Ugh. This goes on
the back burner for now.
- ToDo
- Full MC
- get root and PAW parsers going
- set up 'Large' detector ascii file
- finish cluster cheater and assemble reconstruction (with
track smear)
- provide mechanism for calculating calorimeter and muon strip
channel coordinates from the binary tag words.
- FastMC
2 1/2 weeks until the December SLAC workshop. See preliminary agenda.
- Richard's trip
report from the Frascati ECFA-DESY workshop is available.
- the full simulation code has been assembled on both NT and SLAC unix.
We'll spend a short period with last minute clean-up, then brief
testing and then tar it up for distribution.
- it now contains a muon system, luminosity monitor and masks
- muon strips do now contain the MC history
- the track smearing code has been removed, to go into a
reconstruction step
- calorimeter towers are now described by binary tags and
segmentation is defined in terms of polar, not dip, angle
- the geometry description has been modified to allow inner and
outer skins and endplates (envisaged for the TPC design mostly)
- the full reconstruction cluster-cheater code is mostly done.
Probably another wall clock week before it is ready.
- we are about to embark on creating the full-smear simulation package
- tracking will be taken from Keith Riles' algorithm and
calorimetry initially from the Snowmass smear.f routines.
- still no luck on the StdHep interface to Pythia. If it continues to
resist, we may need to go with the unadorned Isajet generator for
initial studies.
Status & ToDo (11/05/98)
- the (old) code is up and running completely at Penn (AIX) , from
generator to farm batch scripts. At least that's what they tell me!
- it also now running at Michigan (DEC Alpha). Not fully exercised
yet, but looks good so far on muons and pions. A couple of code mods
need to propagate back from there.
- I believe all the technical mods needed to the gismo code to run the
strawman designs have been demonstrated, if not cleanly finished. Rob
will see to the tidying up. I've asked Gary to do some QC. More mods
will be required for the next versions to accomodate requests for
detector design options.
- David Wagner has agreed to write code to do the fast smear for
tracking and calorimetry. I don't know the state of that work. The
tracking module would be used in both the fast smear and 'full'
reconstruction.
- Rob is about to embark on the fake full clustering (rebuild clusters
from the MC hit history).
- Gary is trying to get the Pythia StdHep wrapper to work. It is
resisting him. We need it for Tim's generator interface package.
- Joanne is about to attempt C++-izing her PASS1 clusterer.
- TonyW is remoulding the code to fit into the CVS code repository and
build scheme
- we'll need to modify the output ascii file parsers for new
subsystems and mods to the calorimetry tower IDs.
- the muon hit output does not contain the MC history. I'll try to
code that up on my laptop while I'm away.
- The most current version of the code is on SLDNT17 in S:\gismo. I
think Rob is assembling a unix version.
Status & ToDo (10/31/98)
- Zeroth-order Strawman designs have been created to get the
simulations going. An electronic description will be available
shortly.
- a date has been set for the Simulation Workshop: 7-8 Dec here at
SLAC.
- it is expected that the Strawman designs will be modified by the
working groups after this meeting.
- we are targetting work first that is needed for the Strawman
simulations, then we'll go on to reconstruction and fast-smear
options.
- Mods to GISMO for Strawman
- Masks, Luminosity monitor and muon system installed.
- calorimeter tower ID's changed from decimal packing to binary;
towers now based on polar angle, not dip.
- the iniData geometry parser needs updating to allow some new
fields and to handle the polar angle change
- we will need to update the parsers to handle the new packing
scheme and new subsystems
- we are now including the input detector description file in the
output ascii file. This will allow the parsers to figure out the
geometry and segmentations directly.
- Code export offsite
- the Makefile cleanup was a big success, with the code being
successfully installed at Penn (IBM/SP/AIX) and being brought up
now at U Mich (DEC Alpha). There have been some egcs wrinkles at
UMich that will need looking into (presumably due to the Alphas
being 64 bit machines).
- the generator interfaces and submit scripts have been added to
our web.
