def plot_data(self):
     plt.clf()  # clear the figure
     plt.title("Parent incomes vs. student grade")
     plt.plot(self.incomes,
              self.grades,
              color='orange',
              marker='o',
              linestyle='')
def displayRetireWRate(month, rates, terms):
    plt.figure('retireRate')
    plt.clf()
    for rate in rates:
        xvals, yvals = retire(month, rate, terms)
        plt.plot(xvals,
                 yvals,
                 label='monthly: ' + str(month) + ' rate of:     ' +
                 str(int(rate * 100)))
        plt.legend(loc='upper left')
def displayRetireWMonthlies(monthlies, rate, terms):
    plt.figure('retireMonth')
    plt.clf()
    for monthly in monthlies:
        xvals, yvals = retire(
            monthly, rate,
            terms)  # using base and savings list as x and y values
        plt.plot(xvals,
                 yvals,
                 label='retire with monthly inst of ' + str(monthly))
        plt.legend()
Esempio n. 4
0
def convert_all_to_png(vis_path, out_dir="maps_png", size=None):

    units = {
        'gas_density': 'Gas Density [g/cm$^3$]',
        'Tm': 'Temperature [K]',
        'Tew': 'Temperature [K]',
        'S': 'Entropy []',
        'dm': 'DM Density [g/cm$^3$]',
        'v': 'Velocity [km/s]'
    }

    log_list = ['gas_density']

    for vis_file in os.listdir(vis_path):
        if ".dat" not in vis_file:
            continue
        print "converting %s" % vis_file
        map_type = re.search('sigma_(.*)_[xyz]', vis_file).group(1)

        (image, pixel_size,
         axis_values) = read_visualization_data(vis_path + "/" + vis_file,
                                                size)
        print "image width in Mpc/h: ", axis_values[-1] * 2.0

        x, y = np.meshgrid(axis_values, axis_values)

        cmap_max = image.max()
        cmap_min = image.min()
        ''' plotting '''
        plt.figure(figsize=(5, 4))

        if map_type in log_list:
            plt.pcolor(x, y, image, norm=LogNorm(vmax=cmap_max, vmin=cmap_min))
        else:
            plt.pcolor(x, y, image, vmax=cmap_max, vmin=cmap_min)

        cbar = plt.colorbar()
        if map_type in units.keys():
            cbar.ax.set_ylabel(units[map_type])

        plt.axis(
            [axis_values[0], axis_values[-1], axis_values[0], axis_values[-1]])

        del image

        plt.xlabel(r"$Mpc/h$", fontsize=18)
        plt.ylabel(r"$Mpc/h$", fontsize=18)

        out_file = vis_file.replace("dat", "png")

        plt.savefig(out_dir + "/" + out_file, dpi=150)

        plt.close()
        plt.clf()
def displayRetireWMonthsandRates(monthlies, rates, terms):
    plt.figure('retire both')
    plt.clf()
    plt.xlim(30 * 12,
             40 * 12)  # focusing only on the last 10 years of investment
    for monthly in monthlies:
        for rate in rates:
            xvals, yvals = retire(monthly, rate, terms)
            plt.plot(xvals,
                     yvals,
                     label='retire with ' + str(monthly) + ":" +
                     str(int(rate * 100)))
            plt.legend(loc='upper left')
Esempio n. 6
0
def convert_all_to_png(vis_path, out_dir = "maps_png", size = None) :

    units = { 'gas_density' : 'Gas Density [g/cm$^3$]',
              'Tm' : 'Temperature [K]',
              'Tew' : 'Temperature [K]',
              'S' : 'Entropy []',
              'dm' : 'DM Density [g/cm$^3$]',
              'v' : 'Velocity [km/s]' }

    log_list = ['gas_density']

    for vis_file in os.listdir(vis_path) :
        if ".dat" not in vis_file :
            continue
        print "converting %s" % vis_file
        map_type = re.search('sigma_(.*)_[xyz]', vis_file).group(1)

