Manuals are placed at the end of this page
From Epics9.20 (Oct. 2016)
(with Cosmos8.00 or later), Epics complies with the
standard Fortran Grammer and it
can be compiled by GNU Gfortran.
Epics9.30 Epics9.30.tar.gz (202575416 B; May.04 2020).
A data for Alumina + SiO2 is added forElsepa based multiple Coulomb
scattering. Standard Air composition is bit changed (Ar 0.5 %-->1%)
Some update of the direct pair creation is included.
(though new feature is not tested yet fully).
Epics9.28 Epics9.28.tar.gz (174591301 B; Nov. 25 2019).
The process of the direct pair (DP) creation of e- + e+ was implemented.
This may be used when high energy (>> 100 GeV/n hadron(p, He, C...)
hit a detector containing heavy materials).
You can control it by using 3 parameters in epicsfile:
DirecPair (T or F); default T is to include the procces.
BareTaget (F or T): default F is to regard the target be neutral
(i.e, surrounded by electrons) else target is fully ionized.
AdjFac: 1.0. Default 1 means if the DP effect is <1% of
the usual ionization (knock-on) process (2Z^2MeV/(g/cm2),
the DP process is not considered. If AdjFac is not 1,
percentage of AdJFac*1 is employed.
The process name of DP inside Epics is "dpair". (Move%proc
contians "dpair" when DP happens).
Epics9.26 Epics9.26.tar.gz (222798266 B; Feb. 2 2019).
Though, the release of the v9.26 was announced more than month ago,
it could not run correctly (stoped soon; when Eg is very close
to the threshold (~1.022 MeV), the probabilty of pair creation / r.l
becomes very small (< 10^-8) and interplation doesn't work correctly
(resulting in a negative value).
So the next part was revised
and new one was released without changing the version number.
In the photon pair creation cross-section treatment in the partial
screening region, a bug was found; there is small gap in the
x-section between the partial and complete screening regions, but
it is very samll (< 1 %). However, the gap was enlarged depending
on the situation due to the bug.
Although the bug is expected not to give a visible
effect in many applications but in some cases where the pair creation
probablity of gamma rays of energy <100GeV
in some limitted detector regions is important, we may need to
consider its effect.
One example of wrong pair creation probabilty effect is shown
in this PairXsecGap.pdf
For the bug correction, only one subroutine may be updated. But in this
version we added a table based on Tsai's pair creation formula in the partial
screening region. The default table is Nelson's one as in the old
version. To switch to Tsai's one , we may do as follows
(1) Go to Epics/Data/
(2) rm Media
(3) ln -s MediaTsai Media
No recompilation is needed.
The other choice of the pair creation cross-section is to use Xcom data.
This can be done by editting "epicsfile" in your application and
give a value of 50~100 (GeV) to Excom2.
The difference among the three cross-section is amsll (but visible
at Eg ~10 MeV region. This would not lead to visible difference
in the results of many applicatioos. Which one is the best
is not clear.
Epics9.25 Epics9.25.tar.gz (218156246 B; Dec. 7 2018).
Oct.17 version lost by mistake. but this should be the same
Almost no change from v9.24.
Now the user can give "special" in XsecModel in param,
so that the user can give arbitray hadronic x-section
values different from the currently active hadronic
See Epics/UserHook/UseXsecModel (or Cosmos/UserHook/UseXsecModel)
Epics9.20 Epics9.20.tar.gz (97660165 B)
This version should be used with Cosmos8.00. From this version,
Epics follows the formal Fortran grammar and can be compiled
by GNU gfortran too. If you have your own application programs
in the style accepted so far,
they must be modified to comply with the standard formal
Fortran grammar as noted in Cosmos8.00 page.
This version includes a bug fix which may happen when
some components have explicit rotation and a neutral
hadron passes such component and makes multi-particle
production in a non-rotated component; this results in the error as if we had
very large Pt particle produciton.
See also, Epics9.20.pdf.
This version also includes updated use of the "FreeC" parameter (V1ry.pdf),
which can be used to constrain the first interaction point
and interaction type.
Epics9.167 (Feb. 06, 2015) Epics9.167.tar.gz (48054842 B)
It's found that octagon_y cannot be displayed correctly when
intel fortran optimization level is -O2 or -O3. So for the
compilation by intel fortran, optimization for octagon was
made to be -O1 (-O0 is also ok).
All cmplx(x,y) have been replaced by cmplx(x,y,8), though
except for ciecone case as described in 9.166,
no harmul effect has been found.
