B â‹dD)ã@sNddlZddlZddlZddlmZmZddlmZmZGdd„dejƒZ dS)éN)Ú conversionsÚ boundariesé)ÚuniformÚboundedcsTeZdZdZdZd‡fdd „ Z‡fd d „Zd d „Zd‡fdd„ Ze dd„ƒZ ‡Z S)Ú UniformF0Taua² A distribution uniform in QNM frequency and damping time. Constraints may be placed to exclude frequencies and damping times corresponding to specific masses and spins. To ensure a properly normalized pdf that accounts for the constraints on final mass and spin, a renormalization factor is calculated upon initialization. This is calculated numerically: f0 and tau are drawn randomly, then the norm is scaled by the fraction of points that yield final masses and spins within the constraints. The `norm_tolerance` keyword arguments sets the error on the estimate of the norm from this numerical method. If this value is too large, such that no points are found in the allowed region, a ValueError is raised. Parameters ---------- f0 : tuple or boundaries.Bounds The range of QNM frequencies (in Hz). tau : tuple or boundaries.Bounds The range of QNM damping times (in s). final_mass : tuple or boundaries.Bounds, optional The range of final masses to allow. Default is [0,inf). final_spin : tuple or boundaries.Bounds, optional The range final spins to allow. Must be in [-0.996, 0.996], which is the default. rdfreq : str, optional Use the given string as the name for the f0 parameter. Default is 'f0'. damping_time : str, optional Use the given string as the name for the tau parameter. Default is 'tau'. norm_tolerance : float, optional The tolerance on the estimate of the normalization. Default is 1e-3. norm_seed : int, optional Seed to use for the random number generator when estimating the norm. Default is 0. After the norm is estimated, the random number generator is set back to the state it was in upon initialization. Examples -------- Create a distribution: >>> dist = UniformF0Tau(f0=(10., 2048.), tau=(1e-4,1e-2)) Check that all random samples drawn from the distribution yield final masses > 1: >>> from pycbc import conversions >>> samples = dist.rvs(size=1000) >>> (conversions.final_mass_from_f0_tau(samples['f0'], samples['tau']) > 1.).all() True Create a distribution with tighter bounds on final mass and spin: >>> dist = UniformF0Tau(f0=(10., 2048.), tau=(1e-4,1e-2), final_mass=(20., 200.), final_spin=(0,0.996)) Check that all random samples drawn from the distribution are in the final mass and spin constraints: >>> samples = dist.rvs(size=1000) >>> (conversions.final_mass_from_f0_tau(samples['f0'], samples['tau']) >= 20.).all() True >>> (conversions.final_mass_from_f0_tau(samples['f0'], samples['tau']) < 200.).all() True >>> (conversions.final_spin_from_f0_tau(samples['f0'], samples['tau']) >= 0.).all() True >>> (conversions.final_spin_from_f0_tau(samples['f0'], samples['tau']) < 0.996).all() True Zuniform_f0_tauNÚf0Útauçü©ñÒMbP?rc s*|dkrtdƒ‚|dkr tdƒ‚||_||_||||i} tt|ƒjf| Ž|dkr\dtjf}|dkrhd}tj |d|dd|_ tj |d|dd|_ tj   ¡} tj  |¡td|d ƒ} tt|ƒj| d } tj  | ¡| | ¡ ¡} | dkrútd ƒ‚|jt | ¡t | ¡7_t |j¡|_dS) Nzmust provide a range for f0zmust provide a range for taug)g¬Zd;ßï¿g¬Zd;ßï?rr)Z min_boundZ max_boundgð?