Replies: 3 comments
-
|
From: fo***@kn*** (Andrew Foster) Do you have: If that doesn't work, send your BART.cfg and we'll help you troubleshoot. -asdf- On Sat, Oct 17, 2015 at 3:12 AM, Ian Crossfield <ia***@lp***>
-------------- next part -------------- |
Beta Was this translation helpful? Give feedback.
-
|
From: pc***@fu*** (Patricio Cubillos) Hi Ian, Andrew, I removed the “mpi=True” argument from the BART config file because it always runs in multiple cores (I forgot to update the doc because I was finishing the dissertation). Ian, what I think is going on is that before the parallel MCMC, the code generates an opacity grid. We haven’t parallelized this step in particular because it was a lower-priority task. Making the opacity grid may take from ~30 min to a few hours (depending on your set up/CPU). But once you made it, you can just re-use it for future MCMC runs. Regards,
-------------- next part -------------- |
Beta Was this translation helpful? Give feedback.
-
|
From: ia***@lp*** (Ian Crossfield) Hi Patricio, Andrew, Thanks for the tips! My goal was to repeat the Diamond-Lowe et al. analysis of HD209458b $topdir/BART/modules/transit/pylineread/src/pylineread.py -c The configuration files I used are attached. A 14MB ".tli" file was created by #1 above. Many thanks, folks. On 10/17/15 12:11 PM, Patricio Cubillos wrote:
-- -------------- next part -------------- Multiple-value arguments can be set separated either by a white space ora line break.This is the list (and source) of the currently supported Line Lists:Partridge and Schwenke (H2O):- http://kurucz.harvard.edu/molecules/h2o/h2ofastfix.binHITRAN and HITEMP:- http://www.cfa.harvard.edu/hitran/Schwenke (TiO):- http://kurucz.harvard.edu/molecules/tio/tioschwenke.binWith the correspoding partition functions:- http://kurucz.harvard.edu/molecules/h2o/h2opartfn.dat- Total Internal Partition Sums (TIPS) *- http://kurucz.harvard.edu/molecules/tio/tiopart.dat(*): Transit incorporates a modified version of the TIPS codePath(s) to the database linelists:db_list = ./06_hit08_f53.par ./01_hit08.par ./02_hit08_f53.par ./05_hit08.par Path(s) to the partition function files:part_list = ctips ctips ctips ctips Type of each input database:hit: hitran and hitempps: Partridge and Schwenke H2Ots: Schwenke TiOvo: Plez VOdbtype = hit hit hit hit Output TLI filename:output = ./H2O_CO_CO2_CH4_HITRAN_2-30um.tli Initial wavelength in microns:iwav = 2.0 Final wavelength in microns:fwav = 30.0 Verbosity level (0 -- 20):verb = 10 -------------- next part -------------- Transit Configuration-File Template:Comment (# or ;) and empty lines are allowed.To set an argument, write the argument name, followed by theargument value (white−space separated). No need for the ’equal’sign, nor quotes for string values.For the full list of arguments see Transit User Guide or type transit --help:::::::::: Input files :::::::::::::::::::::::::::::::::::::::::::::Path to atmospheric file:atm ../BART/modules/transit/transit/examples/demo/HD209458b_demo.atm Path to transit-line-information (TLI) file:linedb /home/ianc/python/BART_demo/run/H2O_CO_CO2_CH4_HITRAN_2-30um.tli Path to collision-induced-absorption (CIA) file:csfile ../BART/modules/transit/inputs/CIA_H2H2_400-7000K.dat Path to molecular information file:molfile ../BART/modules/transit/inputs/molecules.dat :::::::::: Spectrum sampling :::::::::::::::::::::::::::::::::::::::Lowest wavelength boundary (see also