Running grids of jobs
Often you need to investigate multiple combinations of parameters and/or likelihoods, or explore a range of different options. Using Cobaya grids, you can run and manage a set of runs using the grid scripts. This includes tools for submitting jobs to the cluster, managing running jobs, analysing results, and producing tables of results (like the Planck parameter tables).
To create a grid, you need a setting file specifying which combinations of parameters, likelihoods, etc. to use.
The command line to generate the basic structure and files in the grid_folder directory is:
cobaya-grid-create grid_folder [my_file]
where [my_file] is either a .py python setting file or a .yaml grid description.
This will create grid_folder if it does not exist, and generate a set of .yaml files for running each of the runs in the grid.
There is a simple generic python example and a cosmology yaml example which combines single parameter variations each with two different likelihoods.
Once the grid is created, you can check the list of runs included using:
cobaya-grid-list grid_folder
To actually submit and run the jobs, you’ll need to have a job script template configured for your cluster. You can check which jobs will be submitted using:
cobaya-grid-run grid_folder --dryrun
Simply remove the --dryrun to actually submit the jobs to run each of the items in the grid. Most of grid scripts have optional parameters to filter the grid to only run on specific subsets of items; use the -h option to see the full help.
You can use cobaya-running-jobs grid_folder to and monitor which jobs are queued and running, and cobaya-delete-jobs grid_folder to cancel jobs based on various name filters.
After the main samples are generated, if you have importance_runs set you can do the corresponding importance sampling on the generated chains using:
cobaya-grid-run grid_folder --importance
The grid also generates input files for minimization rather than sampling runs. If you also want best fits, run:
cobaya-grid-run grid_folder --minimize
Any custom settings for minimization come from the minimize_defaults dictionary in the input grid settings.
For best-fits from importance sampled grid combinations, run:
cobaya-grid-run grid_folder --importance_minimize
For any run that is expected to be fast, you can use --noqueue to run each item directly rather than using a queue submission.
Analysing grid results
While jobs are still running, you can use:
cobaya-grid-converge grid_folder --checkpoint
to show the convergence of each chain from the current checkpoint file. Filter to show on running jobs with --running, or conversely --not-running. If runs stop before they are converged, e.g. due to wall time limits, you can use:
cobaya-grid-run grid_folder --checkpoint_run
to re-start the finished runs that are not converged. --checkpoint_run has an optional parameter to an R-1 convergence value so that only chains with worse convergence than that are rerun.
To see parameter constraints, and convergence statistics using the written chain files, use:
cobaya-grid-getdist grid_folder --burn_remove 0.3
This will run GetDist on all the chains in the folder, removing the first 30% of each chain as burn in. You can use this while chains are still running, and incrementally update later by adding the --update_only switch (which only re-analyses chains which have changed). GetDist text file outputs are stored under a /dist folder in each subfolder of the grid results. To view GetDist-generated convergence numbers use:
cobaya-grid-converge grid_folder
Tables of grid results
After running cobaya-grid-getdist, you can combine results into a single PDF document (or latex file) containing tables of all the results:
cobaya-grid-tables grid_folder output_file --limit 2
This example will output tables containing 95% confidence limits by default, use --limit 1 to get 68% confidence tables. Alternatively use --all_limits to include all. By default the script generates a latex file and then compiles it with pdflatex. Use the --forpaper option to only generate latex.
There are a number of options to customize the result, compare results and give shift significances, e.g.:
cobaya-grid-tables grid_folder tables/baseline_params_table_95pc --limit 2 --converge 0.1 --musthave_data NPIPE lowl lowE --header_tex tableHeader.tex --skip_group nonbbn --skip_data JLA reion BK18
If you want a latex table to compare different parameter and data combinations, you can use the cobaya-grid-tables-compare command, see -h for options.
Exporting grid results
To copy a grid for distribution, without including unwanted files, use:
cobaya-grid-copy grid_folder grid_folder_export.zip
Add the --dist option to include GetDist outputs, or --remove_burn_fraction 0.3 to delete the first 30% of each chain file as burn in. You can also copy to a folder rather than .zip.
