Choose timezone
Your profile timezone:
MAD-X Solenoid, radiation, XSuite developments
→
Europe/Zurich
MAD-X PTC and MAD-NG Normal Form matters
Monday 23/5/2022, 10:00. Dedicated meeting on MAD-X and XSuite developments
Riccardo started the zoom meeting by welcoming all (> 20) participants and encouraging active discussions.
The meeting had two presentations. Please see these and the links provided in the presentations for further details. Only a short summary concentrating on some main subjects described and discussed is given here.
MAD-X status: May 2022. Riccardo De Maria for the MAD-X team, slides
------------------------------------------------------------
The MAD-X version scheme has three numbers : the first is for major releases, the second for minor changes and the third for bugfixes. At the time of this presentation the lastest release is 5.08.01.
Of 37 known issues, 19 can be considered as bug, 8 on documentation and 10 as enhancement. The actual status at any time is visible on github. Issues are closed when fixed by a pull request. Riccardo proposes to also close labels when invalid, duplicate or if the issue cannot be addressed in the foreseeable future.
Riccardo shows a list with links of major activities on MAD-X this year. He shows 5 activities that are on the to-do list for the next bugfix release this year.
The request to allow for the arctan2 function that needs two arguments in MAD-X would imply a major change of the parser. The parser is not constructed from a grammar. Rewriting the parser is not considered a major priority. When more flexibility is needed, users are encouraged to consider to go to cpymad.
Several recent upgrade requests and MAD-X improvements like tapering were motivated by electron machines, and in particular FCC-ee and ELETTRA. Further efforts are required and planned (KTAP, EMIT).
Riccardo recalls the definitions of MAD-X variables. Fields are scaled by P0 and physical coordinates by P0*(1+Deltas). Deltas is used to generate dispersive orbits. This also effects the T variable. MAD-X uses canonical variables and symplectic transformations based on expanded Hamiltonians. With radiation, the energy is lost in the direction of the momentum of the particle, which differs from the canonical PX, PY in the presence of non-zero transverse vector potential components (Ax, Ay). More work is also planned on documentation. Laurent will help to provide information on currently undocumented fringe fields.
In the following discussion it was recommend to generally use thick elements in the design phase and to go to thin elements only when needed (tracking). When slicing is done to split long elements in shorter pieces and insert markers, Helmut recommends to consider to use thick slicing (for bends, quads, solenoids).
XSuite development status, Giovanni Iadarola, slides
----------------------------------------------------
In addition to the main developers Giovanni + Riccardo, XSuite profited much by contributions and growing support from a very active and enthusiastic user community. The project was initially launched to rationalize and modernize software for multi-particle simulations, to move from a heterogenous range of programs each with limited capabilities to an integrated modular toolkit. The code is provided in the form of a set of Python packages based on timing critical code written in C. XSuite provides support for modern computing platforms (e.g. GPUs) and simplifies development and maintenance removing duplications. Giovanni shows a table providing an overview of features of MAD-X, Sixtrack, Sixtracklib, PyHEADTAIL and COMBI with their status of porting into XSuite. He illustrates many applications where XSuite has already been used successfully, often in combination with other "pythonic" tools and user provide packages. These include : single particle tracking, HL-LHC dynamic aperture and generation of particle distributions.
The Xdeps package is used to import deferred expressions form MAD-X into XSuite. Knobs can be imported and changed allowing for parameter scans in simulations. Synchrotron radiation emission was introduced based on MAD-X and PLACET implementations. XSuite was also successfully tested for smaller, lower energy machines like the CERN PSB and ELENA.
The collimation team has been very active to adapt XSuite for special needs of collimation, including halo cleaning with hollow electron lenses.
XSuite is also used in collaboration with EPFL on beam-beam simulations. Beamstrahlung effects have been implemented. The photon spectrum agrees well with GUINEA-PIG. The Xsequence package is being developed in the framework of FCC-ee to facilitate lattice conversions and provides translations between lattice descriptions used by different programs. An effort is ongoing to provide "pythonic" tools to generate and manage large sets of jobs (PyLHCSubmitter, tree_maker, Xdyna).
Rogelio asks if XSuite can also track thick elements. Giovanni replies yes, this can be done in principle. Thick maps have not yet been implemented in XSuite and help to do so by volunteers would be most welcome.
Felix Soubelet asks to automatically circulate XSuite related release information.
Riccardo closes the meeting (at 12:10) encouraging further feedback, collaboration and says we should not hesitate to file issues.
There are minutes attached to this event.
Show them.
