Cometary Trails

[cometarytrails]

GSoC 2014: Julian - Update 6

Progress since last update:

Data handling

Add an (optional) filter that only saves particles during planetary encounters during the saving period. This is useful if the integrator is used only to analyze meteor showers or impacts on planets and the dust environment in the rest of the solar system is not of any interest.

Physics

The effects of general relativity were implemented with a formula that mimics these effects in a relatively efficient yet accurate way.

After a number of tests and corrections, a test using the perihelion precession of Mercury showed an excellent match of the changes that are expected. (See also: Tests of general relativity: Perihelion precession of Mercury and the plots below)

[cometarytrails]

GSoC 2014: Julian - Update 5

Progress since last update:

Integrator

Tests showed that the performance of the RK7(6) integrator was severely impated by a low quality of the interpolation of bodies in cases were particles came close to planets.

A 5th order interpolation algorithm was therefore implemented to replace the 2nd order interpolation that was done before for most purposes. It uses both the positions and the speed vectors of three states now to calculate the positions at the intermediate Runge-Kutta steps.

The step saving algorithm had to be rewritten to give more control over the amount by which the number of saved steps increases. A maximum was added to avoid the creation of unnessecary amounts of saved and returned data. It can now be easily configured in the .ini file.

Program structure

The interpolation algorithm added another two functions that can be chosen to get planetary positions, increasing the total to five. To avoid the limitations of function pointers, a separate function was therefore written to select between these functions depending on the requirements of the integrator and the interpolation algorithm.

A basic setup to read command line options was added. it is currently only used for testing purposes when the application is called from inside MATLAB.

Notes

Linux builds are now using static linking for all libraries to avoid problems with missing or old libraries on some systems.

Counting the integration steps is now done in the same way in both integration algorithms to avoid confusion and reduce the probability that bugs are introduced.

[aerospaceresearch-blog]

This is why we ask #RosettaAreWeThereYet; new input! Good Speed, @ESA_Rosetta. @IRS_RZBW @Uni_Stuttgart @esa. We did this video to explain what our CometTrails app does, how our users helped us in here and because the Rosetta probe reached its desitnation August 6.

If you want to get the CometTrails code @smmrmx @JulianHzg did during @gsoc check @github -> http://fb.me/1tIn9i1vn

And if you like the music played in our @ESA_Rosetta video http://youtu.be/FY0vjbBp4eg , it is Hope For Rebirth by @celestialaeon found on @Jamendo

[cometarytrails]

GSoC 2014: Julian - Update 4

Progress since last update:

Features

Visualizations showed that particles that were perturbed by planets did often not have sufficient numbers of saved steps to be able to assume a Kepler orbit in-between two saved states. An algorithm was therefore implemented to increase the number of saved states if a significant portion of the acceleration comes from an object other than the Sun. This feature required some fine tuning with the help of constellation as it can only be tested properly with a large number of particles.

A number of steps were taken to make the application run on XP platforms again. In the process, the build setup was updated to make building both 32 bit and 64 bit versions easier.

Notes

Warnings on gcc were resolved that were not resolved before because the (very minor) issues were accepted by the Visual Studio compiler.

The release type is now printed in the console together with the version number.

The implementation for ending at a precise point in time was fixed, it had lacked a variable update that was necessary to produce correct results in the last step.

Enabled gdb debugging symbols on Linux.

[cometarytrails]

Test of the lifetime of a dust particle under the influence of only the gravity of the sun and Poynting-Robertson drag.  The black line shows the properties of the particle over time without the influence of PR drag, the red line including this influence. The results match analytical predictions of lifetimes well.

[cometarytrails]

GSoC 2014: Julian - Update 3

Progress since last update:

Integrator:

The 6th/7th order Runge-Kutta method is running and meets speed expectations. A number of fixes and improvements had to be added compared to the last update.

Poynting-Robertson drag

Fixes were applied to the formula to correct errors.

It was combined with the calculation of the effect of solar radiation pressure for more consistency and to be able to work with the actual solar GM for other possible effect.

Solar wind drag

This effect behaves similarly as Poynting-Robertson drag, it was added to the formula as a multpilier.

Both effects can be switched on and off using CMake options.

New features/behaviour

Option to end the simulation precisely on a given point in time rather than at some point afterwards that depends on the current stepsize and randomness.

Never print NANs into result files (these do not occur in nominal situation but can occur in some test scenarios, MATLAB has problems reading those files correctly if they include NANs).

[cometarytrails]

GSoC 2014: Max - Update 2

Progress since last update:

ParticleAnalyser:

Spacecraft relative computation and particle filtering now fully functioning

Node computation now fully functioning (numerical implementation)

Added option for pointcache2 output file-format for visualisation purposes using Blender

[cometarytrails]

GSoC 2014: Julian - Update 2

Progress since the last update:

Integrators:

Restructure information exchange: Struct pointers are handed to the integrators and downwards from there so that there is little duplicate data and everything is called by reference. This also reduces the amoung of arguments for several functions significantly.

Poynting-Robertson drag

Implementation in calc_accel

Particle-constant part of PRD is calculated in a separate function to save time

Tests show a difference in results compared to papers (particle lifetime on eccentric orbit)

Configuration updates:

Separate options for different Integrators (precision)

Selection of algorithm

Other:

Add option to integrate with stationary, SSB-centered bodies rather than moving ones for fast testing and testing of the long-term development of orbit properties

[cometarytrails]

Rosetta approaching Kevin! Our GSOC students code the Comet Trails application, which simulates the durst streams generated by comets. And as XKCD fans, we slected 67P/Churyumov–Gerasimenko aka Kevin as our first target. So what will happen to ESA's Rosetta mission when passing the simulated dust stream? Let's find out! :)

[cometarytrails]

GSoC 2014: Julian - Update 1

Progress of the first weeks of coding:

Linux compatibility and build environment changes:

General linux compatibility of functions and code used

Switched configuration query library to an open-source alternative that works on Windows and Unix systems

Set up a CMake build environment that works on different platforms and doesn't lose any functionality in comparison to the manually created Visual Studio project

Project management:

Introduce git version control (together with other project members)

Move the project to github (together)

Find a suitable license (together)

Move some of the code to include files for better overview

Integrator environment update:

Speedup for the acceleration calculation

Integrator update:

Coded a basic, working implementation of a Runge-Kutta-Nystrom 7(6) Integrator with adaptive stepsize control

[max-sommer-deactivated20160406]

Work progress so far:

- added an option to the ParticleIntegrator app: “save_as_binary” If set, the app now produces one big binary file that contains all of a work unit’s particles instead of heaving one text file for each particle in the WU. All in all this, reduced data size by a factor of 6...