## Feature #1887

### 2016/04 Run Settings

Status:
Closed
Priority:
Normal
Assignee:
-
Start date:
07 November 2016
Due date:
% Done:

100%

Estimated time:

Description

Run settings for 2016/04

Files

#### Updated by Rogers, Chrisalmost 6 years ago

Some supporting documentation

#### Updated by Rogers, Chrisalmost 6 years ago

Version 1 of the 2016/04 set.

I included three run sets, "low force", "intermediate force", "high force", grouped according to the force limit in the calculation (as shown in column S). I only expect one to be chosen, subject to discussion. I am sure this will not be the final scheme!

I can upload some plots/etc in the morning.

#### Updated by Rogers, Chrisalmost 6 years ago

Nb: I had trim supplies on for these settings - probably in the end we won't want this. It probably doesn't affect the optics much.

#### Updated by Rogers, Chrisalmost 6 years ago

Also see the slides/notes here:

2016-11-07-settings

#### Updated by Rogers, Chrisalmost 6 years ago

Discussion with MIPO - request for two new sets both in flip mode; no trims; 3 T:

1. Force between FC and SSU with all coils powered of 15 T excluding PRY (in addition to min/max on M1 and M2 etc)
2. No force limit between FC and SSU - but continue with min/max on M1 and M2 etc

#### Updated by Liu, Aoalmost 6 years ago

Quote from C. Rogers:
15 T is roughly
M1*FC < 18500 in solenoid mode
M1*FC < 28000 in flip mode

The muon beams at TOF1 are attached. Comments or questions welcome.

#### Updated by Rogers, Chrisalmost 6 years ago

Thanks Ao.

Nb: we decided that lattice figures of merit should be

• Emittance reduction of good muons just across the absorber
• Transmission
• Emittance change between TKU and TKD
• Force between SSU and FC

where emittance reduction etc is assessed against the beam files provided by Ao. There is some discussion over which has priority...

#### Updated by Rogers, Chrisalmost 6 years ago

Also, we agreed that we should study two different samples:

• Particle pz falls within peak pz +/- 5 MeV/c cut
• Particle pz falls within peak pz +/- 10 MeV/c cut

#### Updated by Rogers, Chrisalmost 6 years ago

I attach a few files:

• Beam optics quantities from recent running for each sample provided by Ao; two files are provided, "data_summary.csv" is in simple columns format; data_summary.json is in json format. For each file I provide three sets of quantities; optics for "all events", optics for "events in +/- 10 MeV/c cut", optics for "events in +/- 5 MeV/c cut" of the peak.
• plots.tar.gz has plots of pz, x, y, px, py, r for each file; Black is "all events", Blue is "events in +/- 10 MeV/c cut", Green is "events in +/- 5 MeV/c cut.

#### Updated by Rogers, Chrisalmost 6 years ago

Ao, what is the z position you use for TOF1, relative to the absorber say? In your backwards extrapolation, do you assume any energy loss (or other materials processes); or do you only assume effects of the magnetic fields?

#### Updated by Liu, Aoalmost 6 years ago

TOF1 was at -4023 mm rel. to abs. center;

I assumed scattering and E loss. World material was Air.

#### Updated by Liu, Aoover 5 years ago

Preliminary talk slides.

#### Updated by Rogers, Chrisover 5 years ago

Lattice file and tracking output for Jaroslaw's lattice

#### Updated by Rogers, Chrisover 5 years ago

Attached - summary table of performance of various different lattices, along with performance plots (e.g. emittance vs z, etc). Based on tracking through a simple lattice, using MAUS.

#### Updated by Liu, Aoover 5 years ago

• File deleted (Nov14_2016_runSettings.pptx)

#### Updated by Rogers, Chrisover 5 years ago

As per point 16, but with 10x statistics (50k events per simulation run).

#### Updated by Rogers, Chrisover 5 years ago

• transmission is the transmission between z=-2900 and 2900 (fractional)
• delta_emittance_absorber is the emittance change (fractional) between -100 and +100 mm. This is for ALL muons, i.e. no "good muon cut" was applied. This means it is not biased by scraping, but it is not an observable quantity.
• nominal_emittance is the nominal input emittance; where this is -1 Ao's 140 MeV/c beam was used. Else a matched beam with the given nominal emittance and pz = 140 MeV/c was used.
• delta_emittance_tracker is the emittance change from upstream tracker to downstream tracker, for good muons only.
• beta is optical beta function at the absorber.

Forces and currents are in there also.

#### Updated by Rogers, Chrisover 5 years ago

New lattice from Ao.

#### Updated by Rogers, Chrisover 5 years ago

Ao asked for simulation output for this last dataset. i uploaded it as

http://heplnv152.pp.rl.ac.uk/analysis/maus_output-v10.tar

Nb: the input beam is recorded as the MC event "primary" object.

#### Updated by Rogers, Chrisover 5 years ago

Another setting from Ao...