Epics9.166 (Jan. 27, 2015) Epics9.166.tar.gz (47580197 B)
The difference from v9.165 is that all correction shown by
Note in the 9.165 part are included.
Epics9.165 (Jul. 5, 2014) Epics9.165.tar.gz (47745377 B)
This icludes update for the photon LPM effect treatment as mentioned
above, and related utility routines (in Epics/Util/Elemag/BremPair)
such as for creating
the sampling table of a new material.
For versions from 9.15 to 9.164.
The LPM effect for the photon is not properly treated (earlier
versions are OK).
To include the LPM effect for
photons with energy > 4TeV in heavy material such as W, Pb etc
must be expanded in Epics/Data/Media (Go to there, and
use the "tar" command as "tar xvfz path_to_this_file",
where path_to_this_file is ~/To_be_put_inside_Media.tar.gz, if you
place the file in your home.
Visible effects may be expected if incident
photon energy is much higher than 5 TeV
and heavy targets are involved.
For hadron incident case, fluctuation is large so that
we expect such energy scale would be > 100 TeV/n.
In versions > 9.164, update has been included.
Epics9.164 (Jun. 26, 2014)
Epics9.164.tar.gz (56354279 B)
In this update, there is nothing which may affect the simulation
result for existing applications. Updated points are 1)
the user can use any volume-shape (such as box fpolygon etc)
as a world with _w. Then, an implicit world so far may be
specified as an explicit world with _w. 2) the max name length
of a volume-shape is now 16 (old one 12) so as to be
compatible with e.g, honeycomb_yz_w. 3) With implicit worlds
we cannot deal with correct sub-detector hirerarchy information.
So if we need correct sub-detector hierarchy information,
we have to use explicit world. Now it is possible.
4) So far, no method has been supplied from Epics for knowing
the component number of a particular object, e.g
N-th SciFi at L-th layer. Now such a method is
available, and it uses sub-detector hierarchical
information. Also a utiltiy command (tree2index) is
5) It is now possible to give aliases to a medium name.
6) A manual for these is ready:
7) Two new media are added (some wood and steel) being
use by the Tibet experiment.
Epics9.163 (May. 21, 2014)
Epics9.163.tar.gz (55730891 B)
This is an update for better performance for the Intel compiler
by using -O3 option (Don't use -O3 option for the older versions--
it will make the speed 10 times slower).
Now all component list will be printed in defalut when the
program reads config file.
This version must be used with Cosmos7.645 (though 7.644 may be
This version can run on JAXA super computer environment (use
site.configJaxaflat or site.configJaxaVISIMPACT).
Epics9.161 (Dec. 03, 2013)
Epics9.161.tar.gz (55728175 B)
This version must be used with Cosmos v7.641 or higher.
Although very rare, in earlier versions, the first interaction
point informtion might be wrong (<< 1/1000 events; so we cannot
see the difference in the first interaction point distributions.
simulation itself was OK). This was fixed.
QGSJET-II-04 and EPOS-lhc-v3700 are now usable in EPICS,
too. However, Sibyll2.2 cannot be used since it is almost dedecated
to Air target (so for Air it can be used).
Quenching treatment was changed.
It must be managed in epGUI( in epUI.f) of each UserHook.
UserHook/Geometry/ was updated to print r.l in addition.
Geometry program got neurotic in some case when
incident position is just on the boundary of two
or more components. This was fixed by updating prog/KKlib/kxplhorse.f
Also some workaround was prepared for such case.
When some components (B's) are not fully contained by
anothr component (A), B's must be contained by
still another component (X). If X is the world,
the user may list B's like
321 box_w sp 0 0 0 / 0 0 0 / list of B's.
The size of the world could be explicitly given
as in earlier versions.
Epics9.15 (Apr. 7, 2013)
For v.9.15 before Apr. 7, 2013, and other earlier versions,
Intel Fortran compiler v.13 is not usable.
Epics9.15.tar.gz (46014211 B)
The maximum heavy ion projectile for dpmjet3 is now Pb if
the user uses Data/Media (default). Cosmos version 7.633 should be used,
A number of improvment for programs
in Util or Util/Gencol, Util/Geomview has been done.
Some new matterial files have been
added (Au, variant of Al, Steel).
For more details, recent new stuff
and recent new volume-shapes
Epics9.131 (Jul. 10, 2012)
Epics9.131.tar.gz (16899867 B)
The usage of PHTIS which is to be used below 2~2.5GeV was improved
and low energy neutron treatment became better.