é)ÚsizezWthe normalization is < then the norm_tolerance; try again with a smaller nrom_tolerance)Ú ValueErrorÚrdfreqÚ damping_timeÚsuperrÚ__init__ÚnumpyÚinfrZBoundsÚfinal_mass_boundsÚfinal_spin_boundsÚrandomZ get_stateÚseedÚintÚrvsZ set_stateÚ _constraintsÚsumZ_lognormÚlogÚexpZ_norm)Úselfrr Ú final_massÚ final_spinrrÚnorm_toleranceÚ norm_seedZ parent_argsÚsZnsamplesÚdrawsZnum_in)Ú __class__©úd/work/yifan.wang/ringdown/master-ringdown-env/lib/python3.7/site-packages/pycbc/distributions/qnm.pyrhs4     zUniformF0Tau.__init__cs&tt|ƒ |¡}|r"|| |¡M}|S)N)rrÚ __contains__r)rÚparamsÚisin)r%r&r'r(szUniformF0Tau.__contains__c Csª||j}||j}t d|j¡rJ|j d¡}t|dƒt|dƒ}}nd\}}tjddBtj ||||d}tj ||||d}|j   |¡|j   |¡@} WdQRX| S) Nzf_\d{3}Zf_rr)r r Úignore)Úinvalid)ÚlÚm)rrÚreÚmatchÚstriprrZerrstaterZfinal_mass_from_f0_tauZfinal_spin_from_f0_taurr(r) rr)rr Úmoder-r.ÚmfZsfr*r&r&r'r“s    zUniformF0Tau._constraintsrc s„t|ƒ}dd„|jDƒ}tj||d}|}d}xP|r~tt|ƒj|d}| |¡}| ¡}||||||…<||7}||}q0W|S)aDraw random samples from this distribution. Parameters ---------- size : int, optional The number of draws to do. Default is 1. Returns ------- array A structured array of the random draws. cSsg|] }|tf‘qSr&)Úfloat)Ú.0Úpr&r&r'ú ³sz$UniformF0Tau.rvs..)Údtyper)r ) rr)rZzerosrrrrr) rr r8ZarrÚ remainingZkeepidxr$ÚmaskZaddpts)r%r&r'r¥s   zUniformF0Tau.rvsc Cs|}t| tj¡ƒ}t |||d¡}t |||d¡}| |d|¡rR| |d|¡}nd}| |d|¡rt| |d|¡}nd}|t||gƒkstdƒ‚t |||d¡} t |||d¡} i} | |d|¡rØt | |d|¡ƒ| d<| |d |¡rüt | |d |¡ƒ| d <|f||| | ||d œ| —ŽS) aInitialize this class from a config file. Bounds on ``f0``, ``tau``, ``final_mass`` and ``final_spin`` should be specified by providing ``min-{param}`` and ``max-{param}``. If the ``f0`` or ``tau`` param should be renamed, ``rdfreq`` and ``damping_time`` should be provided; these must match ``variable_args``. If ``rdfreq`` and ``damping_time`` are not provided, ``variable_args`` are expected to be ``f0`` and ``tau``. Only ``min/max-f0`` and ``min/max-tau`` need to be provided. Example: .. code-block:: ini [{section}-f0+tau] name = uniform_f0_tau min-f0 = 10 max-f0 = 2048 min-tau = 0.0001 max-tau = 0.010 min-final_mass = 10 Parameters ---------- cp : pycbc.workflow.WorkflowConfigParser WorkflowConfigParser instance to read. section : str The name of the section to read. variable_args : str The name of the variable args. These should be separated by ``pycbc.VARARGS_DELIM``. Returns ------- UniformF0Tau : This class initialized with the parameters provided in the config file. rr rrz8variable args do not match rdfreq and damping_time namesrr r!r")rr rr rr) ÚsetÚsplitÚpycbcZ VARARGS_DELIMrZget_param_bounds_from_configZhas_option_tagZ get_opt_tagr r4r) ÚclsÚcpÚsectionZ variable_argsÚtagrr rrrr Z extra_optsr&r&r'Ú from_configÀs6)  zUniformF0Tau.from_config)NNNNrr r r)r) Ú__name__Ú __module__Ú __qualname__Ú__doc__Únamerr(rrÚ classmethodrBÚ __classcell__r&r&)r%r'rsL# r) r/rr=rrÚrrZUniformrr&r&r&r'Ús