To extract a set of files from a grid, e.g. all GetDist .margestats table outputs and .covmats, use e.g.:
cobaya-grid-extract grid_folder output_dir .margestats .covmat
The cobaya-grid-cleanup script can be used to delete items in a grid_folder, e.g. to free space, delete incorrect results before a re-run, etc.
Grid script parameters
usage: cobaya-grid-create [-h] [--read-only]
[--install-reqs-at INSTALL_REQS_AT]
[--install-reqs-force] [--show-covmats]
batchPath [settingName]
Initialize grid using settings file
positional arguments:
batchPath root directory containing/to contain the grid (e.g.
grid_folder where output directories are created at
grid_folder/base/base_xx)
settingName python setting file for making or updating a grid, a
py filename or full name of a python module
options:
-h, --help show this help message and exit
--read-only option to configure an already-run existing grid
--install-reqs-at INSTALL_REQS_AT
install required code and data for the grid in the
given folder.
--install-reqs-force Force re-installation of apparently installed
packages.
--show-covmats Show which covmat is assigned to each chain.
usage: cobaya-grid-run [-h] [--converge CONVERGE] [--nodes NODES]
[--chains-per-node CHAINS_PER_NODE]
[--cores-per-node CORES_PER_NODE]
[--mem-per-node MEM_PER_NODE] [--walltime WALLTIME]
[--combine-one-job-name COMBINE_ONE_JOB_NAME]
[--runs-per-job RUNS_PER_JOB]
[--job-template JOB_TEMPLATE] [--program PROGRAM]
[--queue QUEUE] [--jobclass JOBCLASS] [--qsub QSUB]
[--dryrun] [--no_sub] [--noqueue [NOQUEUE]]
[--subitems] [--not_queued] [--minimize]
[--importance_minimize] [--minimize_failed]
[--checkpoint_run [CHECKPOINT_RUN]]
[--importance_ready] [--importance_changed]
[--parent_converge PARENT_CONVERGE] [--parent_stopped]
[--noimportance] [--importance [IMPORTANCE ...]]
[--importancetag [IMPORTANCETAG ...]]
[--name NAME [NAME ...]] [--param PARAM [PARAM ...]]
[--paramtag PARAMTAG [PARAMTAG ...]]
[--data DATA [DATA ...]]
[--datatag DATATAG [DATATAG ...]]
[--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]]
[--skip-data SKIP_DATA [SKIP_DATA ...]]
[--skip-param SKIP_PARAM [SKIP_PARAM ...]]
[--group GROUP [GROUP ...]]
[--skip-group SKIP_GROUP [SKIP_GROUP ...]] [--notexist]
[--notall NOTALL]
batchPath
Submit jobs to run chains or importance sample
positional arguments:
batchPath directory containing the grid
options:
-h, --help show this help message and exit
--converge CONVERGE minimum R-1 convergence
--nodes NODES
--chains-per-node CHAINS_PER_NODE
--cores-per-node CORES_PER_NODE
--mem-per-node MEM_PER_NODE
Memory in MB per node
--walltime WALLTIME
--combine-one-job-name COMBINE_ONE_JOB_NAME
run all one after another, under one job submission
(good for many fast operations)
--runs-per-job RUNS_PER_JOB
submit multiple mpi runs at once from each job script
(e.g. to get more than one run per node)
--job-template JOB_TEMPLATE
template file for the job submission script
--program PROGRAM actual program to run (default: cobaya-run -r)
--queue QUEUE name of queue to submit to
--jobclass JOBCLASS any class name of the job
--qsub QSUB option to change qsub command to something else
--dryrun just test configuration and give summary for checking,
don't produce or do anything
--no_sub produce job script but don't actually submit it
--noqueue [NOQUEUE] run directly, not using queue, ignoring most other
arguments (e.g. for fast tests). Optional argument
specifies how many to run in parallel.To use mpi,
include mpirun in the --program argument.
