get_conf

Synopsis

Return information regarding the confidence method.

Syntax

get_conf()
get_conf_results()
get_conf_opt( [name] )

The get_conf command returns the list of current settings defining the confidence estimation method (see "ahelp conf"); these settings may be modified by the user. This function accepts no arguments. The returned information is:

eps - precision of resulting limits, default=0.01
fast - if 'True', the fit optimization method is switched to a faster option to reduce the time per fit; default=False, so that the current optimization method is used
max_rstat - maximum possible value of the reduced statistic, default=3
maxfits - maximum number of tries to fit the parameter space, default=5
maxiters- maximum number of tries to bracket desired limits, before returning, default=200
openinterval - how the confidence function should behave when it cannot locate the root within the open interval; see "ahelp conf" for a detailed explanation; default=False
parallel - switch which allows confidence to be run as a parallel process; default=True
numcores - number of cores to use when conf command is run in parallel
remin - if during search, a new fit statistic is found that is smaller than original best fit, the difference between new and original statistics must be greater than this value, default=0.01
sigma - to get 1, 2, or 3-sigma limits, default=1
soft_limits - a switch which allows the search to go all the way out to the hard parameter limits; if switched on, the search will be restricted to occur within the soft limits, which can be changed by the user, default=False
tol - tolerance for the fit, default=0.20000000000000001
verbose - print information to the screen?; default=False

get_conf_results

The get_conf_results command accesses results from the last run of the confidence command. The function accepts no arguments. The results include the following attributes:

datasets - Data sets in fit
methodname - Estimation method name
iterfitname - Iterative fitting method name, if one specified
fitname - Fitting method name
statname - Statistic name
sigma - Change in statistic
percent - % confidence interval corresponding to sigma
parnames - Model parameter names
parvals - Model parameter fit values
parmins - Model parameter minimum values
parmaxes - Model parameter maximum values
warnings - Warning messages

get_conf_opt

The get_conf_opt command returns the value of a confidence option by name. If the name isn't specified, it returns a dictionary of all options that govern how confidence is run.

eps - precision of resulting limits, default=0.01
fast - if 'True', the fit optimization method is switched to a faster option to reduce the time per fit; default=False, so that the current optimization method is used
max_rstat - maximum possible value of the reduced statistic, default=3
maxfits - maximum number of tries to fit the parameter space, default=5
maxiters- maximum number of tries to bracket desired limits, before returning, default=200
openinterval - how the confidence function should behave when it cannot locate the root within the open interval; see "ahelp conf" for a detailed explanation; default=False
parallel - switch which allows confidence to be run as a parallel process; default=True
numcores - number of cores to use when conf command is run in parallel
remin - if during search, a new fit statistic is found that is smaller than original best fit, the difference between new and original statistics must be greater than this value, default=0.01
sigma - to get 1, 2, or 3-sigma limits, default=1
soft_limits - a switch which allows the search to go all the way out to the hard parameter limits; if switched on, the search will be restricted to occur within the soft limits, which can be changed by the user, default=False
tol - tolerance for the fit, default=0.20000000000000001
verbose - print information to the screen?; default=False

Example 1

sherpa> get_conf()

Return a variable listing all of the settings that govern how the confidence method calculates confidence limits. Sample output is:

sherpa> print(get_conf())
name         = confidence
verbose      = False
openinterval = False
max_rstat    = 3
maxiters     = 200
soft_limits  = False
eps          = 0.01
fast         = False
maxfits      = 5
remin        = 0.01
tol          = 0.2
sigma        = 1
parallel     = True
numcores     = 2

sherpa> set_conf_opt("maxiters",300)

sherpa> print(get_conf())
name         = confidence
verbose      = False
openinterval = False
max_rstat    = 3
maxiters     = 300
soft_limits  = False
eps          = 0.01
fast         = False
maxfits      = 5
remin        = 0.01
tol          = 0.2
sigma        = 1
parallel     = True
numcores     = 2

Example 2

sherpa> get_conf_results()

Return a variable listing all of the results from the last run of the confidence command. Sample output is:

sherpa-> print(get_conf_results())
datasets    = (1,)
methodname  = confidence
iterfitname = none
fitname     = levmar
statname    = chi2gehrels
sigma       = 1
percent     = 68.2689492137
parnames    = ('p1.gamma', 'p1.ampl')
parvals     = (2.1585155113403327, 0.00022484014787994827)
parmins     = (-0.082785567348122591, -1.4825550342799376e-05)
parmaxes    = (0.083410634144100104, 1.4825550342799376e-05)
nfits       = 13

Example 3

sherpa> get_conf_opt()
sherpa> get_conf_opt("sigma")
1

When called without a specific option name, the get_conf_opt command returns the dictionary of all options that govern how confidence is run. If an option name such as "sigma" is specified, it returns the value of the sigma option, which is "1" in this example. Sample output of get_conf_opt() is:

sherpa> print(get_conf_opt())
{'verbose': False, 'openinterval': False, 'max_rstat': 3, 'maxiters': 200, 'soft_limits': False, 'eps': 0.01, 'fast': False, 'maxfits': 5, 'remin': 0.01, 'tol': 0.20000000000000001, 'sigma': 1, ''numcores': 2, parallel': True}

Bugs

See the bugs pages on the Sherpa website for an up-to-date listing of known bugs.