- Fast Smear and Reconstruction
- David Wagner (Colorado) has agreed to modernize the Snowmass
smear.f to include a new tracking smear from Keith Riles. This
package will have the same inputs and outputs as the full
simulation package, and is planned to component interchangeable
with it.
- initial looks at D0's TRF++ package suggest that it will be
non-trivial to set up. We are moving it to the back burner for now
(sharing that spot with GEANT4).
- we will embark on a "poor man's clustering" while we
wait for the full package to happen. This will involve using the
full MC hit history to rebuild the clusters. We can then degrade
the clusters as we wish to allow for overlaps, splitoffs, etc.
This should start next week.
Status & ToDo (10/14/98)
- Our work was well received at Keystone. Some 20 groups have signed
on to do some analysis and/or simulation work. The outcome
of Keystone was:
- define two initial detector designs ("S" &
"L") and four reference reactions
- create a completely fast-smear simulation path to complement the
GISMO full simulation
- run off the 8 full simulation datasets by the end of November
and have a workshop at SLAC on Dec 7-8 for people working on the
simulations.
- Penn and U Michigan have volunteered to be guinea pigs in bringing
up the code away from SLAC. Joanne is going through a round of cleanup
of the Makefiles to grease the wheels. They should be ready any time
now.
- we are now in the process of planning the upgrades to the full
simulation and design of the fast-smear option.
- Full simulation:
- add muon system (tail catcher calorimeter + muon tracking)
- allow for endplates and inner/outer skins on the detector
components
- allow for tapered cylinders for the VXD and Central Tracker
- remove the smeared Tracks (TrackSystem & ReconTrack
classes) from the GISMO stage, to go to a RECON stage
- change the calorimeter tower ID from using tanl to q, with
projective towers, and from decimal packing to binary
- allow termination of particles upon entry to the calorimeter
for the fast smear option
- Recon (of GISMO output and via fast smear)
- define steering routine that allows input from StdHep (ie
directly from the generator for full smearing) or from GISMO's
Root object file (for mixed smear/recon or full recon)
- transfer fast track smear from GISMO
- adapt calorimeter fast smear from Snowmass smear.f.
- Ongoing work
- Joanne continues to develop her clustering algorithm
- we are patching up bugs from the bug list
- develop background overlay and setup of the generators
- parser for PAW ntuples (waiting for redefinition of
calorimeter tower ID)
- it is a stretch to believe we'll get all this done in time!!
Status & ToDo (09/24/98)
Keystone - 2 days!
- Root object file is ready; SUSY and SM datasets have been rooted.
- Gary has completed the Java thrust finder
- Joanne has her first results from the cluster finding
- another bug was found in the tracking code: electrons which brem in
the tracker can give rise to multiple 'tracks'. GISMO terminates them
and creates new daughters of the photon and the electron itself. If
the 'daughter' collects enough hits, it can cause a track to happen
too.
- talks have been prepared for Keystone and we're just about on our
way.
Status & ToDo (09/12/98)
- Root object file
- the parser is essentially ready. After interactions with the
root author, it was found that our root code has chosen AIX to be
the only platform on which to crash. It works fine on Solaris and
we have successfully converted a large event file.
- Rene Brun suggested a few improvements to our class layout, so
we'll do that, clean up the code a bit and then translate all the
dat files. Then produce the root automatic html documentation for
both the classes and a sample macro to read the object files.
- have installed the latest version of root (2.00/11) on SLAC unix,
both for AIX and Solaris. Due to lack of space in the nld group
area, I have put it in sld group space: /afs/slac/g/sld/nld/root200v11/@sys/root.
(the '@sys' is set at SLAC to account for architecture).
- analysis tools
- Java
- Gary has designed and implemented 3- and 4-vector classes
and is moving on to creating a thrust class, to be followed by
jet finding.
- Tony's new student, Azhar, is working on improvements to the
event display.
- Root
- Rob will probably write a wrapper for Fortran versions of
LUTHRU and the JADE jet finder. We can adapt Gary's Java
version when it is done.