--subitems include sub-grid items
--not_queued
--minimize Run minimization jobs
--importance_minimize
Run minimization jobs for chains that are importance
sampled
--minimize_failed run where minimization previously failed
--checkpoint_run [CHECKPOINT_RUN]
run if stopped and not finished; if optional value
given then only run chains with convergence worse than
the given value
--importance_ready where parent chain has converged and stopped
--importance_changed run importance jobs where the parent chain has changed
since last run
--parent_converge PARENT_CONVERGE
minimum R-1 convergence for importance job parent
--parent_stopped only run if parent chain is not still running
--noimportance original chains only, no importance sampled
--importance [IMPORTANCE ...]
data names for importance sampling runs to include
--importancetag [IMPORTANCETAG ...]
importance tags for importance sampling runs to
include
--name NAME [NAME ...]
specific chain full name only
(base_paramx_data1_data2)
--param PARAM [PARAM ...]
runs including specific parameter only (paramx)
--paramtag PARAMTAG [PARAMTAG ...]
runs with specific parameter tag only (base_paramx)
--data DATA [DATA ...]
runs including specific data only (data1)
--datatag DATATAG [DATATAG ...]
runs with specific data tag only (data1_data2)
--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]
include only runs that include specific data (data1)
--skip-data SKIP_DATA [SKIP_DATA ...]
skip runs containing specific data (data1)
--skip-param SKIP_PARAM [SKIP_PARAM ...]
skip runs containing specific parameter (paramx)
--group GROUP [GROUP ...]
include only runs with given group names
--skip-group SKIP_GROUP [SKIP_GROUP ...]
exclude runs with given group names
--notexist only include chains that don't already exist on disk
--notall NOTALL only include chains where all N chains don't already
exist on disk
usage: cobaya-grid-converge [-h] [--converge CONVERGE] [--exist]
[--checkpoint] [--running] [--not_running]
[--stuck] [--noimportance]
[--importance [IMPORTANCE ...]]
[--importancetag [IMPORTANCETAG ...]]
[--name NAME [NAME ...]]
[--param PARAM [PARAM ...]]
[--paramtag PARAMTAG [PARAMTAG ...]]
[--data DATA [DATA ...]]
[--datatag DATATAG [DATATAG ...]]
[--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]]
[--skip-data SKIP_DATA [SKIP_DATA ...]]
[--skip-param SKIP_PARAM [SKIP_PARAM ...]]
[--group GROUP [GROUP ...]]
[--skip-group SKIP_GROUP [SKIP_GROUP ...]]
batchPath
Find chains which have failed or not converged, and showGelman-Rubin R-1
values for each run. Note need more than onechain for getdist to calculate
R-1.
positional arguments:
batchPath directory containing the grid
options:
-h, --help show this help message and exit
--converge CONVERGE minimum R-1 convergence
--exist chain must exist
--checkpoint use R-1 stored in checkpoint files (rather than
getdist output)
--running only check running chains
--not_running only check chains that are not running
--stuck finds chains with big spread in the last update time
--noimportance original chains only, no importance sampled
--importance [IMPORTANCE ...]
data names for importance sampling runs to include
--importancetag [IMPORTANCETAG ...]
importance tags for importance sampling runs to
include
--name NAME [NAME ...]
specific chain full name only
(base_paramx_data1_data2)
--param PARAM [PARAM ...]
runs including specific parameter only (paramx)
--paramtag PARAMTAG [PARAMTAG ...]
runs with specific parameter tag only (base_paramx)
--data DATA [DATA ...]
runs including specific data only (data1)
--datatag DATATAG [DATATAG ...]
runs with specific data tag only (data1_data2)
--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]
include only runs that include specific data (data1)
--skip-data SKIP_DATA [SKIP_DATA ...]
skip runs containing specific data (data1)
--skip-param SKIP_PARAM [SKIP_PARAM ...]
skip runs containing specific parameter (paramx)
--group GROUP [GROUP ...]
include only runs with given group names
--skip-group SKIP_GROUP [SKIP_GROUP ...]
exclude runs with given group names
usage: cobaya-grid-getdist [-h] [--update_only] [--make_plots] [--norun]
[--burn_removed] [--burn_remove BURN_REMOVE]
[--no_plots] [--delay DELAY] [--procs PROCS]
[--base_ini BASE_INI] [--command COMMAND] [--exist]
[--noimportance] [--importance [IMPORTANCE ...]]
