What is the LHC On-Line Modeling?

1. What is the LHC On-Line Modeling? • The LHC machine will be a very demanding accelerator with large nonlinearities to control. Particle loss in the LHC must be actively controlled to avoid damage to the machine. Therefore any relevant adjustment to the machine must be checked beforehand with a proper model of the LHC. • ABP is convinced that this machine will require a powerful on-line modeling tool that is tested and ready for use in the control room before the LHC will be commissioned. • Main design features: • A special version of MAD-X will be the principle engine of the LHC on-line model. • A complete and trustworthy non-linear model of the LHC will be composed with the databases of the measured magnet errors and beam-based measurements. • This tool is NOT conceived as a real-time control tool of the LHC. • Instead the goal is to speed-up the off-line analysis time such that it can be done within minutes directly in the control room. • Experimenters and operators should profit from a fully functional on-line tool! Reference: http:/cern.ch/fidel, L. Bottura & L. Deniau; Wise: P. Hagen, JP. Koutchouk, E. Todesco FiDel - 18 January 2007

2. The Team • Frank Schmidt - Overall Responsibility, MAD-X General Support • Werner Herr – SDDS MAD-X Version, Closed Orbit Correction • Ilya Agapov (fellow Feb. 2007) – Beam-Based Measurements, SDDS Interfaces, Application Development, Optimization • Rogelio Tomás – Beta-Beating • Akio Morita (associate from KEK) – Beta-Beating, Experience from KEKB On-Line Modeling • Jörg Wenninger – SPS Operation, Standardization of SDDS • LSA Team (Mike Lamont, Greg Kruk, Sandeep Bharade) – General SDDS Support • Alastair Bland – TECHNET Support FiDel - 18 January 2007

3. The Tasks • The SDDS (Self Described Data Set) is the interface between MAD-X (or any other tool) and the control system. • The overall and detailed goals of the LHC On-Line Modeling to be discussed early 2007 in collaboration with OP. In particular, which application will be needed, present plan: Closed – Orbit, Coupling, Driving Term, Corrector Powering and Knob Settings. • The CAMIF (Cern Accelerator Mad-X Input Files) CVS repository has been defined and is being filled for all machines and transfer lines relevant for the LHC. • It is also planned to use CAMIF for storing LHC On-Line Models ( to be discussed). • Data Long-Term Storage via CASTOR and Data Reduction (10Gbytes one night for CNGS mostly zeros!). FiDel - 18 January 2007

4. Special SDDS MAD-X Version • Werner Herr has prepared this special version of MAD-X. • The idea is to read and write SDDS files from MAD-X and to transform the back and forth between Twiss Tables. Only 1-dimensional SDDS arrays will be considered. • Input example: • select,flag=sdds,clear; • select,flag=sdds,pattern="^b.*"; • select,flag=sdds,pattern="^d.*"; • ! convert: read SDDS and write TFS files • sddsin,file="cngs-test.sdds",table="sdds_tfs"; • write, table="sdds_tfs",file="tfs1"; • Reference: W.Herr and F. Schmidt, “Using SDDS data sets with MAD-X”, AB-Note-2006-043 (ABP) FiDel - 18 January 2007

5. Next Steps • Consolidation of the SDDS tools. • Definition of Structure of the SDDS for various Accelerators and Transfer Lines ensuring Flexibility and as much Uniformity as possible. • Early 2007: Defining Goals of the LHC On-line Modeling in collaboration with OP. • Development of the Applications needed for the On-Line Modeling including the SDDS Interfaces. • Testing at existing Accelerators. • Deadline: Be ready for the LHC Commissioning Start-Up in 2007! FiDel - 18 January 2007

6. Requirements for FiDel • The On-line modeling should have a response time of minutes  We will need from FiDel a complete magnet error model at specific times in the cycle • Thys Risselada is in charge to do the interface between FiDel and MAD-X • We are in close collaboration with OP and our development plan is in step with them • One of our main goals is to complement the FiDel magnet error model with a fudge factor data base derived from beam-based measurements •  Feed-back to FiDel after a year or so FiDel - 18 January 2007