Running ROHSA

ROHSA

User Parameters

Input file

Once the source code is compiled, you can run ROHSA using the parameters.txt file.

&user_parameters
   filename = ''           !! name of the data file
   ,fileout = ''           !! name of the output result
   ,filename_noise = ''    !! name of the file with STD map (if noise .eq. true)
   ,n_gauss = 8            !! number of gaussian to fit
   ,lambda_amp = 1.d0      !! lambda for amplitude parameter
   ,lambda_mu = 1.d0       !! lamnda for mean position parameter
   ,lambda_sig = 1.d0      !! lambda for dispersion parameter
   ,lambda_var_amp = 0.d0  !! lambda for dispersion amplitude parameter
   ,lambda_var_mu = 0.d0   !! lambda for dispersion mean position parameter
   ,lambda_var_sig = 1.d0  !! lambda for dispersion variance parameter
   ,amp_fact_init = 0.66d0 !! times max amplitude of additional Gaussian
   ,sig_init = 5.d0        !! dispersion of additional Gaussian
   ,init_option = "mean"   !! init ROHSA with the mean or the std spectrum
   ,maxiter_init = 15000   !! max iteration for L-BFGS-B alogorithm (init mean spectrum)
   ,maxiter = 400          !! max iteration for L-BFGS-B alogorithm
   ,m = 10                 !! number of corrections used in the limited memory matrix by LBFGS-B
   ,noise = .false.        !! if false --> STD map computed by ROHSA with lstd and ustd (if true given by the user)
   ,regul = .true.         !! if true --> activate regulation
   ,descent = .true.       !! if true --> activate hierarchical descent to initiate the optimization
   ,lstd = 1               !! lower bound to compute the standard deviation map of the cube (if noise .eq. false)
   ,ustd = 20              !! upper bound to compute the standrad deviation map of the cube (if noise .eq. false)
   ,iprint = 1             !! print option
   ,iprint_init = -1       !! print option init
   /

parameters.txt contains all the free parameters of ROHSA.

Warning

If one of the parameters if missing, ROHSA wil take the default value encoded in the main.f90 source file.

Print option (L-BFGS-B argument)

iprint is an INTEGER variable that must be set by the user. It controls the frequency and type of output generated:

iprint<0    no output is generated;
iprint=0    print only one line at the last iteration;
0<iprint<99 print also f and |proj g| every iprint iterations;
iprint=99   print details of every iteration except n-vectors;
iprint=100  print also the changes of active set and final x;
iprint>100  print details of every iteration including x and g;
When iprint > 0, the file iterate.dat will be created to summarize the iteration.

Note

For any question about the parameters, please contact us at the following address : antoine.marchal@ias.u-psud.fr

Data format

ROHSA does not support FITS file in version 1.2.0. It wiil be added in a next release. You can find a python code : fits2dat.py in src/ directory to convert your .fits file into a .dat file.

Running ROHSA

You can now run your Gaussian decomposition with ROHSA. Enjoy !

./ROHSA parameters.txt

(marchalenv) dapmcw133:src antoinemarchal$ ./ROHSA parameters.txt 

 opening file and reading data
 -------------------------------------------------------------------------
25 September 2018   5:09:11.592 PM

   ____     ___    _   _   ____       _    
  |  _ \   / _ \  | | | | / ___|     / \   
  | |_) | | | | | | |_| | \___ \    / _ \  
  |  _ <  | |_| | |  _  |  ___) |  / ___ \ 
  |_| \_\  \___/  |_| |_| |____/  /_/   \_\ 

  Version 1.0.0
  ROHSA is released as open source code
  Check out the documentation: https://antoinemarchal.github.io/ROHSA/

 run: ./ROHSA parameters.txt
 -------------------------------------------------------------------------
 fileout = 'result.dat'

 ______Parameters_____

 n_gauss =            6
 lambda_amp =    1.0000000000000000     
 lambda_mu =    1.0000000000000000     
 lambda_sig =    1.0000000000000000     
 lambda_var_amp =    0.0000000000000000     
 lambda_var_mu =    0.0000000000000000     
 lambda_var_sig =    1.0000000000000000     
 amp_fact_init =   0.66000000000000003     
 sig_init =    5.0000000000000000     
 init_option = mean    
 maxiter_init =        15000
 maxiter =           20
 lstd =            1
 ustd =           20
 noise =  F
 regul =  T
 descent =  T

 dim_v, dim_y, dim_x =          101          16          16

 number of los =          256
 nside =            4

 Reshape cube, new dimensions :
 dim_v, dim_y, dim_x =          101          16          16

 Compute mean and std spectrum
                     Start iteration

 Start hierarchical descent
 Init mean spectrum

 Update parameters level            0 >           1

 Update parameters level            1 >           2

 Update parameters level            2 >           4

 Update parameters level            3 >           8

 Update parameters level            4 >          16

 Reshape cube, restore initial dimensions :
 dim_v, dim_y, dim_x =          101          16          16

 Update last level ...


 _____ Write output file _____

 ##################################################################
25 September 2018   5:09:11.654 PM
  Terminate
                         ROHSA ALGORITHM

 ##################################################################

User Guide