ó áp7]c@`s”dZddlmZmZmZddlZddlmZddl m Z dZ dZ e e BZ defd „ƒYZd efd „ƒYZdS( s{ State Space Representation, Kalman Filter, Smoother, and Simulation Smoother Author: Chad Fulton License: Simplified-BSD i(tdivisiontabsolute_importtprint_functionNi(tKalmanSmoother(ttoolsitSimulationSmoothercB`s_eZdZdddgZdddd„Zddddd„Zd„Zdddd„ZRS( s State space representation of a time series process, with Kalman filter and smoother, and with simulation smoother. Parameters ---------- k_endog : array_like or integer The observed time-series process :math:`y` if array like or the number of variables in the process if an integer. k_states : int The dimension of the unobserved state process. k_posdef : int, optional The dimension of a guaranteed positive definite covariance matrix describing the shocks in the measurement equation. Must be less than or equal to `k_states`. Default is `k_states`. simulation_smooth_results_class : class, optional Default results class to use to save output of simulation smoothing. Default is `SimulationSmoothResults`. If specified, class must extend from `SimulationSmoothResults`. simulation_smoother_classes : dict, optional Dictionary with BLAS prefixes as keys and classes as values. **kwargs Keyword arguments may be used to provide default values for state space matrices, for Kalman filtering options, for Kalman smoothing options, or for Simulation smoothing options. See `Representation`, `KalmanFilter`, and `KalmanSmoother` for more details. tsimulate_statetsimulate_disturbancet simulate_allcK`sntt|ƒj|||||dkr4t}n||_|dk rO|n tjjƒ|_i|_ dS(N( tsuperRt__init__tNonetSimulationSmoothResultstsimulation_smooth_results_classRtprefix_simulation_smoother_maptcopyt _simulators(tselftk_endogtk_statestk_posdefR tsimulation_smoother_classestkwargs((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyR 6s   cK`s§|dkr£d}|r%|tO}n|r8|tO}n|rK|tO}n|dkr£t|dk|dk|dkgƒ }|r”tdƒ‚n|j}q£n|S(sÛ Get simulation output bitmask Helper method to get final simulation output bitmask from a set of optional arguments including the bitmask itself and possibly boolean flags. Parameters ---------- simulation_output : integer, optional Simulation output bitmask. If this is specified, it is simply returned and the other arguments are ignored. simulate_state : boolean, optional Whether or not to include the state in the simulation output. simulate_disturbance : boolean, optional Whether or not to include the state and observation disturbances in the simulation output. simulate_all : boolean, optional Whether or not to include all simulation output. \*\*kwargs Additional keyword arguments. Present so that calls to this method can use \*\*kwargs without clearing out additional arguments. isKInvalid simulation output options: given options would result in no output.N(R tSIMULATION_STATEtSIMULATION_DISTURBANCEtSIMULATION_ALLtallt ValueErrortsmoother_output(Rtsimulation_outputRRRRt argument_set((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pytget_simulation_outputJs      c C`sñ|jƒ\}}}}} |jd|ƒ||jksQ||j|jk räd} d} |j} |j} |j}|j}|j}|j }|j }|j |}||j || | |||||| | |ƒ |j|R cB`seZdZd„Zed„ƒZejd„ƒZed„ƒZejd„ƒZed„ƒZejd„ƒZed„ƒZ e jd „ƒZ ed „ƒZ ed „ƒZ ed „ƒZ ed „ƒZ ed„ƒZed„ƒZed„ƒZddded„ZRS(s… Results from applying the Kalman smoother and/or filter to a state space model. Parameters ---------- model : Representation A Statespace representation simulation_smoother : {{prefix}}SimulationSmoother object The Cython simulation smoother object with which to simulation smooth. Attributes ---------- model : Representation A Statespace representation dtype : dtype Datatype of representation matrices prefix : str BLAS prefix of representation matrices simulation_output : integer Bitmask controlling simulation output. simulate_state : boolean Flag for if the state is included in simulation output. simulate_disturbance : boolean Flag for if the state and observation disturbances are included in simulation output. simulate_all : boolean Flag for if simulation output should include everything. generated_measurement_disturbance : array Measurement disturbance variates used to genereate the observation vector. generated_state_disturbance : array State disturbance variates used to genereate the state and observation vectors. generated_obs : array Generated observation vector produced as a byproduct of simulation smoothing. generated_state : array Generated state vector produced as a byproduct of simulation smoothing. simulated_state : array Simulated state. simulated_measurement_disturbance : array Simulated measurement disturbance. simulated_state_disturbance : array Simulated state disturbance. cC`sm||_|j|_|j|_||_d|_d|_d|_d|_d|_ d|_ d|_ dS(N( tmodelR R!t_simulation_smootherR