Dear Users,
1. Fixed an issue pointed out in http://afsad1.kek.jp/redmine/issues/230, a mishandling of FRINGE parameter in MULT, The ON/OFF of Maxwellian/Linear fringes of K0/SK0 have been wrong. This bug was there since 1.0.10.5.5a60 as written above.
The table above looks strange due to the contraction of spaces by the server. What I meant was
Axis: axis[0]
Both : major + minor
All: major + minor + axis[0]
Automatic: major + axis[0]
Dear Users,
These changes are for k64, but hope to be included in all versions later.
1. A new flag RADTAPER is defined to enable "automatic tapering", which scales the strength of all magnets except solenoids along with the local energy deviation of the closed orbit during the tracking and the emittance calculation. To use that, first the 6D closed orbit must be calculated and stored in Twiss["DDP"] by EMIT or Emittance[] with flags
RADCOD; RFSW; CODPLOT; RADTAPER;
Then do tracking with RAD . Please not that since either CALC or GO overwrites Twiss["DDP"] by 4D orbit, so do not do CALC or GO after EMIT.
RADTAPER defaults OFF.
2. The option GridLines in plots now can take new values Axis, Axis[z], All . Axis draws a straight line at zero if it is in the plot range. Axis[z] draws a straight line at z, if it is in the plot range. All draws all of Major, Minor, Axis.
These values can be put in a list.
major minor axis
Minor *
Major *
Both * *
Axis *
All * *
Automatic * *
None
3. Some sqrt's in the code are replaced with hopefully-more-efficient routines, but the effect will be minor.
Dear Users,
The changes below are for k64, but should be applied for any version.
(New Functions)
1. Tr[m] to return the trace of a matrix m. Tr[m, f] uses f as the head, instead of List.
2. Diagonal[m] to return a vector with the diagonal components of a matrix m.
3. Rational[n,d] to represent a rational n/d. A few primitive arithmetics are implemented.
4. FromRational[Rational[n,d]] to return a real number n/d.
5. ContinuedFraction[x] to return the continued fraction of a real number x as a List. ContinuedFraction[x,n] returns the firs n components.
6. FromContinuedFraction[f_] returns the Rational corresponding to the continued fraction f.
7. PeriodicR[m] to return the x-y coupling R-matrix to decouple m into diagonal matrices A and B, It returns {R,A,B}. It may return Complex when m is unstable. Only 4x4 is allowed for m.
8. PeriodicOptics[m] to return the periodic Twiss parameters for a 4x4 matrix m. It returns in the same format of the output of Twiss with 27 components. PeriodicOptics[m,d] uses the solution of (I-m).z == d for DX,DPX,DY,DPY parts of the output.
9. FindRoot is extended to accept the form eqs1==eqs2, where eqs1 and eqs2 are lists. Also it allows complex values for the equations. If Complex does not appear in the initial value of the equations, a option Complex->True should be specified. Also Complex->True should be used when the parameters have upvalue definitions for Set. The parameters must be always Real.
10. ArcTan now takes Complex for the arguments.
(Bug fixes)
1. A bug was found in Canvas$Legend to destroy a symbol `y.
2. A bug was found in Complex[a,b] when a or b is not a number.
3. Twiss returned wrong values when the Keyword was "*" (k64 only).
4. MinMax was wrong when the argument was Complex.
Dear Users,
1. (k64 only) Corrected a bug in Sum/Product to produce wrong results. This was there since initial transfer to k64.
2. Signs of CHI1/CHI2 of MULT are now reversed when the component has negative orientation. They should be so as the general rule of the reversal applies. There still remain some exceptions such as DPX/DPY of SOL, but they are not touched yet.
3. The print out of the result of DynamicApertureSurvey was changed for better visibility.
4. A bug was found in saving parameters of MARK, when it is not at the beginning of line.
answer for my question is clear (see below). so we hope that "alternative" way will be discovered soon.
thanks.
Yury.