[--importancetag [IMPORTANCETAG ...]]
[--name NAME [NAME ...]]
[--param PARAM [PARAM ...]]
[--paramtag PARAMTAG [PARAMTAG ...]]
[--data DATA [DATA ...]]
[--datatag DATATAG [DATATAG ...]]
[--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]]
[--skip-data SKIP_DATA [SKIP_DATA ...]]
[--skip-param SKIP_PARAM [SKIP_PARAM ...]]
[--group GROUP [GROUP ...]]
[--skip-group SKIP_GROUP [SKIP_GROUP ...]]
[--notexist]
batchPath
Run getdist over the grid of models. Use e.g. burn_remove=0.3 to remove 30% of
the chain as burn in.
positional arguments:
batchPath directory containing the grid
options:
-h, --help show this help message and exit
--update_only only run if getdist on chains that have been updated
since the last run
--make_plots run generated script plot files to make PDFs
--norun just make the .ini files, do not run getdist
--burn_removed if burn in has already been removed from chains
--burn_remove BURN_REMOVE
fraction of chain to remove as burn in (if not
importance sampled or already done)
--no_plots just make non-plot outputs (faster if using old
plot_data)
--delay DELAY run after delay of some number of seconds
--procs PROCS number of getdist instances to run in parallel
--base_ini BASE_INI default getdist settings .ini file
--command COMMAND program to run
--exist Silently skip all chains that don't exist
--noimportance original chains only, no importance sampled
--importance [IMPORTANCE ...]
data names for importance sampling runs to include
--importancetag [IMPORTANCETAG ...]
importance tags for importance sampling runs to
include
--name NAME [NAME ...]
specific chain full name only
(base_paramx_data1_data2)
--param PARAM [PARAM ...]
runs including specific parameter only (paramx)
--paramtag PARAMTAG [PARAMTAG ...]
runs with specific parameter tag only (base_paramx)
--data DATA [DATA ...]
runs including specific data only (data1)
--datatag DATATAG [DATATAG ...]
runs with specific data tag only (data1_data2)
--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]
include only runs that include specific data (data1)
--skip-data SKIP_DATA [SKIP_DATA ...]
skip runs containing specific data (data1)
--skip-param SKIP_PARAM [SKIP_PARAM ...]
skip runs containing specific parameter (paramx)
--group GROUP [GROUP ...]
include only runs with given group names
--skip-group SKIP_GROUP [SKIP_GROUP ...]
exclude runs with given group names
--notexist only include chains that don't already exist on disk
usage: cobaya-grid-tables [-h] [--converge CONVERGE] [--limit LIMIT]
[--all_limits] [--bestfitonly] [--nobestfit]
[--no_delta_chisq]
[--delta_chisq_paramtag DELTA_CHISQ_PARAMTAG]
[--changes_from_datatag CHANGES_FROM_DATATAG]
[--changes_from_paramtag CHANGES_FROM_PARAMTAG]
[--changes_adding_data CHANGES_ADDING_DATA [CHANGES_ADDING_DATA ...]]
[--changes_replacing CHANGES_REPLACING [CHANGES_REPLACING ...]]
[--changes_only]
[--changes_data_ignore CHANGES_DATA_IGNORE [CHANGES_DATA_IGNORE ...]]
[--systematic_average] [--shift_sigma_indep]
[--shift_sigma_subset]
[--paramNameFile PARAMNAMEFILE]
[--paramList PARAMLIST]
[--blockEndParams BLOCKENDPARAMS]
[--columns COLUMNS]
[--compare COMPARE [COMPARE ...]] [--titles TITLES]
[--forpaper] [--separate_tex]
[--header_tex HEADER_TEX] [--height HEIGHT]
[--width WIDTH] [--noimportance]
[--importance [IMPORTANCE ...]]
[--importancetag [IMPORTANCETAG ...]]
[--name NAME [NAME ...]] [--param PARAM [PARAM ...]]
[--paramtag PARAMTAG [PARAMTAG ...]]
[--data DATA [DATA ...]]