        (image, pixel_size, axis_values) = read_visualization_data(vis_path+"/"+vis_file, size)
        print "image width in Mpc/h: ", axis_values[-1]*2.0

        x, y = np.meshgrid( axis_values, axis_values )

        cmap_max = image.max()
        cmap_min = image.min()


        ''' plotting '''
        plt.figure(figsize=(5,4))

        if map_type in log_list:
            plt.pcolor(x,y,image, norm=LogNorm(vmax=cmap_max, vmin=cmap_min))
        else :
            plt.pcolor(x,y,image, vmax=cmap_max, vmin=cmap_min)

        cbar = plt.colorbar()
        if map_type in units.keys() :
            cbar.ax.set_ylabel(units[map_type])

        plt.axis([axis_values[0], axis_values[-1],axis_values[0], axis_values[-1]])

        del image

        plt.xlabel(r"$Mpc/h$", fontsize=18)
        plt.ylabel(r"$Mpc/h$", fontsize=18)

        out_file = vis_file.replace("dat", "png")

        plt.savefig(out_dir+"/"+out_file, dpi=150 )

        plt.close()
        plt.clf()
Esempio n. 7
0
def drawAdoptionNetworkMPL(G, fnum=1, show=False, writeFile=None):
    """Draws the network to matplotlib, coloring the nodes based on adoption. 
    Looks for the node attribute 'adopted'. If the attribute is True, colors 
    the node a different color, showing adoption visually. This function assumes
    that the node attributes have been pre-populated.
    
    :param networkx.Graph G: Any NetworkX Graph object.
    :param int fnum: The matplotlib figure number. Defaults to 1.
    :param bool show: 
    :param str writeFile: A filename/path to save the figure image. If not
                             specified, no output file is written.
    """
    Gclean = G.subgraph([n for n in G.nodes() if n not in nx.isolates(G)])
    plt.figure(num=fnum, figsize=(6,6))
    # clear figure
    plt.clf()
    
    # Blue ('b') node color for adopters, red ('r') for non-adopters. 
    nodecolors = ['b' if Gclean.node[n]['adopted'] else 'r' \
                  for n in Gclean.nodes()]
    layout = nx.spring_layout(Gclean)
    nx.draw_networkx_nodes(Gclean, layout, node_size=80, 
                           nodelist=Gclean.nodes(), 
                           node_color=nodecolors)
    nx.draw_networkx_edges(Gclean, layout, alpha=0.5) # width=4
    
    # TODO: Draw labels of Ii values. Maybe vary size of node.
    # TODO: Color edges blue based on influences from neighbors
    
    influenceEdges = []
    for a in Gclean.nodes():
        for n in Gclean.node[a]['influence']:
            influenceEdges.append((a,n))
    
    if len(influenceEdges)>0:
        nx.draw_networkx_edges(Gclean, layout, alpha=0.5, width=5,
                               edgelist=influenceEdges,
                               edge_color=['b']*len(influenceEdges))
    
    #some extra space around figure
    plt.xlim(-0.05,1.05)
    plt.ylim(-0.05,1.05)
    plt.axis('off')
    
    if writeFile != None:
        plt.savefig(writeFile)
    
    if show:
        plt.show()
Esempio n. 8
0
def dynamic_img_show(img,title_str='',fig_size=[14,8],hide_axes=True):
    '''Show image <img>. If called repeatedly within a cycle will dynamically redraw image.
    #DEMO
    import time

