Proposal for tests and changes in the MADX code

coldfix commented 12 years ago

Issue migrated from trac ticket # 172

component: core_sequence | priority: major

2012-07-18 13:10:51: stephane.fartoukh commented

date: 2008.12 Please find below a complete symplecticity and consistency check for single element a) defined by a 6D dX offset , R and T matrix b) defined with a linear or non-linear transfert map M: X0 à X=M(X0)

The things to be checked are

a) consistency dX=M(X0=0) R_ij = (dM_i/dXj)(X0=0) T_ijk =1/2 (d2M_i/dX_j/dXk)(X0=0)

b) Symplecticity transpose(R) J R =J transpose (Tk) J R + transpose (R) J Tk =0 for k=1..6

where Tk is the 6*6 matrix defined by (Tk)_ij= Tijk and J is the usual 2 by 2 block diagonal symplectic matrix with the blocks [[0,1],[-1,0]] on the diagonal.

From laurent.deniau I added the attached document gently provided by Stephane for history and backup. The tests above where used to test the RF-multipoles of Andrea. Other points like sequence reflector will be studied in some future.

2012-07-18 13:12:15: @ldeniau uploaded file `beam2-beam4.doc` (76.0 KiB)

2015-09-11 10:49:55: @ldeniau commented

This improvement requires careful check of the code in twiss. This could be done step by step as the elements are reviewed and cleaned (or moved to the new MAD). LD.

2015-09-11 10:49:55: @ldeniau set owner to alatina

2015-09-11 10:49:55: @ldeniau changed status from new to assigned

ldeniau commented 7 years ago

Optimised routines exist in MAD-NG. We may import them in MAD-X such that with proper flags, the checks could be done after each element and before concatenation of the maps.

rdemaria commented 2 years ago

Good ideas, but I am closing now as it is a project in itself. Happy to reopen when somebody finds the time.

MethodicalAcceleratorDesign / MAD-X