Index of /~dlr/g-2/bmad_tracking_code/g-2/documentation
May 17, 2014 - Notes about tracking programs
Main = tracking_master
All subroutine are in code/ directory
Need to figure out how to rebuild finalfocus/ and magneticfield/ after having put all subroutines into code/
tracking_master now propagates through backleg and inflector
Inflector apertures at each end are defined in lattice file for rectangular aperture.
The "D" shape of the aperture is accessed in TRACK1_CUSTOM - for these elements (check TRACK1_CUSTOM to see whats incorporated)
Also scattering in coil ends and cryo window is from TRACK1_CUSTOM
Scattering is no longer controlled in input.dat. It can be turned off by removing the marker elements from the lattice file or
editing TRACK1_CUSTOM
MARK_CRYO_US
MARK_INFLECTOR_US
MARK_INFLECTOR_DS
If in the lattice file
inj: line = (inject, inj_to_ring, ring_br) then particles are tracked through the backleg
inj: line= (injinf, inj_to_inrg, ring_br) then particles are tracked using taylor maps
In both cases, the distribution emittance and twiss parameters is defined at the midpoint of the inflector.
The distribution is propagated to the end of the inflector and then back to the opening of the hole in the iron.
Note that the taylor maps are not precisely symplectic so the inverse is approximate.
If the number of muons is one
then the trajectory through the backleg and fringe and inflector is written to unit 71
and the twiss parameters are computed assuming the initial values in the lattice file and written to unit 72.
The beam parameters computed from the distribution after propagating through injection line to end of inflector is
written to EndInfAfterScatter.dat
August 11, 2014
- The subroutine optimize_twiss finds the twiss parameters at the start of the injection line (30 cm upstream of iron) that give the specified (in the input file) twiss parameters at the end of element INFA in the inflector.
August 12, 2014
- The routine get_branch_inverse_matrix inverts the transfer matrix through the injection line. It had been using the full line, from 30cm upstream of iron to end of inflector. Set the endpoint in this routine to be at element 'INFA'.
August 14, 2014
Modify tracking_master to write phase space for each element in the injection line. Files are named ele%name//"phase_space.dat"
Plot using plotting_scripts/hist_any_file.gnu for example like so
gnuplot -e "filename='CRYO_USphase_space.dat'" ../plotting_scripts/hist_any_file.gnu
Add ability to use matrices for propagation through injection channel. The flag in the input file is inj_matrix_tracking.
Definitely more particles make it through. Remember that the map for the fringe field cuts off at 6mm, and 25mm.
The code assumes that beyond the boundaries of the map the field is uniform. But of course this gets the focusing effect wrong.
Of course the transfer matrices won't be right for the 36mm aperture inflector since the reference trajectory will go beyond the 6mm boundary.
August 15, 2014
The field map for the fringe field cuts off at +6mm in x with respect to the center line
and +-25mm in y.
Beyond the map we assume B has the value at the boundary. This means that for the
nominal 9mm inflector that which the reference orbit is all within the bounds of the map,
that will not be true for all of the particles in the distribution.
1. Perhaps a reasonable approximation
is to use the matrices computed with respect to the reference to propagate the particle distribution.
2. But for the 18mm inflector even the reference orbit is outside the bounds of the map. Are the
matrices associated with this reference valid? If not then using matricies to propagate phase space in this
case is not sensible. Let's check it out. Compare propagation of beta for 18mm inflector and 9mm.
The answer to this question is that the propagation of the twiss functions is very different using
these approximations for the field map. It therefore is not sensible to use the related matrices to propagate phase space.
In order to calculate with the larger inflector we need a better map.
Try to figure out how to read HeHe's map.
Meanwhile, update repository. Check out on linux. Repeat calcs in injection_line/ and they look OK.
Next step to update g-2
Sanity check.
With a single muon, fort.71 is trajectory through injection channel, and fort.72 is the beta.
August 20, 2014
Add comments to "create_phase_space"
Check that random distribution of muons gives capture efficiency consistent with VD...
Note that the emittances in the input file are for 1 sigma and the 95% is 2 sigma (or 4 times the emittance)
August 21, 2014
Add lattice name to input.dat
Add comments to subroutine collect_times
2014_0904
NSF: Adding Wuzheng's larger B-field map for the ring/backleg/hole. The first 3 lines of the file describe
the number of gridpoints in the {xyz}-directions, the number of field components, and the arrangement of the
data in the rest of the file. The spatial extent of the map can be read from the first/last data lines after
the header:
711 cm < x < 731 cm
0 cm < y < 3 cm
-430 cm < z < 10 cm
The B-field unit is again Gauss, as it was with the smaller map. Wuzheng did not include Hy(Oe) in the
larger map, which helps reduce file size, but makes it more difficult to tell when muons are in the backleg
steel (Hy<By).
September 6, 2014
Have rearranged reading field maps for fringe and inflector to generalize for other maps, like Wuzeng's larger map. In particular the 2D map that goes all the way
back to the iron. It just occurs to me that I will have to read three different maps to get the hole, fringe from 2D and inflector.