- Calorimeter Clustering
- Joanne is underway in coding an algorithm. It will involve some
binning of towers, meaning that a second pass clusterer will be
needed later. It is not expected that her code will be ready for
significant use before Keystone.
Status & ToDo (09/06/98)
- ~25 fb-1 of SUSY Point 3 and Standard Model backgrounds have been
simulated.
- work is under way on:
- first clustering algorithm in the calorimeter
- parser to produce Root object file
- analysis tools like thrust axis and jet finders.
- CHEP98 news
- the beta release of GEANT4
is out. The first production release is due out at the end of '98.
- Root appears to be gaining ground quickly in HEP. FNAL will
likely support it as the PAW replacement. CDF is using it for its
object I/O.
- D0 has produced a track finding/fitting package, called TRF++,
that is advertised as portable. We definitely should take a look
at it. The author felt it could also be applied to a muon system.
- Solaris update
- a new batch system binary was installed and a test run of one
W-pair job that failed consistently ran to successful completion.
We'll need to run some of the other failed jobs to see if the
problem has been fixed.
Status & ToDo (08/26/98)
[one month until Keystone]
- we have completed 25k SUSY-point3 and about 8k standard model
background events so far
- Solaris version
- the above events have been mostly generated on the AIX farm.
- the pion non-interaction bug was squished. We are currently
having a problem wherein about 1/3 of the jobs crash - but due to
the batch process dying, not the nld executable. This has been
reported to the batch system vendor.
- we have done some preliminary analysis of the point 3 data. We
easily see selectron pairs and their distinctive box momentum
distribution. The calorimetry simulation was used for some crude
electron ID and looks OK.
- In progress:
- putting a wrapper on PYTHIA, for its 2gamma process.
- Joanne has studied the SLD calorimeter clustering code and is
close to starting to code a first algorithm for nld.
- trying to decide on how to implement thrust axis and jet
finding: wrap Lund/YCLUS Fortran utilities or try to come up with
C++/Java versions.
- work is underway to produce a Root object file from the ascii
file.
- David Wagner at Colorado will build a PAW ntuple from the ascii
file.
- we got 10 GB of NFS space from SCS for this work and must
transfer existing output to it. Also to work out a scheme for
backing those files off to tape via the staging system.
It's Friday
- we have launched the first test run of 5k SUSY Snowmass 'Point 3'
events! If no glaring problems are found, we'll just continue
submitting. These are being run on the SLAC AIX batch farm in the nldq
queue. A scholarly discussion is underway with Colorado to settle on
the value of electron beam polarization. If that resolves itself to a
different value than the test jobs use, we'll start over today.
- as noted above we are doing Isajet Point 3. We have, as per The
Plan, put in -90% electron beam polarization (for now, see above).
- the current idea is to run Point 3 as signal and W-pairs (also
with Isajet) as background. We'll determine whether we need other
backgrounds as well (eg 2-gamma).
- we still are not doing beamsstrahlung
- we installed isajet 7.37 in our nld space. We are now up to date
with it.
- Joanne has completed the Solaris build of nld. We still have the
same pion non-interaction problem.
- GISMO energy cutoff raised
- we raised Gismo::ke_cutoff from 0.5 to 1 MeV (matching GEANT's
default). This gained us about 30-50% in execution time, with
(presumably) no real loss in shower simulation quality.
- we need to put in a higher cutoff for neutral hadrons (notably
neutrons) of 10 MeV, again matching GEANT's. These suckers rattle
around for a long time needlessly!
- Root I/O
- Toby Burnett and (mostly) his student Andy Haas have inserted
Root I/O into GLAST's GISMO. We will be able to piggyback on this
development. They finessed the iostream conflict between TObject
and the standard library typedefs of the iostreams.
- TonyJ yesterday gave us a demo of Java Analysis Studio tuned
to NLD. We will use it to do checks on the simulation output and to
start development of the reconstruction package.
- we need to clean up after the rush to launch
- isajet generator wrapper is in ~grb
- we need a better output file destination that the farm work
disk!
- Rob is embarking on output file parser #2 to get us to a Root
object file.
Is it Friday yet?
