Geometry Hierarchy Queuing LHC Model L Nevay Royal

Geometry Hierarchy, Queuing, LHC Model L. Nevay Royal Holloway, University of London 17 th July 2014

Current Hierarchy • Only element geometry constructed • Tunnel doesn’t work • Cubes / Trapezoids used for marker volumes Marker / Sensitive Volume Tunnel (Doesn’t work) General Geometry Beam Pipe Vacuum 2

Proposed Hierarchy • Vacuum, Beam Pipe & Section Volume minimum • General & Tunnel optional • G 4 Cut. Tubs used for Section & Tunnel volumes Section Volume (G 4 Cut. Tubs, solid, vacuum) Tunnel (G 4 Cut. Tubs, tube, material) General Geometry Beam Pipe Vacuum 3

Differences Currently only make ‘sectional’ volume if it’s a marker should always do – more efficient geometry navigation can then simply paste on sensitive detector(s) can be difficult to create geometry that matches at each plane • not a problem with G 4 Cut. Tubs • Simplifies Detector. Construction() • Tunnel geometry will work! • • • Use 3 Vector of reference line at start & finish of each element • Makes grouping several elements (say 20) together easy ― improves geometry performance for very large lattices 4

Plan View • Each beam line element has it’s own outer section volume tunnel beam pipe vacuum magnet section volume 5

Different Queues TRACKER ONLY TRACKER + COLL BDSIM-RING BDSIM while ( particle radius < beam pipe radius ){ tracker } while ( particle radius < beam pipe radius ){ tracker + G 4 collimators } normal BDSIM run record digital losses shift to BDSIM record digital losses • Condition to move to different queue may be based on rigidity as well as position o for BDSIM-RING only • Simple (low CPU) test to account for motion in field • Shift to BDSIM world – exactly the same tracking, but overall slower 6

Different Queues TRACKER ONLY for comparison of tracking routines TRACKER + COLL more accurate due to scattering a la Six. Track 7 BDSIM-RING BDSIM the same, but slower

Constructing LHC Model • Previous model ends meet within ~100 um • This is the normal accuracy of the MADX model ― Sixtrack input is prepared from this – why don’t they see it? • The model before excluded: ― ― ― multipole kicks – by virtue of ‘thick’ lattice multipoles rf cavities sextupoles collimators • Were replaced with drifts • Adding these in produces offset of 14 m in both x & y 8