Meanwhile, there is some error in the mac version that effects when the kicker turns off. Tracking through inflector looks OK
but no turns. linux version works fine.
Note that the trajectory through the fringe, etc. every few cm in fort.71 will be the same as element by element in single_particle_by_element.dat
or track_backleg_inflector.dat as long as SCATTERING IS TURNED OFF.
check out Wuzeng larger map from repository.
September 7, 2014
FIeld names with maps of the magnetic fields in the fringe and inflector are now read from the input.dat file.
There are 2 fringe maps. Both generated by Wuzeng. The larger map has more extensive boundaries. The regions
of overlap have the same fields. Everything works on linux but there is a funy bug on the mac where
the kicker peak time is screwed up and the kicker is on during the second turn so everything exits.
Also notice looking at time distributions that the zero time is not the time when the particles enter the ring.
This should be fixed. Maybe by using a marker at the end of the injection line with custom_tracking, and custom_tracking sets time to zero.
The kicker timing error seems to have fixed itself. ?
October 7, 2014
Discovered that I had dropped one of the numbers in the twiss line of the input.dat file.
The result was that etayp was set to 0.7 (should have been the phase).
Rerun standard old-inf and find 4.6% capture.
Start to rerun various optimizations
Also change lattice file for use with wide aperture inflector.
There had been 5cm horizontal aperture on kicker 1. Change to 10cm.
Some small effect.
November 3, 2014
- Trying to understand
1. Nature of residual CBO for stored beam
2. Nature of residual beam size modulation of stored beam
- Experiment 1- initial distribution has 50mm-mrad 95% emit, but no energy spread - modulation is very small
- Experiment 2- initial distribution has 1% energy spread but 0.5mm-mrad emittanc - modulation is large
- Tune dispersion but best is ~ 8mm. Tune eta', and no help
- Perhaps the problem is the dispersive kick from the kickers? To test this hypothesis, try to reduce kick but changing 7.7cm
displacement of branch 0 to branch 1 to 1cm. Sure enough, the modulation (with beam with large espread) is reduced to 1mm.
test: Change dispersion in inflector from 8 to 2 and adjust kics to reduce CBO. Now the modulation is big.
test: Set dispersion back to 8. Sure enough modulation is now very small. This makes good sense.
test: Set offset back to 7.7cm and kicks back to nominal for wide inflector - And yes, back to 10mm.
Possibility that beam is running into edge of quad or kicker upstream and that is why the smaller kick works better.
Rather than because dispersion is generated by kick. But then why does large emittance no energy spread work?
__________________________________________________________
June 14, 2015
Preparing for CD 2/3 review
Plot sigma_x through injection line and into ring. Find that if betax=1.5 and alpax=0 at end
of inflector that entrance to inflector is limiting aperture. Use $ps/twiss_mismatch.gnu to plot.
(Input file, ring_twiss = 'open'). Add an extra flag to input file:
<use_lattice_twiss>: logical - if true use twiss parameters defined in lattice file to propagate through injection line.
so that it is possible to use twiss set in lattice file. This does not converge so well. Return
to varying betax and alphax at end of inflector through input file and find that if
twiss = 1.7642, 10.0, -0.7502, 0, 2.0, 0., 0., 0., 0.7, 0.7, 0., 0.,
beam just fits through inflector with max of 44.6mm in ring. f=45/7.7 =>95%
emit=34 mm-mrad - Note this is independent of the distribution. (f defined in plotting script)
Now it would be nice to see if these better optimized values give more muons.
Create new directory BetaTest
Copy input file and lattice file from ../KickTimeTest/127cm-20153028
Set optimized beta in input file - 348 stored ? seems low
----------------------------
I would like a fresh calculation of capture with old kicker vs new (including realistic pulse)
-create /KickTimeTest/821kick-20150614. Copy input from ../../BetaTest. Restore nominal twiss.
and run as a check - ge 398. Looks reasonable and the 348 with the new betas really was low ???
- Meanwhile, I found that KickTimeTest/127cm_table_20150507/Kick1-3Scan/
has the appropriate run with the full pulse
- Back to /KickTimeTest/821kick-20150614. Set input for 821 kicker circuit
- We get 217 stored with kickers at kicker_params%kicker_field = 210.0e-4 210.0e-4 210.0e-4
- Now do scan
**************
Comment out line 175 in fields.f90 that writes to unit 344. This information gives kicker pulse but generates a very large file
**********
------------------------------------------------------
October 12, 2015
code/fields.f90
Edit calculation of quadrupole efield so that
- between quad plates (0.05) and chamber wall (0.1) the efield is taken to be plate_voltage/(distance to wall)
- beyond the vacuum chamber wall, quad field is set to zero. (This is important for electron tracking)
- programs/efield_test.f90 calls fields.f90 and can compute fields.
___________________________________________
December 23, 2015
Add bmad style electrostatic multipoles to lattice file bmad_es_multipoles.