- Joanne ran into a hitch with compiling the areve/arve library. egcs
does not recognize the namespace syntax used there in Warning.h. This
was a minor roadblock. Of the arve libraries, CLHEP still is to be
done - it is quite different from older versions; nld remains on our
side.
- Gary is still working on the wrapper to the generators
- Tracking mods
- Keith Riles delivered a new track smearing algorithm, which has
been incorporated into ReconTrack. For the moment, the smearing
parameters are hard-wired in to static variables.
- he recommended 600 um Si and 1,1 mm G10.
- this algorithm uncovered a glitch in the track hits: albedo
particles deep in a shower can leak back into the central detector
and leave a track. However its MCParticle object can be deleted in
cleanup after the shower initiator is done. The hits were modified
such that if it is albedo, the MParticle pointer in the hit list
is set to zero ('background', until we can think of something
better to do).
- Reference Geometry error
- a big overlap was found for the HAD barrel and endcap.
Calorimeter behaviour changed somewhat after repairing the error,
in that a large number of very low energy hits appeared in the hit
list. These were infinitesimal in terms of energy sum and an
energy threshold has been instituted of approximately 1% of min-I.
- a quick test of 500 5 GeV pions pretty much reproduces previous
results.
- SLAC unix batch farm
- this was negotiated with SCS and BABAR (Charlie Young). We have
been granted 75% of the AIX farm over the next 6 weeks, BABAR
needs one week, soon - perhaps this week. We assumed we would have
teething problems getting going and this would not be a real
conflict.
- a special queue is being created for us to facilitate the
operation.
- Randy Melen will try to find us 10 GB of NFS space, but may not
succeed in which case we would need to use the staging system.
- We are continuing to run test jobs on the PC's to try to squish as
many bugs as possible before unix launch.
Status & ToDo (08/10/98)
- unix g++ compiler
- Joanne has installed the egcs 1.03 compiler on AIX; Solaris to
follow tomorrow.
- we will need to see if the odd pion behaviour under Solaris
persists.
- we are proceeding with porting the latest release to unix.
- Gary has linked and run isajet and pythia with output to stdHEP
files.
- some work on the wrapper for passing in a random number seed,
output file, etc is still needed.
- Output ASCII file
- the format is complete (for the moment!).
- Revised 'tracks'
- the previous scheme was replaced with a count on deposited hits
before creating a track.
- Keith Riles (co-ordinator of the LC Tracking working group) has
promised us materials and a smearing scheme.
- Particle position at front face of the EM Calorimeter
- a thin (currently 100 um) air layer has been placed as the
front layer. A special digitizer stores the position if the
particle is outgoing. The position and drc's are tacked on to the
MCParticle output from MCPrint.
- Event Generation
- we have cloned (sic) the SLD MC farm scheme of defining task
files to describe an MC run. Perl scripts have been written to
submit bunches of jobs with the parameters set in the task file.
Farm code resides in $NLDROOT/farm et seq.
- after much putzing, we have generated a test sample of more than
50 isajet events on SLDNT0 (NT).
Status & ToDo (08/05/98)
- unix g++ compiler
- the original plan for the SLAC g++ maintainers to handle this
fell through. Joanne is going to install the egcs 1.03 compiler on
both AIX and Solaris.
- we will need to see if the odd pion behaviour under Solaris
persists.
- Gary has started on a wrapper for isajet so it can be called with a
random seed and output an stdHEP file. He ran into problems trying to
link to the SLAC isajet libraries.
- Output ASCII file
- we have generated a couple of passes through it, finding bugs as
we go.
- Tony has written a parser to convert the ASCII file into Java
objects that JAS can use
- we had hoped we could compress the ascii file on the fly via
unix pipes, but no luck so far. We should be able to zip it after
the fact. isajet events are running about 300-400 kB; ~100 kB
after zip.
- we added a run number and random seed (placeholder) as optional
arguments to main. EventMarker will put out a run number in
addition to the event number.