[--datatag DATATAG [DATATAG ...]]
[--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]]
[--skip-data SKIP_DATA [SKIP_DATA ...]]
[--skip-param SKIP_PARAM [SKIP_PARAM ...]]
[--group GROUP [GROUP ...]]
[--skip-group SKIP_GROUP [SKIP_GROUP ...]]
batchPath latex_filename
Make pdf tables from latex generated from getdist outputs
positional arguments:
batchPath directory containing the grid
latex_filename name of latex/PDF file to produce
options:
-h, --help show this help message and exit
--converge CONVERGE minimum R-1 convergence
--limit LIMIT sigmas of quoted confidence intervals
--all_limits
--bestfitonly
--nobestfit
--no_delta_chisq
--delta_chisq_paramtag DELTA_CHISQ_PARAMTAG
parameter tag to give best-fit chi-squared differences
--changes_from_datatag CHANGES_FROM_DATATAG
give fractional sigma shifts compared to a given data
combination tag
--changes_from_paramtag CHANGES_FROM_PARAMTAG
give fractional sigma shifts compared to a given
parameter combination tag
--changes_adding_data CHANGES_ADDING_DATA [CHANGES_ADDING_DATA ...]
give fractional sigma shifts when adding given data
--changes_replacing CHANGES_REPLACING [CHANGES_REPLACING ...]
give sigma shifts for results with data x, y, z
replacing data y, z.. with x
--changes_only Only include results in the changes_replacing set
--changes_data_ignore CHANGES_DATA_IGNORE [CHANGES_DATA_IGNORE ...]
ignore these data tags when mapping to reference for
comparison
--systematic_average Combine two chains and quote results for the
combination, e.g. as a crude way of including
systematic errors between likelihood versions. Used
with --changes_replacing or similar.
--shift_sigma_indep fractional shifts are relative to the sigma for
independent data (sigma^2=sigma1^2+sigma2^2
--shift_sigma_subset fractional shifts are relative to the sigma for
stricly subset data (sigma^2 = |sigma1^2-sigma2^2|,
regularized to sigma/20)
--paramNameFile PARAMNAMEFILE
.paramnames file for custom labels for parameters
--paramList PARAMLIST
.paramnames file listing specific parameters to
include (only).
--blockEndParams BLOCKENDPARAMS
a semi-colon separated list of parameters marking the
end of distinct parameter blocks (e.g. physical vs
nuisance parmeters, sampled vs derived)
--columns COLUMNS
--compare COMPARE [COMPARE ...]
--titles TITLES
--forpaper
--separate_tex
--header_tex HEADER_TEX
--height HEIGHT
--width WIDTH
--noimportance original chains only, no importance sampled
--importance [IMPORTANCE ...]
data names for importance sampling runs to include
--importancetag [IMPORTANCETAG ...]
importance tags for importance sampling runs to
include
--name NAME [NAME ...]
specific chain full name only
(base_paramx_data1_data2)
--param PARAM [PARAM ...]
runs including specific parameter only (paramx)
--paramtag PARAMTAG [PARAMTAG ...]
runs with specific parameter tag only (base_paramx)
--data DATA [DATA ...]
runs including specific data only (data1)
--datatag DATATAG [DATATAG ...]
runs with specific data tag only (data1_data2)
--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]
include only runs that include specific data (data1)
--skip-data SKIP_DATA [SKIP_DATA ...]
skip runs containing specific data (data1)
--skip-param SKIP_PARAM [SKIP_PARAM ...]
skip runs containing specific parameter (paramx)
--group GROUP [GROUP ...]
include only runs with given group names
--skip-group SKIP_GROUP [SKIP_GROUP ...]
exclude runs with given group names
usage: cobaya-grid-tables-compare [-h] [--params PARAMS [PARAMS ...]]
[--single_extparams SINGLE_EXTPARAMS [SINGLE_EXTPARAMS ...]]
[--compare COMPARE [COMPARE ...]]
[--nobestfits] [--limit LIMIT]
[--latex_filename LATEX_FILENAME]
[--math_columns] [--endline ENDLINE]
[--paramNameFile PARAMNAMEFILE]
[--noimportance]
[--importance [IMPORTANCE ...]]