Although data (Epics/Data/Media1/BGO.GLB etc) itself is not included in this
version, now dpmjet3 can accept projectile heavier than Fe (upto Pb). In such a case,
Cosmos7.631 must be used.
IncGP to speicfiy the photo-hadron interaction (and also muon N.I) is changed.
Now standard value is 3. See UserHook/Template/epicsfile, or
Cosmos/Version/Readme. Some related bug was fixed.
Polyvinylchloride, CH2CHCl needed in some
applications. Exapand it in Epics/Data/Media/. For safety, the file PVC (up to line 19) may
also be copied to Epics/Data/BaseM/.
For Epics9.13, you may need to issue a command
"manageIntInfo.sh" in your
UserHook/xxx, where xxx is your application forlder other than
those contained in the distribution. (If you have already done so,
you don't need to do so again--though doing twice is not harmful).
The new features in 9.084 and 9.13 are mainly about the inelastic cross-sections
and some details are
The differences between 9.084 and 9.13
are basically the same as those between EPICS9.08x and EPICS9.1x
as has been shown below. (The essenatial difference is that 9.08 series
cannot use AutoEmin. Epics9.1x series can use it.
Probably we don't need to use 9.08x series now.
AutoEin=0 corresponds to Epics9.08x series (i.e, no automatic Emin
AutoEmin=1: fix the Emin by looking at detector shape
and media. This choice gives almost the same result as the v9.08x series
and also as the case of AutoEmin=0.
AutoEmin=2: Besides automatic Emin,
when tracking a particle, particle death and energy
deposit calculation take into account both the distance to the
boundary of the crrent component and the range of the particle.
The difference of AutoEmin is reflected to the lateral distribution of
cascade showers. It is very small and if we plot the lateral distribution
(rho(r)) of energy depoit, using log scale for the ordinate, we cannot see the
difference at all. However, if we look at the ratio of rho(r) as a function of r,
AutoEmin=2 case shows some systematic difference from AutoEmin=0 or 1;
(At small r, rho2 < rho0 or rho1 while at larger r, rho2> rho0 or rho1 systematicaly,
although it is 1 % to 2 % level. The difference seems to become larger
at deeper depths). This difference could be exhibited clearly in some kind of
Currently, we don't know which of the AutoEmin is better. For the
time being we take AutoEmin=1 as default .
The cosmos version7.62 can be
used both for Epcis9.084 and 9.13. (If you find something strange
in Epics9.084, Cosmos7.583 may be employed for 9.084).
Epics9.083 and 9.12are updated version of
Epics9.082 and 9.11; they include updates shown below
Some bug fixes
Knock-on process by heavy ion bug fix
must be modified so that
*   temp * masele * media.basearea * 2 / beta2
*   temp * masele * media.basearea * 2 / beta2
*   * aPtcl.charge**2
Or equivalently, you may replace it by
Photo-hadron produciton usage bug fix
Nproduced = 0 must be put in subroutine cphotop
(at it't top part) which is in Cosmos/Tracking/cintePhoton.f
Or equivalently, you may replace it by
Although this is the file in Cosmos, Cosmos itself is free from
dpmjet3 usage bug fix
For heavy ion collisions, the Glauber data made by the fordpmjet
command was wrong. So for any of the recent Epics versions,
you may use this patch
to expand it in Epics/Scrpt/ (i.e., tar xvfz fordpmjet.tar.gz).
in your application directory (usually under UserHook). (config
is the path to a detector configutation file).
EPICS9.081 / 082/083 and EPICS9.10 / 11/12
Release (Sep 12/Oct.2,2011/Dec.2011).
9.082 and 9.11 are updated versions of 9.081 and 9.10; they can
treat a new volume-shape
(sqTccl) which converts a circle shape into
a square shape).
There are also some upgrades which are not essential. (say,
"fordpmjet" command becomes more versatile. The hookIsOutSide.sh
supports an easy way to place user's application
in a directory other than EPICSTOP/UserHook. A short description
Below, ..Diff.tar.gz files can be used to upgrade exisiting 9.081/9.10
to 9.082/9.11 by unpacking them inside the Epics top directory.
The one without Diff is a complete set.
EPICS9.081/9.082/9.083 must use Cosmos7.582
EPICS9.10/9.11/9.12 must use Cosmos7.61
There are rather big differences between EPICS9.08x and EPICS9.1x.