- Revised 'tracks'
- the present scheme is a kludge: a 1mm cube is defined around the
IP; when a particle passes through it, a track is created based on
the MCParticle parameters. I would like to revise that to count
hits per MC particle in the tracking system. However, I have been
unable to get the STL sort stuff to work and am hoping Toby will
ride to the rescue!
- Same ToDo as last week for generating events.
- beampipe, VXD and tracking systems have been added, both for volumes
and a digitization scheme. The default geometry consists of cylinders,
while the digitizations are (layer#, x,y,z, Eloss, MC ptr) for each
particle intersection with a layer.
- the KAL digitization upgrade is complete; the MC history (assigning
shower particles' energies to the particle initiating the shower) is
done.
- Rob has merged his stdHEP/HepEvt code into the S: disk newarve
version; the new geometry and digitizations mods need to go in too.
- TonyJ is going write a Java Analysis Studio interface (DIM) to the
ASCII file. These files are big!
- Paul Kunz has agreed to install the current egcs g++ compiler on
SLAC Solaris; when the new version comes out, Matt Langston (the g++
maintainer) will install it on AIX and Solaris. This is needed for the
latest version of arve from GLAST.
- Rob is starting to look at why the Solaris version of the code
doesn't do pions correctly.
- Gary will make sure that we have the latest Pythia and Isajet
generators available on SLAC AIX and Solaris. He will also provide a
wrapper with which we can set the initial random seed and output an
stdHEP file.
- the Colorado folks are still seeing significantly less visible
energy in their GEANT3 simulation than we see with GISMO. They see e/mu
~ 0.7 and pi/mu ~ 0.4. For the Snowmass calorimeter, we see e/mu~0.85
and pi/mu~0.75. This is still quite a disturbing difference.
- GISMO uses the straight Bethe-Block ionization loss formula (as
given in the PDG book, for example), while GEANT3 has an elaborate
scheme for determining dE/dx and comments in the manual on the
importance of the steps they take. The concern is mostly for very
low energy particles; a specific worry is whether the range of
nuclear breakup fragments like deuterons and tritons are handled
correctly in GISMO. See Material::dedx for how GISMO computes
energy loss.
- this is the current worry about the validity of GISMO (even
though we ran it against the SLD calorimeter and got decent
agreement). Who knows?
- the parametrized tracks currently being written out need revision.
The kludge by which the trackable MC particles are selected is a tad
too crude!
- Generating events
- we are carrying on with the schedule to start generating events
soon, perhaps starting 8/10 or so.
- we need
- to iron out some procedural details like writing to tape on
unix, setting up scripts for job handling etc
- to decide on what generator/parameters we will run with, for
how many events
- get the revised code back to unix and working on Solaris
- what else?
Status & ToDo (07/23/98)
- Root I/O
- we have decided to abandon embedding root into GISMO for now.
Too many problems with compiler versions and stdlib conflicts to
deal with. We'll stay with an expanded ASCII format for now.
- this also means we can migrate to the egsc C++ compiler on unix
and not worry about whether it is root compatible or not (for
now).
- stdHEP/hepevt ready
- gismo/hepevt is now able to accept and propagate events
generated by isajet and stored in the stdHEP /HEPEVT/ format.
- final checks are underway
- KAL Digitization update complete
- a new digitization structure based on the proposed format
has been implemented and is being written out to an ASCII file.
- still need to get mods in to set SHOWER status for MCPartcles so
that proper parentage can be assigned and the particle tree
appropriately trimmed.
- also need to write an ASCII-to-root ntuple translator, based on
the prototype.
- unix makefile problems
- Joanne discovered that the unix makefiles were badly set up and,
in fact, didn't do library updates properly (wrong options to the
archive command). This has been fixed and some other cleanup has
been performed (explicit reference to libraries rather than
searches; use of gmake variables improved - eg $(@F) to get
the filename rather then $@ which is the full target filespec:
this gave odd behaviour depending on whether the target .o file
existed or not).
- code management
- Tony Waite has been working on a Rexx interface which will do
release control on top of CVS.
- Chad wrote up the Visual Studio configuration.