[--importancetag [IMPORTANCETAG ...]]
[--name NAME [NAME ...]]
[--param PARAM [PARAM ...]]
[--paramtag PARAMTAG [PARAMTAG ...]]
[--data DATA [DATA ...]]
[--datatag DATATAG [DATATAG ...]]
[--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]]
[--skip-data SKIP_DATA [SKIP_DATA ...]]
[--skip-param SKIP_PARAM [SKIP_PARAM ...]]
[--group GROUP [GROUP ...]]
[--skip-group SKIP_GROUP [SKIP_GROUP ...]]
batchPath
Compare parameter constraints over set of models, producinglatext column
content for tables (but not full tables)
positional arguments:
batchPath directory containing the grid
options:
-h, --help show this help message and exit
--params PARAMS [PARAMS ...]
list of parameters to compare in the output tables
--single_extparams SINGLE_EXTPARAMS [SINGLE_EXTPARAMS ...]
list of single parameter extensions to restrict to
--compare COMPARE [COMPARE ...]
list of data tags to compare
--nobestfits
--limit LIMIT
--latex_filename LATEX_FILENAME
name of latex file to produce (otherwise stdout)
--math_columns whether to include $$ in the output for each column
--endline ENDLINE
--paramNameFile PARAMNAMEFILE
Optional parameter name file giving latex labels for
each parameter
--noimportance original chains only, no importance sampled
--importance [IMPORTANCE ...]
data names for importance sampling runs to include
--importancetag [IMPORTANCETAG ...]
importance tags for importance sampling runs to
include
--name NAME [NAME ...]
specific chain full name only
(base_paramx_data1_data2)
--param PARAM [PARAM ...]
runs including specific parameter only (paramx)
--paramtag PARAMTAG [PARAMTAG ...]
runs with specific parameter tag only (base_paramx)
--data DATA [DATA ...]
runs including specific data only (data1)
--datatag DATATAG [DATATAG ...]
runs with specific data tag only (data1_data2)
--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]
include only runs that include specific data (data1)
--skip-data SKIP_DATA [SKIP_DATA ...]
skip runs containing specific data (data1)
--skip-param SKIP_PARAM [SKIP_PARAM ...]
skip runs containing specific parameter (paramx)
--group GROUP [GROUP ...]
include only runs with given group names
--skip-group SKIP_GROUP [SKIP_GROUP ...]
exclude runs with given group names
usage: cobaya-grid-copy [-h] [--converge CONVERGE] [--dist] [--chains]
[--sym_link] [--no_config]
[--remove_burn_fraction REMOVE_BURN_FRACTION]
[--file_extensions FILE_EXTENSIONS [FILE_EXTENSIONS ...]]
[--skip_extensions SKIP_EXTENSIONS [SKIP_EXTENSIONS ...]]
[--max_age_days MAX_AGE_DAYS] [--dryrun] [--verbose]
[--zip] [--noimportance]
[--importance [IMPORTANCE ...]]
[--importancetag [IMPORTANCETAG ...]]
[--name NAME [NAME ...]] [--param PARAM [PARAM ...]]
[--paramtag PARAMTAG [PARAMTAG ...]]
[--data DATA [DATA ...]]
[--datatag DATATAG [DATATAG ...]]
[--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]]
[--skip-data SKIP_DATA [SKIP_DATA ...]]
[--skip-param SKIP_PARAM [SKIP_PARAM ...]]
[--group GROUP [GROUP ...]]
[--skip-group SKIP_GROUP [SKIP_GROUP ...]]
batchPath target_dir
copy or zip chains and optionally other files
positional arguments:
batchPath directory containing the grid
target_dir output root directory or zip file name
options:
-h, --help show this help message and exit
--converge CONVERGE minimum R-1 convergence
--dist include getdist outputs
--chains include chain files
--sym_link just make symbolic links to source directories
--no_config don't copy grid config info
--remove_burn_fraction REMOVE_BURN_FRACTION
fraction at start of chain to remove as burn in
--file_extensions FILE_EXTENSIONS [FILE_EXTENSIONS ...]
extensions to include
--skip_extensions SKIP_EXTENSIONS [SKIP_EXTENSIONS ...]