EPICS9.10 introduced many new features such as automatic Emin,
ability to fix some parameters by specifing a paticular detector component
for particle tracking. Also it changed the treatment of "pipe_y" type
shape specification. Therefore, there might be
some unforeseenable defect. In such a case, we must go back to EPICS9.08.
However, we need to use new component shapes (honeycomb/octagon) which
requires the revised method of treating "pipe_y" type notation.
So, that feature only has been
implemented in EPICS9.08 and it was renamed to EPICS9.081.
See a manual updated parts of EPICS
In this connection, the max character length for the volume-shape
(box etc) is changed from 8 to 12. (Say, "honeycomb" can be
treated without short form such as honeycm). If the user
uses the query routine (epqstruc) to get the volume-shape, the character
length to receive the shape must be >= 12.
For details, See a manual:
About new treatment of underscore in voluem name
(octagon and honeycomb)
Cosmos7.581 has very small difference form 7.58: the jam code
fails to annihilate stopped anti-proton. The conde conversion
of eta and lamda0 from QGSJETII is wrong.
These were corrected
Cosmos7.60 can use "sofia" code for photo-hadron production.
The jam code in these versions are the same as in older versions.
Release May. 11 2011.
FirstKiss example uses FirstM which is not usable for e/g incident.
So previous 9.08 (Jan 2011) was replaced without chagining the v.n.
Also some messages related to the boundary search problem will not
Interaction model 'jam' (2GeV~* TeV) can be used for hadrons.
For heavy ion, under investigation. jam seems to be little
bit better(?) than dpmjet3. Large memory size problem
written for direct pair creation by muon in 9.07 was fixed.
Typical interaction model with jam is
IntModel='"phits" 2.5 "jam" xxx "dpmjet3"' (xxx may be 5 to 100;). At very high energies, dpmjet3 would not be very good.
Release Jan. 21 2011.
The version is corrupted: Not usable
Some garbage was remvoed. Diret e+/e- creation by muon at lower
energies (< 150 GeV) was not accurate so it was improved. (To see
muon single peak, it has no effect). At present, it uses large
memory size; this will be updated in near future.
Release Dec. 28 2010.
Util/Gencol/Gencol.f has bug for moving target and nucleus projectile.
This was corrected.
Release Dec. 26 2010.
prog/Light/epLightCoundDE.f has a bug about mnCounter (by Akaike).
This was corrected.
Release Nov. 16 2010.
The x-section calculation for phits was wrong (total xs= inela xs +
2*ela xs was corrected to be total xs = inela xs + ela xs).
UserHook/GetXsec is added to get x-section for various interaction models.
Release Nov. 13 2010.
Now phits and Jam are stable. (9.00/9.01 has some problem
related to stopping muon treatment).
Release Oct. 17 2010.
Versions before Oct. 17 cannot treat heavy ion by phits.
Now, low energy hadronic interaction models employed in Cosmos7.53
are availble for use. (i.e, you need Cosmso7.53 or newer).
Scintiltion and Cherenkov light generation and propagation can be treated.
You nees intel fortran compiler version 11. For Epics, it might be
possible to use non-zero optimization, but default is -O0.
Release Jan. 30 2010.
fordpmjet command cannot treat Air*0.564 type notation. This was
epicsfile in FirstKiss has wrong values for RecoilKeMin
; 100 should be changed to 100e-6 or 0. Related documents
Release Jan. 17 2010.
dE/dx related stuff has been updated to use most recent data.
Release Jun. 11 2009.
Intel compiler 10.1.021 results in loop So
don't use it. Versions 10.1.012/10.1.022 seem to work.
Release Nov. 9 2008.
For compound material and hadron projectile, cross-section treatment
is made to be more accurate.
For this version of Epics, you better to use
Cosmos version7.38 or later for stable execution (for media containing H;
when pi- and lambda bar collide with H, infinite loop may happen in the older
Release Jul. 09 2008.
Stability increase in accordance wiht Cosmos.uv7.36. fordpmjet works
for complex config, too.
Release Feb. 20 2008.
Userdefined parameters are now definable only in sepicsfile.
Release Jan 26 2008.
Sorry, previous 8.73's was wrong. This version should work.
In uv8.72, one include file was updated but not included. This was corrected.
Release Nov 24 2007.
IncGP=t is not effective for photo-hadron production during some recent
versions. This was corrected.
Characeristic X-ray emisson at photoelectric effect is now considered.
If you have a valid uv8.71, the difference fileEpics.uv8.72Diff.tar.gz
(50496 B) may be used. (expande it inside 8.71)
Release Jun. 25 2007.