    for i in range(10):
        img = np.zeros([50,50])
        img[:i*5]=1
        dynamic_img_show(img,'iter=%s'%i)
        time.sleep(0.1)
    '''
    plt.clf()
    plt.title(title_str)
    plt.imshow(img)
    plt.xticks([]); plt.yticks([]);
    plt.gcf().set_size_inches(fig_size)
    display.display(plt.gcf())
    display.clear_output(wait=True)   
def displayRetireWMonthsandRates2(monthlies, rates, terms):
    plt.figure('retire better')
    plt.clf()
    plt.xlim(30 * 12, 40 * 12)
    monthLabels = ['r', 'b', 'g', 'k']
    rateLabels = ['-', 'o', '^']
    for i in range(len(monthlies)):
        monthly = monthlies[i]
        monthLabel = monthLabels[i % len(
            monthLabels
        )]  # using remainder to pick new label for each new month choice
        for j in range(len(rates)):
            rate = rates[j]
            rateLable = rateLabels[j % len(
                rateLabels)]  # if more months, cycles back to the beginning
            xvals, yvals = retire(monthly, rate, terms)
            plt.plot(xvals,
                     yvals,
                     monthLabel + rateLable,
                     label='retire: ' + str(monthly) + " : " +
                     str(int(rate * 100)))
            plt.legend(loc="upper left")
Esempio n. 10
0
        xx = xx.astype(np.float)
        yy = yy.astype(np.float)
       
        dimx = float(dimx)
        dimy=float(dimy)        
        nTimesInX = np.floor(xx / M).max() + 1
 
        seg_cpu = np.floor(yy / M)  * nTimesInX + np.floor(xx / M)
        seg_cpu = seg_cpu.astype(np.int32)
        return seg_cpu


def random_permute_seg(seg):
    p=np.random.permutation(seg.max()+1)   
    seg2 = np.zeros_like(seg)
    for c in range(seg.max()+1):             
        seg2[seg==c]=p[c]
    return seg2.astype(np.int32)


if __name__ == "__main__":  
    tic = time.clock()
    seg= get_init_seg(500, 500,17,True)      
#    seg= get_init_seg(512, 512,50,False)  
    toc = time.clock()
    print toc-tic 
    print 'k = ', seg.max()+1
    plt.figure(1)
    plt.clf()  
    plt.imshow(seg,interpolation="Nearest")
    plt.axis('scaled') 
Esempio n. 11
0
        yy = yy.astype(np.float)

        dimx = float(dimx)
        dimy = float(dimy)
        nTimesInX = np.floor(xx / M).max() + 1

        seg_cpu = np.floor(yy / M) * nTimesInX + np.floor(xx / M)
        seg_cpu = seg_cpu.astype(np.int32)
        return seg_cpu


def random_permute_seg(seg):
    p = np.random.permutation(seg.max() + 1)
    seg2 = np.zeros_like(seg)
    for c in range(seg.max() + 1):
        seg2[seg == c] = p[c]
    return seg2.astype(np.int32)


if __name__ == "__main__":
    tic = time.clock()
    seg = get_init_seg(500, 500, 17, True)
    #    seg= get_init_seg(512, 512,50,False)
    toc = time.clock()
    print(toc - tic)
    print('k = ', seg.max() + 1)
    plt.figure(1)
    plt.clf()
    plt.imshow(seg, interpolation="Nearest")
    plt.axis('scaled')
    def fit(self, use_prior=False, proposal_scale=0.001, use_local=False):
        nLevels = self.nLevels
        tw = self.tw
        sigma_lm = self.sigma_lm
        self.use_local = use_local

        inference_record = Bunch()
        inference_record.nLevels = nLevels

        inference_record.tw_args = tw.args
        inference_record.steps = []
        inference_record.use_prior = use_prior
        inference_record.proposal_scale = proposal_scale
        inference_record.sigma_lm = sigma_lm

        try:
            run_lengths = self.run_lengths
        except AttributeError:
            self.set_run_lengths([500] * self.nLevels)
            run_lengths = self.run_lengths