==================================================================================
No. : 10079
Title : Shutdown of "www-acc-theory.kek.jp"
Date : 2015/02/13(Fri) 18:19:21
Name : KAMADA Susumu <susumu.kamada@kek.jp>
URL :
The server "www-acc-theory.kek.jp" stopped its function forever.
Accordingly, SAD source can not be downloaded by using CVS access.
The alternative way will be prepared some day by somebody.
==================================================================================
> is the access to CVS repo of SAD is still avaliable?
>
> i'm trying to get latest snapshot of repository, but CVS login fails because of host resolve errros (output below).
>
> [1]-i-(6016)-/tmp% cvs -z3 -d :pserver:anoncvs@www-acc-theory.kek.jp:/Users/cvs/SAD checkout -P oldsad
> Unknown host www-acc-theory.kek.jp.
> [1]-i-(6017)-/tmp% host www-acc-theory.kek.jp
> Host www-acc-theory.kek.jp not found: 3(NXDOMAIN)
> [1]-i-(6018)-/tmp% traceroute6 www-acc-theory.kek.jp
> traceroute: unknown host www-acc-theory.kek.jp
>
> but for another hostname I can obtain IP adress.
>
> [1]-i-(6019)-/tmp% host acc-physics.kek.jp
> acc-physics.kek.jp is an alias for kekacclserver.kek.jp.
> kekacclserver.kek.jp has address 130.87.106.66
is the access to CVS repo of SAD is still avaliable?
i'm trying to get latest snapshot of repository, but CVS login fails because of host resolve errros (output below).
[1]-i-(6016)-/tmp% cvs -z3 -d :pserver:anoncvs@www-acc-theory.kek.jp:/Users/cvs/SAD checkout -P oldsad
Unknown host www-acc-theory.kek.jp.
[1]-i-(6017)-/tmp% host www-acc-theory.kek.jp
Host www-acc-theory.kek.jp not found: 3(NXDOMAIN)
[1]-i-(6018)-/tmp% traceroute6 www-acc-theory.kek.jp
traceroute: unknown host www-acc-theory.kek.jp
but for another hostname I can obtain IP adress.
[1]-i-(6019)-/tmp% host acc-physics.kek.jp
acc-physics.kek.jp is an alias for kekacclserver.kek.jp.
kekacclserver.kek.jp has address 130.87.106.66
Dear Users,
1. A problem was found in the handling of ElementValues, when multiple components belong to an element. Corrected Couple.n .
Dear Users,
1. More corrections are made for trackd, to avoid unnecessary wait for children which do not exist.
Dear Users,
1. Mod is modified to work for nested lists such as Mod[{{1,2},3,4,5},3] .
2. RealListQ is changed to return False for a structure whose head is not List, eg., RealListQ[a[1,2,3]] .
3. Plot is changed to use the real part if the result of function becomes complex.
This bug was reported at:
http://afsad1.kek.jp/redmine/issues/151
Dear Users,
1. A bug in DynamicApertureSurvey (trackd.f) was corrected. The bug has been reported since a long time ago, but now patched, hopefully. It is common for both k64 and legacy versions.
> Dear SAD experts,
>
> I have installed SAD main trunk (not a k64 branch) to KEKCC, which is operated with 64-bit Scientific Linux 5.6.
>
> I installed and compiled gcc 4.9.2 with multilib option, and used it to compile SAD, with following additional option in sad.conf:
> FC=gfortran
> ABI=32
>
> I need to copy libtcl8.5 by hand to $SAD_ROOT,
> and need to call 32bit version of libgfortran.
>
> Then I can run bench2.sad successfully except for one message:
>
> Note: The following floating-point exceptions are signalling: IEEE_OVERFLOW_FLAG IEEE_DENORMAL
>
This message is generated by GNU Fortran runtime.
See following document in detail
https://gcc.gnu.org/gcc-4.9/changes.html
https://gcc.gnu.org/onlinedocs/gcc-4.9.0/gfortran/Debugging-Options.html
Dear SAD experts,
I have installed SAD main trunk (not a k64 branch) to KEKCC, which is operated with 64-bit Scientific Linux 5.6.
