What’s new in 1.4.1

Summary

General

  • All tofu object have overloaded repr method

  • units tests improved, Continuous Integration more complete, several bugs fixed

  • More robust wndows pip installation

  • More robust saving to .mat

  • tofuplot and tofucalc now provide simple bash interfaces

The geometry module is now more complete

  • A Config class handles the tokamak geometry

  • A CamLOS1D class handles arbitrary sets of LOS

  • A CamLOS2D class handles 2D cameras approximated as a set of LOS

  • Basic reflexions are taken into account

  • Ray-tracing, LOS-campling and LOS-integration algorithls have been accelerated and parallelized

  • Several utility function implemented for easy instance creation and useful geometrical computations

  • The basics of a magnetic field-line tracing algorithm and plotting routine have been implemented

IMAS-interfacing

  • More robust MultiIDSLoader class for IMAS interfacing

  • Can import from IMAS and export as tofu classes

  • Very flexible and customizable

  • Some classes have a save_to_imas() method

Next to come

  • Solid angles, solid angles, solid angles

Detailed Changelog

Enhancements

Geometry

Algorithms to compute distance between LOS and a flux surfaces

  • Compute distance between LOS and a circle (+ vectorial version)

  • Test if a LOS and a circle are close to a epsilon (+ vectorial version)

  • Compute distance between LOS and VPoly (+ vectorial version)

  • Test if a LOS and a VPoly are close to a epsilon (+ vectorial version)

  • Compute which LOS is closer to a VPoly

  • Compute which VPoly is closer to a LOS

Algorithms for discretization and sampling

  • Optimization of discretization of 1D lines (simple + complex version), 2D lines, 2D polygons

  • Optimization of LOS sampling and all the different rules: parallelized and using less memory.

Calc signal (function integration over LOS):

  • LOS_calc_signal now cythonized for isotropic and anisotropic emissivity and several sampling / integration strategies

Algorithms to check if visible

  • Optimization LOS_isVis_PtFromPts_VesStruct checks if a point is visible with respect to a list of points and taking into account a vessel (with structures, limits, etc.) : function cythonized properly, speed up ~50%

  • LOS_areVis_PtsFromPts_VesStruct equivalent to last function but input is a list of points

  • Dust_calc_SolidAngle is now using new versions of above mentioned functions

Vignetting

  • Added functions to compute the 3d bounding box of a polygon in 3d space

  • Added functions for the triangulation of a polygon by earclipping

  • Added function that tests if intersection between ray and triangle in space

  • Added functions for vignetting : for a set of rays and a sets of polygons, test if the rays go through the polygons or not.

Structures and Configuration

  • tf.geom.Struct._get_phithetaproj(): to get the (phi, theta) limits in a project of each toroidal occurence of Struct, with respect to a refpt in (R,Z)

  • tf.geom.Struct._get_phithetaproj_dist(): + get the distance

  • tf.geom.dist._get_phithetaproj_dist(): generalize to all Struct of a Config

  • Config has set_colors_random()

Solid angles

  • Added _GG.Dust_calc_SolidAngle for computation of a solid angle on dust particle (multi-pos, multi-pts, with and without vect, with and without approx, with and without block)

Basic Reflexions

  • Rays subclasses can now have added reflections (self.add_reflections())

  • Reflections can alos be extracted (instead of stored) as a list of separated cameras using Rays.get_reflections_as_cam()

  • Reflections can be of type ‘specular’, ‘diffusive’ or ‘ccube’

  • The reflection type can be forced by the user (identical for all LOS of a camera) or automatically set by the Structural elements each LOS touches

  • There can be any number of reflections

  • Rays.plot() and Rays.plot_touch() have been updated to properly represent reflections

  • Rays.calc_signal() and Rays.calc_signal_from_Plasma2D() have been updated to account for reflections with a user-provided coefficient

Cythonization

  • Re wrote some numpy function to avoid over heading: hypothenus and composition of minimum of hypotenus, minimum of vector, tile,

Extra

  • Erased old functions:

    • SLOW_LOS_Calc_PInOut_VesStruct

    • Calc_LOS_PInOut_Lin

    • Calc_LOS_PInOut_Tor

  • extract_svd() now compatible with Py27 and Py36

  • Rays.dgeom[‘indout’][0,:] now refers to the global index of each structure (as in config.lStruct)