#            raise Exception("self.set_run_lengths was not called yet")

        for i in range(nLevels):
            #        for i in range(nLevels+5):

            if i < nLevels:
                level = i
                if level == 0:
                    theta = tw.ms.L_cpa_space[level].get_zeros_theta()
                else:
                    theta_fine = tw.ms.L_cpa_space[level].get_zeros_theta()
                    tw.ms.propogate_theta_coarse2fine(theta_coarse=theta,
                                                      theta_fine=theta_fine)
                    theta = theta_fine
                _sigma_lm = sigma_lm
            else:
                _sigma_lm *= 0.9

            print '-' * 10, 'level', level, '-' * 10
            cpa_space = tw.ms.L_cpa_space[level]
            print cpa_space

            data = {
                'src': self.src,
                'dst': self.dst,
                'transformed': self.transformed
            }

            if use_prior:

                lp_func = LP(ms=tw.ms,
                             msp=tw.msp,
                             level=level,
                             SDLP=SDLP,
                             required={})

            else:

                lp_func = None
            sampler = Metropolis(
                ll_func=LL(
                    ms=tw.ms,
                    level=level,
                    SDLL=SDLL,
                    data=data,
                    required={
                        'sigma_lm': _sigma_lm,
                        'params_flow_int': tw.params_flow_int_coarse
                    },
                ),
                proposal=Proposal(ms=tw.ms,
                                  msp=tw.msp,
                                  level=level,
                                  scale=proposal_scale,
                                  use_local=use_local),
                #                                 proposal=ProposalSphericalGaussian(ms=tw.ms,
                #                                                                    level=level,
                #                                                   scale=0.1 / (1.2**level)
                #                                                   scale=0.01,
                #                                          scale=0.1 / (1.1**level)
                #                                           scale=0.1
                #                                                   ) ,
                lp_func=lp_func,
                wlp=self.wlp)

            sampler.set_theta(theta)
            run_length = run_lengths[level]
            sampler.run(run_length)
            theta = sampler.theta_current  # prepare for next iteration

            inference_record.steps.append(sampler.get_record())

            if i >= nLevels - 1:
                plt.figure(i + 1)
                plt.clf()
                self.disp(sampler=sampler)

        inference_record.theta = theta.copy()
        steps = inference_record.steps
        nAccepted = [step.nAccepted for step in steps]
        run_lengths = [step.N for step in steps]
        times = [step.runs[0].time for step in steps]
        total_time = sum(times)

        print "run_lengths:", run_lengths
        print "nAccepted:", nAccepted

        print 'times:'
        print_iterable(times)
        print 'time total:', total_time

        return theta.copy(), inference_record
Esempio n. 13
0
def main(data, infernece_params, dispOn):

    img1 = data.img1  # src
    img2 = data.img2  # dst

    if img1.shape != img2.shape:
        raise ValueError(img1.shape, img2.shape)

    nRows, nCols = img1.shape[:2]
    nPts = nRows * nCols

    reg = Register(nRows=nRows,
                   nCols=nCols,
                   base=inference_params.base,
                   nLevels=inference_params.nLevels,
                   tess='I',
                   zero_v_across_bdry=[False, False],
                   sigma_signal=inference_params.sigma_signal,
                   scale_spatial=inference_params.scale_spatial,
                   scale_value=inference_params.scale_value,
                   wlp=inference_params.wlp,
                   ll_type=inference_params.ll_type,
                   only_local=False,
                   valid_outside=infernece_params.valid_outside)

    reg.set_dense(domain_start=0, domain_end=nPts)

    reg.set_data(x=reg.tw.pts_src_dense,
                 signal_src=img1,
                 signal_dst=img2,
                 isbinary=inference_params.isbinary)

    print reg

    #    reg.set_run_lengths([50000]*reg.nLevels)
    reg.set_run_lengths([infernece_params.MCMCniters_per_level] * reg.nLevels)