#### Updated by Rogers, Chrisover 5 years ago

Beta = 1000 mm setting from Jaroslaw

#### Updated by Rogers, Chrisover 5 years ago

Rates/etc for the best 140 MeV/c settings we have to date:

 Tag Runs lmc1234 sum TOF1 Triggers TOF2 Triggers time taken number in muon peak number at tof2 number passing all cuts number passing all cuts / p in tku [MeV/c^-1] number passing all cuts / tof01 [ns^-1] 3-140+M3-Test2 [8541, 8542, 8543] 1638216 291246 147415 2 hrs 42 mins 86947 69369 25704 2898.35 41730.0 6-140+M3-Test2 [8593, 8595] 1046237 359945 152032 2 hrs 58 mins 130235 69146 34050 3783.56 59420.0 10-140+M3-Test2 [8609, 8610] 128161 75665 11301 0 hrs 32 mins 37936 3819 3775 428.57 8380.0
• Tag is the beamline tag
• Runs is a list of runs considered to make following calculations (I didn't use all runs for a particular tag)
• TOF1 Triggers is the sum of the CDB TOF1 triggers scalar; note this is trigger requests (does not include dead time/etc)
• TOF2 Triggers is the sum of the CDB TOF2 triggers scalar; note this is trigger requests (does not include coincidence with TOF1/etc)
• lmc1234 sum is the sum of the CDB luminosity monitor scalar
• time taken is the sum of the times listed (time between "start run" and "stop run")
• number in muon peak is the number reporting tof01 between 28 and 33 ns
• number at tof2 is the number reporting at least one space point in tof1 and exactly one space point in tof2
• number passing all cuts is the number passing:-
• exactly one track in TKU
• tof01 between 28 and 33 ns
• TKU total momentum between 135 and 145 MeV/c
• Exactly one space point in TOF0 and TOF1
• number passing all cuts/p in tku:- I histogram p in TKU, then find the peak bin, then report number in the peak bin divide bin width
• number passing all cuts/tof01:- as per previous entry, but this time using tof01 histogram

#### Updated by Rogers, Chrisover 5 years ago

Plots to support previous entry

#### Updated by Rogers, Chrisover 5 years ago

We also need to do some off-momentum alignment runs. The point is to scan momentum and look at the offset of the beam for different momentum particles. From 2016-03, we have following settings/rates (see entry 24 for meaning of each heading):

 Tag Runs lmc1234 sum TOF1 Triggers TOF2 Triggers time taken number at tof2 3-170+M3-Test1 [8449] 936636 338776 105834 2 hrs 15 mins 29987 3-200+M3-Test1 [8463] 1223556 667060 103122 3 hrs 5 mins 44128 3-240+M3-Test1 [8451] 504169 418820 100491 1 hrs 10 mins 44616

#### Updated by Long, Kennethover 5 years ago

Spreadsheet to compare recent settings. Assimilation of v8, v11 and v12. Nothing new. Just that I found it easier to compare.

#### Updated by Rogers, Chrisover 5 years ago

I did some head scratching on the difference between Ao and my simulations. One thing I noted was that I assume a 3 T field for the input beam, even when sometimes it is a bit less. Also, I made summary plots of performance. See the plots tarball.

One feature that I noted was that the lattice 1.1 was sometimes running very slowly and timing out - not sure why that was. So I ended up running with reduced statistics (5k instead of 50k in the input). I had better dig into the slow running. Even like this, some jobs ran for 30 mins and others for 150 mins, for no apparent reason. Something weird is going on.

#### Updated by Liu, Aoover 5 years ago

Chris,

Thanks for scratching your head. Yes the initial beam will have a little bit of difference with different field, but I remember they were all close to 3 T so there must be something else adding up the discrepancy.

1. Would you also mind explaining what's new in your new performance log file? Also, please note that the log file had some std output mixed inside, please check.
2. Your fractional emittance plot - I presume that was based on matched beam wasn't it?
3. Lattice 1.1 accepts the mismatched beam better so when using a mismatched beam there might be more computing involved. Also it might be that a matched beam in 1.1 falls outside the "good volume".

#### Updated by Rogers, Chrisover 5 years ago

Would you also mind explaining what's new in your new performance log file? Also, please note that the log file had some
std output mixed inside, please check.

I think I just added some plotting. Sorry about the messed up std::out, I forgot to flush my buffers. I hope it is still readable.

Your fractional emittance plot - I presume that was based on matched beam wasn't it?

The disconnected points are for the "unmatched" beam. I assume that the higher emittance beams are well enough matched that there won't be much visible difference between "matched" and "unmatched".

Lattice 1.1 accepts the mismatched beam better so when using a mismatched beam there might be more computing involved.
Also it might be that a matched beam in 1.1 falls outside the "good volume".

Well, I don't think the transmission is different by 5x. This happens for "matched" beams. It feels more like there is showering or something. I have a hard limit of 10000 steps (10 mm max step size) for each track, so can't be one track getting stuck but could be many tracks getting stuck for some reason? I can't think of any regions where processing would be very slow because my geometry is very naive/simple. It is a mystery.

#### Updated by Rogers, Chrisover 5 years ago

Three new lattices from JP.

• 1 is for a 240 MeV/c beam;
• 2 and 3 are optimisations for the 140 MeV/c beam setting

For reference I also included the 1.2 setting on the plots.