Turn on/off bmad multipoles (vs straight expansion in em_field_custom) by setting field_calc = custom (or not) in
definition of FULLRING.
Update set_quad_params to do the right thing with bmad multipoles
The good thing about the bmad style is that the entrance and exit fringe effect is included
The bad thing is that there is no easy way to deal with muons (or positrons) near or outside the plates)
But maybe the correction to the fringe fixed the peculiar energy dependence of omega_a?
Try that out
___________________________________________________________
September 30, 2017
Quads can be represented by
- grid map - based on 2D map
- grid map - slice from center of 3 D map
- bmad standard multipoles - cartesian
- bmad standard multipoles - cylindrical
Cylindrical multipole coefficients computed with programs/fit_mcmillan - fit to slice from center of 3D curviliner map
- track1_preprocess
quad ramping if em_field comes from cartesion grid map - do not have multipole or cylindrical version for ramp
kill particles that are outside vacuum chamber or headed in direction to be lost.
remebering that particles enter ring from outside of Q1-long
- em_field_calc
quad with grid map field calc are designated custom in bmad file
If vec(1) is within boundaries of map, set field_calc to standared and call em_field_calc
If vec(1) is outside boundary set E=0, B=1.45
Then add error fields
- compute_decayTime
After fussing with exponential distribution and gnuplot histogramming realize everything is fine. The problem was that
particle type was set in two places in the lattice file and I had set one to antimuon but the other remained as proton
________________________________________________
October 8, 2017
Add an option to create_phase_space.f90 to 'use_as_is'
If twiss_ref = 'use_as_is' then use the phase space distribution as read from a file or create with no matching to any particular
twiss parameters. The twiss parameters will be that of the distribution.
This is added so that I could read the distribution I wrote at the exit of the inflector and propagate around the ring.
This distribution was created by starting with Dikty's distribution at the end of M5 and propagating through inflector
and then onto design trajectory
_______________________________________________-
October 22, 2017
Reinstate RF quad. Add a couple of lines to track1_preprocess.
This is wrong. The quad must be set in em_field_custom so that the time is updated with each step.
Modify set_quad_params so that each quad can be called individually to set field
And use new routine set_a_quad(ele, verbose) to translate from quad_params%short_quad_field_index
Hardwired for Q2 short only
Also add write to unit 197 in em_field_custom.f90 so that it is possible to
check to see that it is happening properly
Note that if rf_quad frequency is horizontal tune, the output is a bit confusing since only a fraction is sampled
Old way: The frequency is 2 *(2pi* rf_quad()%freq_h - lat%a%tune) that is a small change in the tune and then doubled for beta modulation
If rf_quad()%freq = 0 then rf_quad%freq > 2Q
New way: Frequency is lat%a%tune * rf_quad()%freq_h that is a multiplicative factor times tune.
If rf_quad()%freq = 1 then rf_quad%freq = tune
Also, use twopi - lat%a%tune for horizontal
__________________________________________________________
November 18, 2017
Shape of measured kicker pulse in ring at fermilab by Alex.
Data file with time, volts, bfield: /field_maps/kicker1data.txt - coded in input.dat as 20171113
A modified version with reduced secondary pulse amplitude: /field_maps/kicker1data_halftail.txt - coded in input.dat as 20171118
/code/fields.f90 modified to read files
/programs/modify_kicker_pulse_map.f90
__________________________________________________________
March 16, 2018 and before
Modify create_phase_space to simplify creation of temporal distribution.
Create a separate routine <generate_fime_long_distr>, as a part of the create_phase_space.f90 file to generate temporal distribution and vec(5)
and rearrange create_phase_space so this same routine is called as required
Distinct pulse shapes for the 8 injected pulses based on measurements are in files/pulse_shapes/
Subroutine <muon_pulse_shape> reads one of the eight files <pulseshape_n.dat>, or the average which is the ninth <pulseshape_ave.dat>
and creates random distribution with the right shape.
Modifiy create_phase_space to read Diktys 400k trajectory file 'particles_M4M5End_400_mod.txt'
____________________________________________________________
March 18, 2018
Create a kicker pulse with width modified for 40ns increase of two additional blumlein sections. The file is /field_maps/kicker1data_flattop.dat.
______________________________________________________________
March 19, 2018
Create a new routine to compute energy loss in fiber harps. Use the energy loss distribution generated by gm2ringsim (provided by Antoine).
fiber_energy_loss.f90 generates a random energy loss consistent with gm2ringsim distribution. For the moment it does not account for
the fact the that effective thickness depends on the particular trajectory. If the particle hits the fiber then it applies the
same energy loss regardless. The file files/fiber_energy_loss/eloss.dat is the 'measured' distribution.
![[ICO]](/icons/blank.gif){kind=icon}
![[PARENTDIR]](/icons/back.gif){kind=icon}
![[ ]](/icons/tex.gif){kind=icon}
![[ ]](/icons/layout.gif){kind=icon}
![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/folder.gif){kind=icon}
![[ ]](/icons/unknown.gif){kind=icon}