In version 8.68, LPMeffect was introduced to control the LPM effect;
but due to a mistake, it was not effective (LPMeffect=f dose not work);
this was corrected.
Release Jun. 04 2007.
(May version was not updated fully. The same version number but more complete).
Incident particle specification method is described more clearly
in FirstKiss/sepicsfile. Some confusing specifications such as InputP='gsph',
'sph' are now removed. UserHook/Test-sepicsfile/ contains some document
and explanations. It is the same one as linked at the last part of this
Release Apr. 6 2007.
If you have a valid uv8.68, the difference fileEpics.uv8.69Diff.tar.gz(23582 B) may be used.
A bug fix; component selection by specifing media at config display
became impossible in 8.67 and 8.68. This was corrected.
Default max number of components is now larger than previous ones.
If you have a valid uv8.67, the difference fileEpics.uv8.68Diff.tar.gz(39887 B) may be used.
New epicsfile parameters(Excom1, Excom2 and LPMeffect) can now be used in epicsfile.
For these, see epicsfile in FirstKiss.
Now the parameters in sepicsfile and epicsfile may appear in any order or may not appear.
This version can be compiled on Mac OS X with Intel Fortran.
If you display only a subdetector by 'drawconfig', the media
of each component is shown wrongly. This was corrected.
Subdetector name can be as long as 16 characters now.
The xcom data is used for Xray region (>1keV) treatment. Coherent and
incoherent scattering and photo-electric effect become more accurate.
This verson needs Cosmos.uv7.23 or later.
Util/Gencol is enhanced so that it can generaate collisions which may
change the conditions event-by-event. e+ anihilation cross-section
was wrong; this was corrected. (Essentially no effect is seen).
site.config for 64 bitInterl Fortran is ready.
If you have a valid 8.63,
may be expanded inside 8.63.
Bug fix for treatment of the elliptic cylinder.
+filnename type primary file specification permits now
preceeding folder name(s) ( say, somefoler/another/+primary).
If you have a valid 8.62,
may be expanded inside 8.62.
Bug fix for how to treat time information when an anti particle (
typically anti-proton) stops and anihilates. The older versions
added a big time for a stopping ptcl; this would result in
a strange time info. for descendents. Now, we don't add any time
when anihilation takes place.
In some case when a component is rotated by 180 deg., tracking becomes
wrong. This is corrected. When two cylinders are contact and an incident
particle is exactly on the surface of both cylinders, the tracking
might be fail depending on the particle direction. This was corrected.
A small bug for e+ anihilation angle is corrected.
A routine, epretTrack, is added to make it easier to handle multi-particle
Epics.uv8.54.tar.gz( 9777536 B)
This version supports explicit event genertor by various interacion
models (Util/Gencol). and multi-partilce incident particles in
a file created by, say, Gencol. Gencol can use QGSJET1 and Sibyll
if you have corresponding codes.
Epics.uv8.53.tar.gz( 9771930 B)
This version supports QGSJet-II. You need Cosmos.uv7.01 or later.
Version 8.50 to 8.52 have wrong treatment of xsection of Pb target
Epics.uv8.14.tar.gz( 9766414 B)
Source code files organizaton changed so that unresolved ref. may not
appear when drawconfig, testCnf1, usenewvol etc are used.
A new volume shape, ciecone (cut inclided eliptic cone), is now public.
This can be used to make pants.
Epics.uv8.13.tar.gz( 9667111 B)
Bug for +primary specification (with subKEdir) and uniform spherical input
particle postion was corrected. (8.11 & 8.12 had some tribial mistakes
and deleted )
Epics.uv8.10.tar.gz( 9699879 B)
Util/Anime is added to make a movie which shows time development of cascade
showers. In the angle sampling, 'cos 1' specification was wrong.
It was equivalent to 'cos -1'. This is corrected.
The current version has larger
If you made tables yourself for some media,
please remake them. The most big changes in this version
from the older ones are: pair creation formula by Nelson is
now standard. The absolute value of muon sinlge peak is
corrected by about 5% (increased).
Various checks have been done for the current version and the
results will be presented soon.
In 2002 summer,
due to collapse of 4 different disks in a day (three of them were
raid 5 systems), almost all Epics stuff was lost.
Version uv7.24 was the first recovery version.
WarningIntel fortran (ifort), with -g option,
may lead to any strange resutls. So -g option should not be used. (experience
at ifort v10.1).
The default treatment in site.config is no -g option.
In recent ifort versions, the bug seems absent.