New optics correction techniques, progress with Normal Form and 2022 commissioning update
######
S. Fartoukh, LSS optics correction with orbit bumps
Motivation is to have a technique to correct IR optics locally accessible to all LHC EiCs, as orbit correction is. The idea is to use different orbits, similar as LOCO, adapted to the LHC features. Deploy closed orbit bumps of constant amplitude but varying phase of the orbit bump. Involved IR bumps are IP Separation by about 0.5mm and crossing angle by 50 urad. Between Q5L and Q5R there are 160 observables 14+2 variables and 28 BPM calibration errors. Corrections will be iterative. Simulations are performed adding errors in only one IR and iterating corrections starting with most ideal assumptions and adding imperfections as BPM calibrations and quadrupole longitudinal errors in successive simulations. So far magnet errors, beta* and BPM calibrations are well reconstructed to state-of-the-art level in about 5 iterations. Missing simulated imperfections are: global errors, coupling, BPM tilts and longitudinal shifts, orbit corrector calibrations, non-linear errors and b2 in D1/D2. Most important missing ingredients are coupling (with BPM tilt) and orbit correctors errors.
Felix asks about impact of non-linear errors. Stephane assumes that perturbations from non-linearities will be filtered by the fits. Information from non-linearities might be also extracted via dedicated fit functions (or looking at tunes, etc.).
A reasonable timeline is to aim at the end of the year for the following points:
a) To complete the algebraic structure of the tools to include coupling and COD errors
b) Run few simulations and organize a second meeting with OP showing the results and get them embarked for 2023.
###############################
Tobias, Optics stability at injection
Clear changes up to 4% in betas are observed from beam test data after staying 30h at injection or after a precycle. Beta-beating between Run 2 and the beam test also changed by more than 4%. Stephane added that IR7 optics was different between Run 2 and the beam test. There was yet another change in beta-beating between the beam test in 2021 and the first measurement in 2022. Again by about 4%. It is beam 2 that is the most affected by these drifts. Another increase in injection beta-beating was observed during 2022 3Qy MD that took beam 2 beta-beating near 15%. After this observation a dedicated measurement showed beta-beating back to close to 10%. An energy offset of 1e-4 produces negligible beta-beating of about 0.3%. The reasons for the drift can be: b3 decay plus H orbit (similar to coupling decay), Dipole b2 (assumed very small), Q2 b2 drift (Stephane mentioned 3 units, to be checked with Ezio and Matteo), etc.
Stephane asks about BPM settings. All measurements presented are with pilot bunches and pilot BPM settings.
Stephane extrapolates the coupling decay from b3 + offset to 1-2% beta-beating. So within the order or magnitude. b2 decay in quadrupoles is 3 units from his memory. Ideally a simulation can be done putting these various ingredients together which seem to point that 4% changes are reasonably expected.
###############################
Laurent, progress with the normal form in MADX-PTC and MAD-NG.
The normal form is calculated from the MAD-X tracking using Sussix and the data processing in MAD-NG. These data are compared to the MAD-X PTC results, as Frank has previously identified discrepancies between the MAD-X PTC and SODD versions. Frank supported the implementation of Sussix. Laurent confirmed that asking MAD-X to produce 3rd and 4th order RDTs simultaneously produces wrong Hamiltonian terms in PTC_NORMAL since version 5.02.07. This error was also propagated to the 4th order extraction alone in PTC_NORMAL and PTC_TWISS since version 5.03.04, based on the new PTC. On the other hand, the generating functions are correct (sign error in PTC_NORMAL) for all extraction schemes.
More extensive comparisons between tracking and normal form give discrepancies when the spectral lines overlap (including the tune). However, small discrepancies also appeared elsewhere. This problem should be followed up to see if it can be fully understood or improved.
The amplitude detuning outputs of SODD, PTC, and MADNG are in very good agreement!
Laurent proposes as a correction to MADX-PTC PTC_NORMAL to use the same method as in PTC_TWISS.
Both PTC_NORMAL and PTC_TWISS have a problem with Hamiltonian terms. An easy solution proposed by Laurent is to calculate the H terms from the F terms using the analytical formula.
Laurent presents an application of MADNG with a 1st and 2nd order chromaticity match for FCC-ee with the collaboration of Michael Hoffer. MAD-NG is much faster than MADX-PTC. Exploitation by Michael to come in the future by comparing different approaches.
Frank updated Sussix to include 5th order analysis following Laurent's request to compare a non-trivial odd order with the normal form. The modified Sussix code can be found at:
/afs/cern.ch/project/lhcnap/projects/sussix/src/sussix
and including OPENMP at:
/afs/cern.ch/project/lhcnap/projects/sussix/src/sussix_omp