t"_generated_measurement_disturbancet_generated_state_disturbancet_generated_obst_generated_statet_simulated_statet"_simulated_measurement_disturbancet_simulated_state_disturbance(RRQRK((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyR ;s          cC`s |jjS(N(RRR(R((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyRJscC`s||j_dS(N(RRR(Rtvalue((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyRNscC`st|jt@ƒS(N(tboolRR(R((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyRRscC`s4t|ƒr|jtB|_n|jt@|_dS(N(R[RR(RRZ((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyRVs cC`st|jt@ƒS(N(R[RR(R((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyR]scC`s4t|ƒr|jtB|_n|jt@|_dS(N(R[RR(RRZ((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyRas cC`st|jt@ƒS(N(R[RR(R((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyRjscC`s4t|ƒr|jtB|_n|jt@|_dS(N(R[RR(RRZ((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyRns cC`si|jdkrb|jj|jj}tj|jj| dt ƒj |jj|jjƒ|_n|jS(s~ Randomly drawn measurement disturbance variates, used to construct `generated_obs`. Notes ----- .. math:: arepsilon_t^+ ~ N(0, H_t) If `disturbance_variates` were provided to the `simulate()` method, then this returns those variates (which were N(0,1)) transformed to the distribution above. RN( RSR RQR%RR.R0RRRDR5treshape(Rtend((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyt!generated_measurement_disturbanceus 'cC`si|jdkrb|jj|jj}tj|jj|dt ƒj |jj|jj ƒ|_n|jS(s‰ Randomly drawn state disturbance variates, used to construct `generated_state` and `generated_obs`. Notes ----- .. math:: \eta_t^+ ~ N(0, Q_t) If `disturbance_variates` were provided to the `simulate()` method, then this returns those variates (which were N(0,1)) transformed to the distribution above. RN( RTR RQR%RR.R0RRRDR5R\R(Rtstart((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pytgenerated_state_disturbances 'cC`s7|jdkr0tj|jjdtƒ|_n|jS(s) Generated vector of observations by iterating on the observation and transition equations, given a random initial state draw and random disturbance draws. Notes ----- .. math:: y_t^+ = d_t + Z_t \alpha_t^+ + \varepsilon_t^+ RN(RUR R.R0RRR8R5(R((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyR8¦scC`s7|jdkr0tj|jjdtƒ|_n|jS(s  Generated vector of states by iterating on the transition equation, given a random initial state draw and random disturbance draws. Notes ----- .. math:: \alpha_{t+1}^+ = c_t + T_t \alpha_t^+ + \eta_t^+ RN(RVR R.R0RRR9R5(R((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyR9»scC`s7|jdkr0tj|jjdtƒ|_n|jS(s® Random draw of the state vector from its conditional distribution. Notes ----- .. math:: \alpha ~ p(\alpha \mid Y_n) RN(RWR R.R0RRRFR5(R((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyRFÏs cC`s7|jdkr0tj|jjdtƒ|_n|jS(sî Random draw of the measurement disturbance vector from its conditional distribution. Notes ----- .. math:: \varepsilon ~ N(\hat \varepsilon, Var(\hat \varepsilon \mid Y_n)) RN(RXR R.R0RRt!simulated_measurement_disturbanceR5(R((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyRaâs  cC`s7|jdkr0tj|jjdtƒ|_n|jS(sÓ Random draw of the state disturbance vector from its conditional distribution. Notes ----- .. math:: \eta ~ N(\hat \eta, Var(\hat \eta \mid Y_n)) RN(RYR R.R0RRtsimulated_state_disturbanceR5(R((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyRb÷s  iÿÿÿÿc C`sd|_d|_d|_d|_d|_d|_d|_d|_|jj ƒ\}}}}} |jj d|ƒ|dk r°|j j t j|d|jƒd|ƒn |j jƒ|dk rô|j jt j|d|jƒd|ƒn |j jƒ|j j|ƒdS(sä Perform simulation smoothing Does not return anything, but populates the object's `simulated_*` attributes, as specified by simulation output. Parameters ---------- simulation_output : integer, optional Bitmask controlling simulation output. Default is to use the simulation output defined in object initialization. disturbance_variates : array_likes, optional Random values to use as disturbance variates, distributed standard Normal. Usually only specified if results are to be replicated (e.g. to enforce a seed) or for testing. If not specified, random variates are drawn. initial_state_variates : array_likes, optional Random values to use as initial state variates. Usually only specified if results are to be replicated (e.g. to enforce a seed) or for testing. If not specified, random variates are drawn. R R!R"N(R RSRTRVRURWRXRYRQR#R$RRR4R.R0R!tdraw_disturbance_variatestset_initial_state_variatestdraw_initial_state_variatesR7( RRRDtinitial_state_variatestpretransformed_variatesR R!R?R@RA((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyR7 s*                N(RMRNROR tpropertyRtsetterRRRR^R`R8R9RFRaRbR tFalseR7(((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyR s&.  (ROt __future__RRRtnumpyR.tkalman_smootherRtRRRRRtobjectR (((sMlib/python2.7/site-packages/statsmodels/tsa/statespace/simulation_smoother.pyts  ÷