--max_age_days MAX_AGE_DAYS
only include files with date stamp at most
max_age_days old
--dryrun
--verbose
--zip make a zip file. Not needed if target_dir is a
filename ending in .zip
--noimportance original chains only, no importance sampled
--importance [IMPORTANCE ...]
data names for importance sampling runs to include
--importancetag [IMPORTANCETAG ...]
importance tags for importance sampling runs to
include
--name NAME [NAME ...]
specific chain full name only
(base_paramx_data1_data2)
--param PARAM [PARAM ...]
runs including specific parameter only (paramx)
--paramtag PARAMTAG [PARAMTAG ...]
runs with specific parameter tag only (base_paramx)
--data DATA [DATA ...]
runs including specific data only (data1)
--datatag DATATAG [DATATAG ...]
runs with specific data tag only (data1_data2)
--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]
include only runs that include specific data (data1)
--skip-data SKIP_DATA [SKIP_DATA ...]
skip runs containing specific data (data1)
--skip-param SKIP_PARAM [SKIP_PARAM ...]
skip runs containing specific parameter (paramx)
--group GROUP [GROUP ...]
include only runs with given group names
--skip-group SKIP_GROUP [SKIP_GROUP ...]
exclude runs with given group names
usage: cobaya-grid-extract [-h] [--converge CONVERGE] [--normalize_names]
[--tag_replacements TAG_REPLACEMENTS [TAG_REPLACEMENTS ...]]
[--noimportance] [--importance [IMPORTANCE ...]]
[--importancetag [IMPORTANCETAG ...]]
[--name NAME [NAME ...]]
[--param PARAM [PARAM ...]]
[--paramtag PARAMTAG [PARAMTAG ...]]
[--data DATA [DATA ...]]
[--datatag DATATAG [DATATAG ...]]
[--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]]
[--skip-data SKIP_DATA [SKIP_DATA ...]]
[--skip-param SKIP_PARAM [SKIP_PARAM ...]]
[--group GROUP [GROUP ...]]
[--skip-group SKIP_GROUP [SKIP_GROUP ...]]
batchPath target_dir file_extension
[file_extension ...]
copy all files of a given type from all getdist output directories in the grid
positional arguments:
batchPath directory containing the grid
target_dir
file_extension
options:
-h, --help show this help message and exit
--converge CONVERGE minimum R-1 convergence
--normalize_names replace actual name tags with normalized names
--tag_replacements TAG_REPLACEMENTS [TAG_REPLACEMENTS ...]
XX YY XX2 YY2 replaces name XX with YY, XX2 with YY2
etc.
--noimportance original chains only, no importance sampled
--importance [IMPORTANCE ...]
data names for importance sampling runs to include
--importancetag [IMPORTANCETAG ...]
importance tags for importance sampling runs to
include
--name NAME [NAME ...]
specific chain full name only
(base_paramx_data1_data2)
--param PARAM [PARAM ...]
runs including specific parameter only (paramx)
--paramtag PARAMTAG [PARAMTAG ...]
runs with specific parameter tag only (base_paramx)
--data DATA [DATA ...]
runs including specific data only (data1)
--datatag DATATAG [DATATAG ...]
runs with specific data tag only (data1_data2)
--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]
include only runs that include specific data (data1)
--skip-data SKIP_DATA [SKIP_DATA ...]
skip runs containing specific data (data1)
--skip-param SKIP_PARAM [SKIP_PARAM ...]
skip runs containing specific parameter (paramx)
--group GROUP [GROUP ...]
include only runs with given group names
--skip-group SKIP_GROUP [SKIP_GROUP ...]
exclude runs with given group names
usage: cobaya-grid-list [-h] [--converge CONVERGE] [--exists] [--normed]
[--noimportance] [--importance [IMPORTANCE ...]]
[--importancetag [IMPORTANCETAG ...]]
[--name NAME [NAME ...]] [--param PARAM [PARAM ...]]
