Wiki
Clone wikiC-Development / CommandLineInterface
You can run Fréchet-View in "headless" mode from a command line terminal. The program processes just one input file.
Select an algorithm:
-c
Fréchet distance for curves-p
Fréchet distance für simple polygons-k
k-Fréchet distance
Options -c
and -p
support three variants:
--dec=<eps>
decision problem: is d_F <= epsilon ?--opt
optimization problem: computes the exact value of d_F--app=<accuracy>
approximates d_F by nested intervals
The k-Frechet algorithm -k
accepts only --dec=<eps>
.
It computes the smallest k for a given epsilon.
There is no other optimization variant for the k-Frechet algorithm
(i.e. there is no option to compute the smallest epsilon for a given k).
--cores=<p>
defines the number of parallel threads on a multi-core system.
Results are usually best if you choose the number of physical CPU cores.
Performance does not benefit from hyperthreading on Intel processors.
Example for a quad core processor: --cores=4
is fine, but --cores=8
does not improve.
--large
minimizes memory usage (for really large input data).
--gpu
enables GPGPU support. You need an OpenCL 1.2 compatible card and driver.
Significant speedups require really powerful graphics cards.
--tile=<a,b>
fine-tuning parameter for GPGPU support.
Use --help
to view available options.
Examples:
Frechet-View.app/Contents/MacOS/frechet-view "Demo Data/trousers.jscript" -p --dec=45
Frechet-View.app/Contents/MacOS/frechet-view "Demo Data/trousers.jscript" -p --opt --cores=4
Frechet-View.app/Contents/MacOS/frechet-view "Demo Data/trousers.jscript" -c --approx=1-e6
Frechet-View.app/Contents/MacOS/frechet-view "Demo Data/criss-cross-2.jscript" -k --dec=0.4
Frechet-View.app/Contents/MacOS/frechet-view "Demo Data/trousers.jscript" -p --dec60 --cores=4 --gpu
Updated