… turns.

Python3.9 compatibility

Merger for new SPS OTFB model

Updated the central frequency of the SPS 3- and 4-section cavities.

Python3.9 compatibility

Updated SPS OTFB model

Linux Python 3.9 compatibility

…TFB class.

…ng unittests for llrf.

…e RF current calculation for the llrf.

Index of particles alive is !=0, not ==0.
beam object does not have beam attribute itself.

…particles

Fixed 2 typos in beam.eliminate_lost_particles.

Arrays beam.dt and beam.dE can be updated if more than 0 particles are still alive.
Array beam.id must be updated after deleting particles because array length changes.

… with the coarse grid.

…particles

Fixed bug in beam.eliminate-lost-particles.

…duced voltage.

… the SPS OTFB model.

… This solves the problem that dt and dE shared the same random number stream. This caused completely different bunches if number of macroparticles was just slightly changed

Lw otfb

Removed and added a few comments.

Merge lw_otfb into master

…tion and updated llrf unittests related to the SPS OTFB.

# Conflicts:
#	blond/beam/distributions_multibunch.py
#	blond/llrf/cavity_feedback.py
#	unittests/llrf/test_cavity_feedback.py

…sian

bigaussian initializer: Introduced separate random generators for dt and dE.

Merge blond-admin master back into my master

Merge latest SPS OTFB model

Update boost installation instructions

* adds missing requirements

* adds multi-bunch monitor support

* deletes unneeded version

* minor bug fixes

* not running reduce_histo automatically

* monitors record min max

* updates monitor arguments, adds new plot scripts

* debugs mew monitor args

* new monitor configurations

* mods to support larger number of  particles

* cosmetic improvements

* cuda-shmem-hot-bins.ipynb

* sets up efficient compile mechanism

* updates installation instructions

* Fixes py36 compatibility issue

* updates latest fftw version

* programmatic way to figure out python interpreter

* removes timing directives

* prints pycuda location

* Code clean-up

* fixes typo

* improves compile

* updates gpu main scripts

* CGA class to cupy

* conversion to cupy

* MyGpuarray without subclassing

* stdKernel as function

* minor change

* replaced butils_wrap.stdKernel with cupy.std

* minor fixes to kernel calls

* minor change to dev_my_array

* kernels and fft changes

* scaling to irfft

* elw kernels and fft changes

* fft changes

* cupy.rfft without plan

* remove GPUarray inheritance

* minor change to find cupy path

* cupy_array fix

* cupy_array remake

* cupy_array with unified memory and cleanup

* commented out invalidate_cpu/gpu)

* removed enable_gpucache()

* added None object init handler

* CGA.__init__ changes

* minor changes

* cuda kernels clean-up

* cupy install instructions

* removed pycuda unused imports

* minor change

* ports impedance to new simplified gpu scheme

* ports beam to new simplified gpu scheme

* modifies code to be compatible with cupy

* adds cupy/ numpy functions in bmath dictionary

* ports profile to new simplified gpu scheme

* converts beam_feedback

* same api in physics wrap

* function reposition

* ports rf_params

* cleans up bmath

* updates access to rawmodule

* ports tracker

* gpu code clean-up, ready to test

* debugs example 01, runs correctly

* more debugging, examples 2,3 run correctly

* example 4 works, example 5 needs debugging

* prepares remaining examples to be tested

* debugs until example 7, continue with 8

* minor change

* minor changes to gpu_slice

* transfers to gpu imported objects recursively

* all examples produce correct results

* EX01 supports plotting

* cleaning up

* different cubin name, for multiple GPU archs

* corrects unittests

* minor nested loop optimizations

* bug fixes in cupy backend

* minor fixes

* minor fix

* changes for main testcases

* changes for LHC testcase

* minor fixes

* changes for LHC

* performance improvement in beam_feedback

* reverting commits

* minor LHC fix

* typo corrections

* minor changes to testcases

* cupy v11 release

* changes to rf_params calls

* up to date with latest dev in master

* minor change

* debugging the examples

* tested GPU examples, 3 need debugging

* debugged all example files

* wires inter_const_space to interp

* minor bug fixz

* up to date with latest changes

* fixes failing unittests in bmath

* fixes failing unittests in test_synch_rad

* fixes failing unittests in impedance_sources

* fixes warnings in beam

* fixes file not found errors

* fixes failing tests in tracker

* fixes failing examples

* minor improvements in example files

* check for gpu use

* adding unittests for the GPU parts

* fixes failing cupy fft testcases

* adds cp.convolve unittests

* adds unittests for beam_phase

* name mangling attribute fix

* adds unittests for more physics modules

* switch __device to _device

* finished testing all physics kernels

* allreduce compatible with cupy arrays

* adds unittests for impedance related classes

* tests ind volt track

* adds __init__.py to avoid name collisions

* uses pytest parametrize to reduce code length

* fixes warnings

* adds unittests for impedance with MTW

* workers check

* minor changes

* gpu single precision fix

* cleaning up unused code

* function name changes

* gpu drift minor fix

* reduce_histo cpu transfer

* readme changes

* nvcc from cuda path

* lik simplify

* error corrections

* Disables interactive matplotlib backend

* skips failing unittests, adds FIXME note

* changes np.int (deprecated) to int

* removes python3.6 from windows testing

---------

Co-authored-by: Giotyp <typaldosgeorge@gmail.com>
Co-authored-by: Giotyp <82325747+Giotyp@users.noreply.github.com>

1. Updated outdated syntax: Obtaining ring radius from beam -> Obtaining ring radius from ring object
2. Corrected small bug when trying to pick out the maximum frequency of each particle for calculating the synchrotron frequency. Previously it was possible to pull multiple frequencies if the maximum Fourier amplitude of the FFT repeated itself. Now it will always only pull one value from the frequency array.

Co-authored-by: onaumenk <oleksandr.naumenko@cern.ch>

Sets up gitlab CI/CD

See merge request blond/BLonD!1

Removes various C++ code warnings

See merge request blond/BLonD!2

Fix swap in {fwhm, rms}_multibunch return values

See merge request blond/BLonD!3

changes readme format to md

See merge request blond/BLonD!5

minor modifications to support python3.6

See merge request blond/BLonD!4

Automatic Compilation and Upload of BLonD documentation

See merge request blond/BLonD!6