It will need to updated a little when stdHEP goes public. It also
suggests some minor organizational mods to the projects to improve
clarity.
- Solaris pions still broken
- Tom Pavel remade the f2c library on Solaris, but (as he
predicted) no improvement was seen. We'll have to debug the code
to see what is happening.
- There is a hitch
with the latest GLAST arve
- there is a 30% difference in pion response (lower than old arve).
Electron and muon responses are the same. Under investigation.
- this has been resolved: new arve inits GHEISHA internal to
gismo. It was also being done in main(). When this was
removed, new arve agreed with old.
- Root I/O in GISMO
- stdHEP generator
interface
- Rob is still debugging his modified HepEvt class on unix
- he is having troubles getting the stdHEP code to link on NT
- Solaris version built
- there is a hitch here too. It looks like pions don't interact,
but appear to lose all their energy in the calorimetry via
ionization. We're hoping this is a bug in a system library build
that Tom Pavel will remake shortly.
- Problems with new geometry init package?
- for the first time we have tried running many hundreds of events
through the unix package, with the new file-driven geometry. We
are getting apparent geometry errors. It may be that we have to
take a closer look at it.
Status & ToDo (07/10/98)
- Have obtained the latest GLAST arve and made it work. Not propagated
to the latest nld versions yet.
- Root I/O in GISMO
- the stream conflict remains. TObject defines stream functions as
classes, while the standard library typedefs them
- we may be able to put ifdef's in TOBject to get around this.
- stdHEP generator
interface
- TonyJ has produced an NT version.
- Rob is debugging his modified HepEvt class
- iniData bug
- Joanne discovered a bug in the iniData geometry definition
wherein all materials were sensitive. This fix will be propagated
to the latest versions on unix and NT
- Solaris version built
- the Makefiles were suitably modified to also generate the
binaries on Solaris
- ToDo
- new geometry elements
- (Chad) add beampipe, VXD and tracking layers. Maybe small
angle masks too.
- (Joanne) time benchmarks for nld (GeV/sec), with and without
digitization
- (Richard) complete the TrackSystem class and do the upgrade for
new digitization classes.
Status & ToDo (06/25/98)
- Tony Waite and Joanne Bogart are starting to work on the project.
- the NBeam generator has been installed in the 'prod' nld version on
unix. The code version philosophy for now is to be that the $NLDROOT/gismo
version is the latest and named other versions (eg nld-june) are
older.
- Have not had closure with Toby to get the latest arve, for the
MCParticle stack mods.
- Root I/O in GISMO
- Chad has succeeded (Word
doc file) in making Visual C++ build a root project.
- an attempt to do the simplest root initialization (ie init root
itself) from the nld main routine failed very badly in
compilation, complaining about ambiguous symbols for istream and
ostream. Mail has been sent to root and GISMO centrals.
- stdHEP generator
interface
- Rob has succeeded in doing the 'Hello World' analogue. He can
write out and read back in hand-generated /HEPEVT/ information.
Next steps are to try it out with Isajet and then to revise the
hepevt generator class in GISMO to read in such stdHEP files and
create the appropriate MCParticle list.
Status & ToDo (06/19/98)
- the NBeam generator has been ported to unix. It remains to be
inserted into the public version.
- Toby has finished his revamp of the MCParticle stack mechanism. We
need to obtain the new arve to carry on with the remake of the
digitization scheme.
- Root I/O in GISMO
- Prototype root I/O has been done, with TMaps and TClonesArrays
being written
out and read
in..
- apparently g++ Makefiles exist for AIX and Sun.
- stdHEP generator
interface
Status & ToDo (06/12/98)
- we have a working NBeam generator to replace the single-particle
TestBeam version. It will be installed in the public versions shortly
- Toby has decided to rework the MCParticle stack mechanism so that
shower particles from a given parent are dealt with before moving on
to a new parent. This will make the Particle vs energy map easier to
generate.
- Root I/O in GISMO
- we are starting to prototype embedding root I/O into GISMO. The
first step is to learn how to do root I/O in a root environment.
This will be done progressively with maps and vectors to emulate
the expected data structures.
