HTML automatically generated with rman
Table of Contents
mkspiral - uniform-density test disk plus spiral in spherical potential
mkspiral out=snapshot [parameters=values ...]
sets up a cold disk of test particles orbiting clockwise in the gravitational
potential of a user-supplied potential (in potential(5NEMO)
format). In addition
a spiral is added with constant width in degrees along a galactic circle.
For given ratio of spiral/disk perturbation a and width w (now measured
in radians) the fraction of particles in the disk will be about pi / (pi
+ a.w.sqrt(2.pi)) for not too large w. The inverse wavelenght of the spiral
(related to the pitch angle) will be positive for a trailing spiral pattern.
have to be used if for some reason
the user want finite mass particles and a perfectly centered set of particles.
Note this particle distribution is only useful for display, e.g. spiral structure
in a Longitude-Velocity diagram, and should not be integrated with programs
mkspiral is also used in the standard benchmark as
to measure disk I/O speed on a file just over 2GB.
parameters are recognized in any order if the keyword is also given:
This program has been used as benchmark to
test writing >2GB files (40 models of 1,000,000 particles creates a file
just over this magic limit):
- The data are written to this file in SnapShot(5NEMO)
format [no default].
- Number of disk particles [default: 2048]
- name of the potential, see $NEMO/obj/potential for the current repository.
The user can supply his own, see potential(5NEMO)
. [default: plummer].
- Paramaters to the user supplied potential. The number of parameters depends
on the potential supplied, the first parameter is reserved for a pattern
speed in future implementation. [default: not supplied, parameters as defined
- Data_file(s) to the user supplied potential.
The number of files depends on the potential supplied. [default: not supplied,
datafile(s) defined by potential(5)
- Inner cutoff
radius of test-particle disk. [Default: 0.0].
- Outer cutoff
radius of test-particle disk. [Default: 1.0].
- Total mass of the disk.
- Ratio of maximum spiral perturbation w.r.t. base disk.
- Wavenumber of the spiral pattern, such that 2.pi.k.r
is the phase of the spiral density maximum. Positive means trailing spiral
pattern. (note that this is defined in term of sign=, see below) [default:
- Width of the spiral arm along a galactic circle. Depending on
the next keyword this width is either is degrees along a galactic circle
if that is to be constant, or in length units if the width of the spiral
is to be constant. [default: 15].
- Is the width of the spiral defined
in angular terms, in which case the physical real width linearly increases
with distance from the galactic center, or is is constant as a function
of radius. [default: t, i.e. defined in angular terms].
random number seed. A zero seed will generate a system-time dependant seed.
- Sign of the angular moment vector. Positive means
counter-clock wise rotation. Not
- Velocity dispersion (isotropic)
in the X-Y plane, plus optionally an offset and exponential dropoff: s =
s1 + s2 * exp(-r/s3). Default: 0
- The ratio of the tangential to
radial velocity dispersion. By default 1, but if you need realistic values,
our local solar neighbor measures about 0.5 but the best self-consistent
(despite these models are NOT) estimate can be set by using the value -1,
in which case 0.5*kappa/omega is used. Default: 1
of output models. [Default: 1].
- Text headline for output
file [default: not used].
- Should the spiral be linear or logarithmic.
Although spiral galaxies are mostly logarithmic, for historic reasons the
default here is linear. [Default: t]
- The mode for testing.
test=0 gives the normal test disk; test=1 the 2D coordinates are uniform
random (-1,1); test=2 the 2D coordinates are all set to 1.0; test=3 no initialization
(everything 0) 
% time mkspiral s000 1000000 nmodel=40
Here is a sample on some 2002-class machines:
U-sparc IIi/440 MHz: 161.30u 42.59s 5:04.85 66.8%
PentiumIV 2.2 GHz: 29.54u 10.68s 2:40.14 25.1%
PentiumIV 2.4 GHz: 26.46u 10.14s 1:37.71 37.4%
G5 2.0GHz: 28.400u 24.660s 1:05.41 81.1% (at no load of 0.0)
33.460u 44.950s 13:01.31 10.0% (at full load of 2.0)
P4/1.6 GHz: 69.940u 18.470s 4:50.64 30.4% (laptop)
P4/2.6 GHz + SATA 27.400u 16.320s 0:52.13 83.8% 0+0k 0+0io 300pf+0w
(at no load of 0.0)
27.790u 20.020s 1:52.22 42.6% 0+0k 0+0io 300pf+0w
sum: 79.35u 3.72s 2:17.64 60.3% P2400
tsf: 0.00u 7.59s 1:59.46 6.3% P2400
sum: 13.610u 4.130s 1:10.12 25.2% P2600/SATA
tsf: 0.010u 7.800s 1:09.74 11.1% P2600/SATA
sum: 13.334u 2.915s 0:57.67 28.1% P2800/SATA (Peter’s asus )
tsf: 0.003u 4.696s 0:44.36 10.5% P2800/SATA
sum: 68.63u 22.34s 2:31.54 60.0% U440
tsf: 21.85u 30.73s 2:32.06 34.5% U440
And here a more modern benchmark on a i7-8550U CPU processor with different
% time mkspiral $out 1000000 nmodel=40 seed=123 test=$test
$test $out cpu:
0 . 3.05u 1.09 i5-1135G7
0 . 4.15u 1.56 i5-10xxx
0 s0 6.21u 2.16s 0:08.38 full computation on a i7-8550U CPU
0 . 6.40u 0.69s 0:07.10
1 . 3.49u 0.68s 0:04.17 only using 4 xrandom’s calls
2 . 2.32u 0.71s 0:03.03 constant initialization
2 . 1.43u 0.68s 0:02.12 - same , but with better caching barebody.h
3 . 0.65u 0.56s 0:01.22 no initialization (using calloc)
13-mar-89 V1.0 created PJT
6-apr-89 V1.1 seed=0 implemented PJT
7-apr-89 V1.2 angular keyword introduced PJT
26-feb-93 V1.4 new potential keywrds PJT
16-sep-95 V1.5 added nmodel=, sign= ; radii now random PJT
26-may-02 added benchmark data for >2GB files PJT
3-aug-09 V1.8 and documented forgotten stuff PJT
11-aug-09 V1.9 added sigmator= PJT
24-oct-2020 V1.10 added test= PJT
Table of Contents