LIME 1.7.2

The previous code allowed a situation in raytrace() in which the code could attempt to free the pointers chainOfCellIds and cellExitIntcpts twice in succession without an intervening malloc or set to NULL. This was fixed by NULLing these pointers explicitly for all non-zero status returns of followRayThroughCells().

LIME 1.7.1

This is a patch of a few bugs detected in 1.7, as follows:

• Subtraction of CMB after raytracing has been reinstated (fixes #206).
• Data race removed from raythrucells.c (fixes #203).
• Line feed (\n) removed from the end of the molecule name read from file (fixes #204).
• Some memory leaks have been plugged.

LIME 1.7

• Added facility to read and write the entire grid to FITS file at 4 stages during task execution. (#92)
• A complete new (fast, flexible) gridding algorithm was added, controlled by a new user parameter par.samplingAlgorithm. (#93)
• Many knots, traps and tangles in the code have been cleaned out. (#152)
• Errors in the solution of the RTE corrected. (#153)
• Line amplitude now calculated using an error function lookup. (#153)
• Added facility to use the same random seeds for each solution iteration, which allows convergence to be separated from Monte Carlo noise in the calculated level populations. (#187)
• Grid points labelled non-sink are now labelled as sink points if geometrically they in fact are such. (#188)
• Photon tracks now keep close to the starting direction rather than diffusing as before. (#190)
• Sampling of frequencies for spectral lines is now uniform-random across the specified range. (#191)
• Several minor fixes and additions. (#170, #171, #172, #174, #183, #184, #185, #194, #195, #196)

LIME 1.6.1

• Fix version number (#163)
• Fix a bug with the image-plane interpolation (#162)

LIME 1.6

• Added inclination/PA/azimuth to image parameters (#63, #161)
• Separate the user-settable input parameters from others (#77)
• Allowed LTE as an initial solution (#82)
• Added an optional user function to define the gridding (#103)
• Interpolate collision coefficients on-the-fly (#135)
• Allowed better user control of density/dust calculations (#128)
• Implemented a new optional ray-tracing algorithm (#145)
• Fixed an incorrect calculation of dust density for multiple collision partners (#58)
• Fixed issues with the line polarization (#68, #94)
• Fixed a convergence issue when the density is too low (#85)
• Fixed an issue when the only collision partner is not H2 (#86)
• Fixed a velocity inversion (#88, #98)
• Fixed an X-axis inversion in images (#97)
• Fixed issues with line blending and multiple species (#99, #114, #115, #129)
• Other minor bug fixes and improvements (#69, #73, #74, #104, #106, #110, #111, #113, #121, #122, #123, #126, #127, #130, #131, #136, #137, #138, #148, #157)

LIME 1.5

• Do not rely on environment variables (see #57)
• Parallelize the code (see #56)
• Add an option to turn off ncurses output (see #53)
• Add command line options (see #47)
• Fix the velocity for cubes with an odd number of channels (closes #44)
• Flip the velocity axis in output cubes (closes #38)
• Build the code on Travis CI (closes #34)
• Fix memory leaks and overflows (see #32)
• Speed up the code (see #29 and #46)
• Move the documentation to ReadTheDocs (see #28)