- a hitch has been uncovered: the AIX and Solaris versions of root
were not compiled with g++ and a feature of c++ is that you can't
mix compilers. So I have to hope that the rooters come up with a
g++ version on at least one of our unix platforms. Or I can check
with Toby on the practicalities of recompiling GISMO with xlC.
- stdHEP generator
interface
- next we try to use the Fermilab /HEPEVT/ interface.
- it's not clear whether there is an NT version.
- Public area on SLDNT17
- the folder nld\ contains the GISMO code and a recent version of
root.
Status & ToDo (06/02/98)
- See 5/13/98 report for the big picture.
- we very nearly have a working NBeam generator to replace the
single-particle TestBeam version.
- we have a prototype version of a decomposing the energy
deposited in a given hit into the energy contributions from each
MCParticle. The prototype needs to be shoe-horned into GISMO.
Status & ToDo (05/13/98)
- Chad has moved on and was replaced by Rob Shanks on 5/4.
- the initialization file interface for specifying the detector
geometry is complete
- disk space on the SLAC central unix system has been obtained: /afs/slac.stanford.edu/g/nld.
- the latest version of GISMO/NLD has been ported there. We have
set up the directory tree to allow for architecture-specific
binaries (AIX and SunOS are supported at SLAC).
- root v2.0 has been installed here
- We are in a phase now of making the project suited for exploring the
detector parameter choices. These involve a number of hefty analysis
tasks and infrastructure improvements:
- create a multi-particle diagnostic generator to allow study of
particle separation in the calorimeter
- do clustering in the calorimeter
- do track-cluster association
- untangle overlapping charged & neutral particles in the
calorimeter
- wire up an MC particle history
- replace the ASCII output file with root I/O
- implement FNAL's StdHEP package for I/O of /HEPEVT/. Allows easy
use of most generators.
Status & ToDo (03/20/98)
- GISMO calorimetry seems valid based on its rendition of SLD's liquid
argon/lead calorimeter. Differences
between GISMO and SLD GEANT LAC simulation seem reconcilable. NLD
results appear to be established.
- To Do
- complete.ini file interface for defining detector parameters
- start validating GISMO shower response for e,p,m,g
- create a 'diagnostic' generator resembling SLD's MCDIAG
- create MC interface for study beyond testbeam particle
generation
Status & ToDo (02/13/98)
- The last two months have been devoted to validation
of GISMO's shower code. We have found that the EM response and
resolution match expectations reasonably, but that the hadronic
particle resolution is impossibly optimistic when compared
to SLD's liquid argon/lead calorimeter. Since finding that we have
been trying to track down how it is that the GISMO/GHEISHA code gives
such fine resolutions.
Status & ToDo (11/12/97)
- Done
- code now runs on unix!
- Toby discovered a typo in the writing code
- still need to allow input of detector parameters via .ini
file (Chad)
- To Do
- start validating GISMO shower response for e,p,m,g
- perhaps create a 'diagnostic' generator à la SLD's
MCDIAG
- we need to settle on detector dimensions
- complete .ini file interface for defining all parameters
from file
Status & ToDo (10/24/97)
- Done
- batch interface defined
- discovered problem with unix port in writing output to file.
Doesn't happen on the PC; reported to Toby.
- still need to allow input of detector parameters via .ini
file (Chad)
- To Do
- as for previous status.
- as for previous status.
- we need to settle on detector dimensions
Status & ToDo (10/15/97)
- Done
- ported PC code to AIX unix
- still need to link with batch interface and allow input of
parameters via .ini file (Chad)
- To Do
- Tracks
- define an output format
- update ntupler accordingly
- how to deal with decays of long lived particles (eg p± from
K0's)?
- make ISAJET available to run in GISMO
- has been f2c'ed.
- We are pursuing two pronged approach:
- allow ISAJET to write /HEPEVT/ to an output file,
using FNAL's StdHEP package.
- imbed ISAJET in GISMO directly. This will require
CERNLIB on the PC.