  • Operator overloading of ____repr____ of all tofu objects with attribute get_summary()

Cameras and visualization

  • overloaded add operator to allow for easy CamLOS1D concatenation

  • tf.geom.dist.plot_phithetaproj_dist(): plot a static figure with a place holder for time traces, the (phi,theta) projection (with aspect ratio), the cross and hor Config projections and legend

  • Rays.plot_touch() now handles: angles (wrote docstring)

General

  • get_summary() is now a method of the AbstractToFuObject parent class, implemented for Config, CamLOS1D, Plasma2D

  • saving to .mat file is now more robust (operational)

  • Added benchmarking for calc_signal

IMAS interfacing

  1. libraries that need imas2tofu are only imported if the IMAS module is found (Warning not Error)

  2. MultiIDSLoader class now handles all imas interfacing:

  • allows for loading several ids from different idd

  • dynamic printing of ids loading

  • get_summary() methods giving an overview of content

  • provides a dictionary of shortcuts for fast typing of pre-recorded imas signals

  • provides dictionary of preset ids and signals

  • Flexible instanciation, from idd of idd args, ids or ids args, open or not, get or not

  • provides add_idd(), remove_idd(), add_ids(), remove_ids()

  • Include to_Config(), to_Plasma2D(), to_Diag() to export to tofu objects

  1. **All stored in a single _core.py**:

  • MultiIDSLoader

  • load_Config(), load_Plasma2D(), load_Diag()

  1. save_to_imas()

  • Actual routines implemented in tf.imas2tofu._core

  • methods implemented for Struct, Config, CamLOS1D

  1. Other

  • MultiIDSLoader now has calc_signal()

  • ids = ‘magfieldlines’ returns mag field line tracking from IMAS equilibrium

  • More robust / detailed sanity checks of triangular meshes from IMAS

  • Synthetic data and experimental data are now time-synchronized

  • Synthetic diag operational on WEST for interferometer, polarimeter, bremsstrahlung and bolometers

bash interface:

  • tofuplot debugged

  • tofucalc created

Code structure and format

Restructuring of the geometry modules

Created cython include files *.pxd and *.pyx for the following group of functions: - _basic_geom_tools.*: global variables definition (_VSMALL and _SMALL), and basic geometric tools (vector calculus, path distance point-point, point-vector,…) - _raytracing_tools.*: for intersection of LOS and different objects (bounding boxes, poly, surfaces, …) - _distance_tools.*: distance between LOS and circle, VPoly and so on - _sampling_tools.*: line, LOS, poly, volume, and surface sampling

Installation and portability

  • Had to use some cpp tools, so had to update the setup.py

  • Updated the setup.py to be able to “clean” an installation (erase compilation files)

  • Took out the possibility of not using cython, as this is now impossible

  • git dependency is now optional (issue #67 )

  • changes in setup.py for Windows portability

  • Now only supporting cython versions >=0.26

  • Removed all pandas dependencies

  • Using -O3 flag instead of -O0 for faster execution time even if compilation is slower

  • Removed all Polygon dependencies

Unit test

  • Added unit tests for triangulations, and vignetting

  • Added unit test for computation of kmin, kmax LOS_Calc_kMinkMax_VesStruct

  • Update of in _vessel tests

  • Added tests for is_visble (vectorized and point wise)

  • Now by default python setup.py nosetests with run with: verbose, detailed erorrs, coverage (nose) and other utilities for debugging. But most importantly only tests in ``tofu/tofu/tests/`` will be run.

  • Testing all get_sample options in a short new unit test

  • Testing all options for LOS_calc_signal (method of discretization, of integration, steps relative, absolute, unique or changing for each los, etc.).

Update of documentation

  • Added documentation on how to install on Windows

  • Updated, restructured and adde figures to README. Change of format *.rst to *.md

  • Updated Wiki pages on GitHub

Bugs

  • Found several small bugs in the function that computes for a list of flux surfaces and a list of rays the kmin, kmax

  • There was a bug in some special cases of 1D camera definition, when all Ds are on a similar plane sharing the same D for the 2 first LOS, solved

  • There was a bug in tf.data.plotcombine() (wrong graph), solved (issue #65 )

  • There was another bug with plot_combine() when several equilibria were provided, the reference time was not properly defined (short-term fix, on the long term issue #79 should fix it)