wnhigh):wllow 2.0 Highest wavelength boundary (see also wnlow):wlhigh 30.0 Wavelength conversion factor to cm (default: 1e-4, microns)wlfct 1e-4 Wavenumber sampling interval for output:wndelt 1.0 Wavenumber oversampling factor:wnosamp 2160 Wavenumber conversion factor to cm-1 (default: 1.0):wnfct 1.0 :::::::::: Geometry ::::::::::::::::::::::::::::::::::::::::::::::::Raypath geometry (transit, for transmission; oreclipse, for dayside emission):solution eclipse For eclipse mode, set the angle of the incident ray-paths:raygrid 0 20 40 60 80 :::::::::: Optical Depth :::::::::::::::::::::::::::::::::::::::::::Maximum optical depth:toomuch 10 Line-strength threshold:ethresh 1e-6 :::::::::: Opacity Grid Calculation ::::::::::::::::::::::::::::::::Lowest temperature boundary in K (default: 500)tlow 400 Highest temperature boundary in K (default: 3000)thigh 3000 Temperature sampling in K (default: 100):tempdelt 100 Path to the opacity file (input/output):opacityfile ./H2O_CO_CO2_CH4_HITRAN_2-30um.dat :::::::::: Line Broadening :::::::::::::::::::::::::::::::::::::::::Voigt profile half width (in Doppler or Lorentz HWHM):nwidth 20 :::::::::: Verbosity :::::::::::::::::::::::::::::::::::::::::::::::Verbosity level for on-screen display (0 -- 20):verb 11 :::::::::: Output files ::::::::::::::::::::::::::::::::::::::::::::Emission (eclipse) or modulation (transit) spectrum:outflux ./H2O_CO_CO2_CH4_HITRAN_2-30um_demo_spectrum.dat Location of the output directory:loc_dir = ./spitz_hd209458b/ Tepfile name:tep_name = ../BART/inputs/tep/HD209458b.tep Stellar Kurucz file:kurucz = ../BART/inputs/kurucz/fp00k2odfnew.pck Atmospheric model file:atmfile = ../BART/modules/transit/transit/examples/demo/HD209458b_demo.atm :::::::::: MCMC arguments ::::::::::::::::::::::::::::::::::::::::::Data (eclipse depths):data = 0.00119 0.00123 0.00134 0.00215 0.00315 0.00338 Data 1-sigma uncertainties:uncert = 0.00007 0.00006 0.00035 0.00008 0.00000315 0.00026 Waveband filters:filter = ../BART/inputs/filters/ispitzer_irac1_sa.dat List of moelcules being fit:molfit = CH4 MCMC temperature boundaries:Tmin = 400.0 Temperature-profile model:PTtype = line The model fitting parameters:log(kappa) log(g1) log(g2) alpha beta log(CH4) log(CO) log(H2O) log(CO2)params = -2.0 0.0 1.0 0.0 0.98 -0.5 -0.5 -0.5 -0.5 DEMC setup:numit = 100000 grtest = True Verbosity level:verb = 1 :::::::::: Transit Arguments :::::::::::::::::::::::::::::::::::::::tconfig = ./config_demo_transit.cfg Spectrum array:wllow = 2.0 wndelt = 1.0 Reference pressure where Radius(p0) = Rp:refpress = 0.1 solution = eclipse nwidth = 20 Opacity grid:tlow = 400 Output spectrum:outflux = ./BART_spitz209_demo_eclipse_spectrum.dat |
Beta Was this translation helpful? Give feedback.
-
The following is an archived message from the BART-user mailing list, which has now closed.
From: ia***@lp*** (Ian Crossfield)
Date: Sat, 17 Oct 2015 09:12:26 +0200
Subject: [BART-user] Running BART on multiple cores?
Hi all,
What's the trick to running BART across multiple cores? I assumed that
one merely needed to set:
nchains = 10
... but the process just runs on one core. I didn't immediately see this
issue addressed in the documentation, either. Any suggestions would be
most welcome!
Sincerely,
-Ian
--
Ian Crossfield
http://www.lpl.arizona.edu/~ianc/
U. Arizona, Lunar & Planetary Lab
Kuiper Space Sciences 401
+1 520.626.2083
Beta Was this translation helpful? Give feedback.
All reactions