Status & ToDo (10/9/97)
- Done
- EM & HAD endcaps defined. No known problems.
- ntupler backfitted for new components. We have sample ntuples
(PAW & Root).
- define parametrized tracks
- ToDo
- port PC code to unix
- an old version of the code is on unix and compiled. Have
Makefile from Matt Langston to do the link. Need to understand
it! Or make Toby's simpler version work. (Chad)
- Tracks
- define an output format
- update ntupler accordingly
- how to deal with decays of long lived particles (eg p± from
K0's)?
- make ISAJET available to run in GISMO
- has been f2c'ed.
- We are pursuing two pronged approach:
- allow ISAJET to write /HEPEVT/ to an output file,
using FNAL's StdHEP package.
- imbed ISAJET in GISMO directly. This will require
CERNLIB on the PC.
Status & ToDo (9/24/97)
- Done
- define endcap calorimeters
- clone hadron calorimeters
- ToDo
- Geometry hitch
- there has been a hitch with the geometry. Currently it looks
like a problem in the GISMO handling of the endcap SlicedCone
class when particles come very close to the inner radius edge
of the cone.
- backfit ntupler to accomodate the addition of the region to
the channel ID.
- put code back into MS Sourcesafe for management (Chad).
- we have a version of the code in SourceSafe, but it doesn't
seem very user-friendly!!
- port PC code to unix
- an old version of the code is on unix and compiled. Have
Makefile from Matt Langston to do the link. Need to understand
it! Or make Toby's simpler version work. (Chad)
- define parametrized tracks
- make ISAJET available to run in GISMO
Status & ToDo (9/16/97)
- ToDo
- define endcap calorimeters
- do EM cal (Richard)
- there has been a hitch with the geometry. Currently it
looks like a problem in the GISMO handling of the endcap
SlicedCone class in converting from global to local
coordinate systems (for z).
- clone for HAD (Chad)
- backfit ntupler to accomodate new digitizations plus any
changes to existing ones.
- put code back into MS Sourcesafe for management (Chad).
- our first flailings have failed to give a reliable system.
Keep at it. A manual may be needed!
- port PC code to unix
- code is on unix and compiled. Have Makefile from Matt
Langston to do the link. Need to understand it! (Chad)
- define parametrized tracks
- make ISAJET available to run in GISMO
Status & ToDo (9/8/97)
- Chad Colgur is now on board, as of Sept 2!
- Done
- Toby Burnett is done with defining the layered medium class for
the endcap.
- ToDo (short list!)
- define endcap calorimeters
- do EM cal (Richard)
- clone for HAD (Chad)
- backfit ntupler to accomodate new digitizations plus any
changes to existing ones.
- put code back into MS Sourcesafe for management (Chad).
- port PC code to unix
- code is on unix and compiled. Have Makefile from Matt
Langston to do the link. Need to understand it! (Chad)
- define parametrized tracks
- make ISAJET available to run in GISMO
Status & ToDo (8/8/97)
- Done
- write ntuple-maker from the ASCII output file. (Dave Wagner)
- have looked at 5 GeV p, m, e into the barrel:
- electrons:
e/m~0.9 with s/Root(E)~14%
- pions:
p/m~0.5
- muons: give ~expected min-I response. Note that there are no
fluctuations in Gismo for dE/dx. There could be if we need it.
- ToDo (short list!)
- define endcap calorimeters
- Toby Burnett is nearly done with defining the layered medium
class for the endcap.
- port PC code to unix
- code is on unix and compiled. Awaiting a Makefile from Matt
Langston to do the link.
- define parametrized tracks
- make ISAJET available to run in GISMO
Status & ToDo (7/31/97)
- Done
- barrel EM and HAD calorimeters are defined and being 'read out'
for single particle generator
- ToDo (short list!)
- write ntuple-maker from the ASCII output file. (Dave Wagner)
- define endcap calorimeters
- this awaits Toby Burnett adding a compound medium for the
doughnut shapes we need with the cosq inner radius boundary.
- port PC code to unix
- define parametrized tracks
- make ISAJET available to run in GISMO
|