#### Updated by Rogers, Chrisover 5 years ago

Jaroslaw was interested in the comparison with 1.5 lattice (that we plan to run on Monday). I upload the same v16 plots but this time comparing with 1.5 as a reference.

#### Updated by Liu, Aoover 5 years ago

Thanks Chris. Might be necessary to solve that mystery. In your update #33, you said lattice 1.5, but you actually pointed to 1.6 in your Excel table. I don't know which one you actually referred to but I think it is worth clarifying.

#### Updated by Rogers, Chrisover 5 years ago

I mudged around with my scripts a bit, approaching a "GA" type approach. I fire off loads of simulation runs with relatively low statistics and look at the resultant beta function at the AFC, emittance changes and transmission. I note under this model, the settings proposed by Ao are a bit off the optimum point... we can get same optical function at the absorber but with better transmission if we push FC up a bit.

• I start with a matched 6 mm emittance beam, 200 MeV/c with 5 MeV/c RMS, 1k muons
• I look at the transmission from upstream end of TKU to downstream end of TKD
• I look at the beta at z = 0
• I look at the change in emittance of good muons between tracker reference planes (fractional_)emittance_change_tracker)
• I look at the change in emittance of all muons across the absorber (fractional_emittance_change_absorber)
• I also include plots coloured by focus coil current, SSU force, match coil currents
• I put triangles indicating my simulated performance for Ao's three settings

#### Updated by Liu, Aoover 5 years ago

I'm confused because the FC changes the beta function by almost nothing, even sometimes pushing the beta function higher. I think that worths a check. I'll check the FC setting in my codes and do a comparison.
Also I'm confused that my setting for a "matched beam" was the worst..., and also I'm confused that, my 2.1 and 2.3 look very similar but behaved very differently. Let's check.

#### Updated by Liu, Aoover 5 years ago

OK I have tried my setting with higher FC current as suggested by Chris' analyses. I'm getting a bit worse transmission and less emittance reduction. So I think 56.2 is better than 80 A.

#### Updated by Rogers, Chrisover 5 years ago

I ran a few more options through the full MC. The option

run_plans/MICE_User_Run_2016_04-v4.csv 9
nominal momentum: 200.0 MeV/c
currents
EndCoil2_US: 205.86
CenterCoil_US: 205.86
EndCoil1_US: 205.86
MatchCoil2_US: 216.4
MatchCoil1_US: 243.85
FocusCoil_US: 73.88
FocusCoil_DS: 73.88
MatchCoil2_DS: 0.0
EndCoil1_DS: 205.86
CenterCoil_DS: 205.86
EndCoil2_DS: 205.86


has best performance but the force is illegal - 18.2 tonnes. I must have made a mistake (I checked on JHC's spreadsheet which is the reference I think). So next best is

run_plans/MICE_User_Run_2016_04-v4.csv 8
nominal momentum: 200.0 MeV/c
currents
EndCoil2_US: 206.3
CenterCoil_US: 206.3
EndCoil1_US: 206.3
MatchCoil2_US: 172.9
MatchCoil1_US: 239.2
FocusCoil_US: 56.2
FocusCoil_DS: 56.2
MatchCoil2_DS: 0.0
EndCoil1_DS: 205.8
CenterCoil_DS: 205.8
EndCoil2_DS: 205.8
forces
SSU: 13.36475618
FC: -12.1410265
SSD: -1.74390804


I would run with this one...

#### Updated by Liu, Aoover 5 years ago

Three things as a reminder:
1. (Just got off from the phone call with Chris)
208 A on ECE is fine;
2. Setting 2.2 was done with a mismatched beam, and a different requirement: I required the beam that is within Pz \in [195, 205] MeV/c to be cooled. As it turned out, it can't. But this setting should give the least heating for that slice of the beam;
3. It would be better if tracking with the realistic 200 MeV/c beam can be done for each 2.X setting. The beam can be found in the previous attachments.

BTW, Chris' setting 9 was indeed the best. What I saw was a better transmission with roughly the same emit. reduction.

#### Updated by Rogers, Chrisover 5 years ago

I found the bug in the tools I was using to create the cloud of performance plots (note 35). I had neglected to include FCD in my force calculation. I attach the revised plots and Ao's setting looks to be in a pretty good place.

Updated

#### Updated by Rogers, Chrisover 5 years ago

• File deleted (optimisation-17b.tar.gz)

#### Updated by Liu, Aoover 5 years ago

Jaroslaw's setting plots.

#### Updated by Rogers, Chrisover 5 years ago

• Status changed from Open to Closed
• % Done changed from 0 to 100

#### Updated by Rogers, Chrisover 5 years ago

Final version of the magnet current spreadsheet - for posterity

#### Updated by Liu, Aoover 5 years ago

Chris,The Setting number in the updated run plan table does not match the CC tag we used in the runs - it might cause some confusion. The 2.3 in your current table is the 1.7 we took, and the 2.4 is the 2.4a we took.
Just in case people could get confused (mostly myself?)
Would you like a separate table for the flip mode settings?

Also available in: Atom PDF