[--paramtag PARAMTAG [PARAMTAG ...]]
[--data DATA [DATA ...]]
[--datatag DATATAG [DATATAG ...]]
[--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]]
[--skip-data SKIP_DATA [SKIP_DATA ...]]
[--skip-param SKIP_PARAM [SKIP_PARAM ...]]
[--group GROUP [GROUP ...]]
[--skip-group SKIP_GROUP [SKIP_GROUP ...]]
[--notexist]
batchPath
List items in a grid
positional arguments:
batchPath directory containing the grid
options:
-h, --help show this help message and exit
--converge CONVERGE minimum R-1 convergence
--exists chain must exist
--normed Output normed names
--noimportance original chains only, no importance sampled
--importance [IMPORTANCE ...]
data names for importance sampling runs to include
--importancetag [IMPORTANCETAG ...]
importance tags for importance sampling runs to
include
--name NAME [NAME ...]
specific chain full name only
(base_paramx_data1_data2)
--param PARAM [PARAM ...]
runs including specific parameter only (paramx)
--paramtag PARAMTAG [PARAMTAG ...]
runs with specific parameter tag only (base_paramx)
--data DATA [DATA ...]
runs including specific data only (data1)
--datatag DATATAG [DATATAG ...]
runs with specific data tag only (data1_data2)
--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]
include only runs that include specific data (data1)
--skip-data SKIP_DATA [SKIP_DATA ...]
skip runs containing specific data (data1)
--skip-param SKIP_PARAM [SKIP_PARAM ...]
skip runs containing specific parameter (paramx)
--group GROUP [GROUP ...]
include only runs with given group names
--skip-group SKIP_GROUP [SKIP_GROUP ...]
exclude runs with given group names
--notexist only include chains that don't already exist on disk
usage: cobaya-grid-cleanup [-h] [--converge CONVERGE] [--dist]
[--ext EXT [EXT ...]] [--empty] [--confirm]
[--chainnum CHAINNUM] [--noimportance]
[--importance [IMPORTANCE ...]]
[--importancetag [IMPORTANCETAG ...]]
[--name NAME [NAME ...]]
[--param PARAM [PARAM ...]]
[--paramtag PARAMTAG [PARAMTAG ...]]
[--data DATA [DATA ...]]
[--datatag DATATAG [DATATAG ...]]
[--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]]
[--skip-data SKIP_DATA [SKIP_DATA ...]]
[--skip-param SKIP_PARAM [SKIP_PARAM ...]]
[--group GROUP [GROUP ...]]
[--skip-group SKIP_GROUP [SKIP_GROUP ...]]
batchPath
Delete failed chains, files etc. Nothing is actually deleteuntil you add
--confirm, so you can check what you are doing first
positional arguments:
batchPath directory containing the grid
options:
-h, --help show this help message and exit
--converge CONVERGE minimum R-1 convergence
--dist set to only affect getdist output files
--ext EXT [EXT ...] file extensions to delete
--empty
--confirm
--chainnum CHAINNUM
--noimportance original chains only, no importance sampled
--importance [IMPORTANCE ...]
data names for importance sampling runs to include
--importancetag [IMPORTANCETAG ...]
importance tags for importance sampling runs to
include
--name NAME [NAME ...]
specific chain full name only
(base_paramx_data1_data2)
--param PARAM [PARAM ...]
runs including specific parameter only (paramx)
--paramtag PARAMTAG [PARAMTAG ...]
runs with specific parameter tag only (base_paramx)
--data DATA [DATA ...]
runs including specific data only (data1)
--datatag DATATAG [DATATAG ...]
runs with specific data tag only (data1_data2)
--musthave-data MUSTHAVE_DATA [MUSTHAVE_DATA ...]
include only runs that include specific data (data1)
--skip-data SKIP_DATA [SKIP_DATA ...]
skip runs containing specific data (data1)
--skip-param SKIP_PARAM [SKIP_PARAM ...]
skip runs containing specific parameter (paramx)
--group GROUP [GROUP ...]
include only runs with given group names
--skip-group SKIP_GROUP [SKIP_GROUP ...]
exclude runs with given group names