I installed and compiled gcc 4.9.2 with multilib option, and used it to compile SAD, with following additional option in sad.conf:
FC=gfortran
ABI=32
I need to copy libtcl8.5 by hand to $SAD_ROOT,
and need to call 32bit version of libgfortran.
Then I can run bench2.sad successfully except for one message:
Note: The following floating-point exceptions are signalling: IEEE_OVERFLOW_FLAG IEEE_DENORMAL
Detailed log file can be found at
SAD cluster:~nakkan/Work_SAD/bench2.log_KEKCC
I understand SL5.6 is not supported officially, but if you have any advice to us, it will be more than welcome.
Hiro Nakayama
Dear Users,
1. A bug was found in ListDensityPlot, miss-spelling "If" as "if".
* SAD_1_0_10_5_8a8
** Fix lock race in Shared function family(*1)
* SAD_1_0_10_5_8a9
** Improve GCC support(checking feature support)
* SAD_1_0_10_5_8a10
** Support Tcl/Tk 8.5.18
(*1)
* DESCRIPTION
** SAD shared memory access function(ReadShared[], WriteShared[], Shared[], Shared[_] = val_) has lock race.
* IMPACT
** This errata makes a corrupted data or causes crash in ReadShared[].
* AFFECTS
** MAIN trunk src/tfsahred.f revision 1.10 or prior
** k64-1-6 branch src/tfsahred.f revision 1.10.2.22 or prior
* FIXED
** MAIN trunk src/tfshared.f revision 1.11
Dear Users,
1. Corrections are made to clip the momentum of particles in tracking, which might become less than or equal to zero due to deceleration in CAV, MULT, and wake. This issue is bot for k64 and the legacy version.
The server "www-acc-theory.kek.jp" stopped its function forever.
Accordingly, SAD source can not be downloaded by using CVS access.
The alternative way will be prepared some day by somebody.
Dear Users:
1. A bug was found in calling built-in functions with two args such as ArcTan. This is k64 only.
> 1. The definition syntax of Class is changed to avoid a crash when something like C=Class[{},{},] is declared by mistake. This change should not affect anything for a properly written program.
>
手元の環境でのamorita branch上の検証では、
* gsから実行すると C = Class[{}, {}, ]の評価時に不正命令トラップが発生する
* sad*.exeを直接実行し FFSに入って C = Class[{}, {}, ]を評価するとGeneral::deepになる
という症状で、原因は次のようなものであると推定しています
(*1) C = Class[{}, {}, ]の評価時に、ClassのUpSetDelayed定義にループが存在し式評価が無限ループする
(*2) gsスクリプトでマシンスタックが制限されているため、SADスタックが尽きる前にマシンスタックが尽きている、もしくは、スタックオーバーランの結果メモリ破壊が発生している
Classの定義修正は(*1)に対する修正
Hourly rsync update of SAD CVS mirror pserver is stopped because of missing of login shell on mirror pserver.
Dear Users,
This BBS has been updated to suppress spams. The previous contents are available at
★acc-physics.kek.jp/cgi-bin/SAD/SAD-BBS/wforum.cgi
(★=http://)
スパム対策の為、SAD BBSをこちらの掲示板に移行致しました。
(生出施設長、鎌田先生 確認済)
直近のスレッドはこちらの掲示板にもコピー移行しておりますが、それ以前の投稿履歴をご覧になるには下記アーカイブをご参照くださいませ。
★acc-physics.kek.jp/cgi-bin/SAD/SAD-BBS/wforum.cgi
(★=http://)
Dear Users,
1. The definition syntax of Class is changed to avoid a crash when something like C=Class[{},{},] is declared by mistake. This change should not affect anything for a properly written program.
Dear Users,
1. (k64 only) A bug was found when the second argument of Partition is non-positive.
2. ListContourPlot now takes an option ColorFunctionScaling. If it is False, the value of the list is directly pased to the color function. The default is True.