    #    if inside_spyder():
    #        reg.set_run_lengths([100]*reg.nLevels)
    #    #        reg.set_run_lengths([50000]*reg.nLevels)
    #    else:
    #        reg.set_run_lengths(run_lengths)

    theta_est, inference_record = reg.fit(
        use_prior=infernece_params.use_prior,
        use_local=infernece_params.use_local,
        dispOn=dispOn,
        #interp_type_for_ll=cv2.INTER_LANCZOS4,
        interp_type_for_ll='gpu_linear',
        interp_type_during_visualization=cv2.INTER_LANCZOS4)

    if dispOn:
        plt.figure(1000)
        plt.clf()
        reg.plot_inference_summary(inference_record)

    reg.transformed.gpu2cpu()
    reg.signal.transformed.gpu2cpu()

    inference_record.transformed = reg.transformed.cpu
    inference_record.src = reg.src.cpu
    inference_record.signal = Bunch()
    inference_record.signal.src = reg.signal.src.cpu
    inference_record.signal.dst = reg.signal.dst.cpu
    inference_record.signal.transformed = reg.signal.transformed.cpu

    #    if TF.dump_results:
    #    Pkl.dump(results_filename,inference_record,create_dir_if_needed=1,override=1)

    return reg, inference_record, theta_est
def example(base=[5],
            scale_spatial=100,
            nLevels=2,
            zero_v_across_bdry=[1],
            use_local_basis=True):   
    nPtsDense = 10000   
    tw = TransformWrapper(nCols=100,
                          nLevels=nLevels,  
                          base=base,
                          scale_spatial=scale_spatial,
                          nPtsDense=nPtsDense,
                          zero_v_across_bdry=zero_v_across_bdry)
     
    print_iterable(tw.ms.L_cpa_space)

    seed=0
    np.random.seed(seed)    
               

    
    
    for level in range(tw.ms.nLevels):
        cpa_space = tw.ms.L_cpa_space[level]
        Avees = cpa_space.Avees
        velTess = cpa_space.zeros_velTess()
        
        if use_local_basis:
            if 0:
                tw.sample_gaussian_velTess(level,Avees,velTess,mu=None) 
                Avees*=0.001
                velTess*=0.001  
            else:
                
                if not zero_v_across_bdry[0]:
                    velTess[:]=10*np.random.standard_normal(velTess.shape)
                    cpa_space.velTess2Avees(velTess=velTess,Avees=Avees)
                else:
                    velTess[1:-1]=10*np.random.standard_normal(velTess[1:-1].shape)
                    cpa_space.velTess2Avees(velTess=velTess,Avees=Avees)
                    
            cpa_space.velTess2Avees(velTess=velTess,Avees=Avees)  
            
        else:
            theta= cpa_space.get_zeros_theta()
            tw.sample_gaussian(level,Avees,theta,mu=None)    
#            theta/=10             
            cpa_space.theta2Avees(theta=theta,Avees=Avees)  
        
        # This step is important and must be done 
        # before are trying to "use" the new values of 
        # the (vectorized) A's.            
        tw.update_pat_from_Avees(Avees,level) 
        
        pts_src = tw.x_dense        
            
        tw.calc_v(level=level,pts=pts_src,v=tw.v_dense)
        tw.v_dense.gpu2cpu()
            
        
        pts_fwd = CpuGpuArray.zeros_like(pts_src) # Create a buffer for the output      
        
        tw.calc_T_fwd(pts_src,pts_fwd,level=level)    
        pts_fwd.gpu2cpu()  

        pts_inv = CpuGpuArray.zeros_like(pts_src) # Create a buffer for the output
        tw.calc_T_inv(pts_src,pts_inv,level=level)    
        pts_inv.gpu2cpu()     
     
        plt.figure(level)
        plt.clf()   
        interval = pts_src.cpu # interval doesn't have to be pts_src.cpu
        Visualize.simple(tw.x_dense,tw.v_dense,interval,pts_src,
                         transformed_fwd=pts_fwd,transformed_inv=pts_inv,
                         cpa_space=cpa_space)
        
    return tw
Esempio n. 15
0
def main(data,infernece_params,dispOn ):
 
    img1=data.img1 # src
    img2=data.img2 # dst
    
    if img1.shape != img2.shape:
        raise ValueError(img1.shape,img2.shape)

    nRows,nCols = img1.shape[:2]  
    nPts = nRows * nCols 
      
    reg = Register(nRows=nRows,
                   nCols=nCols,
                   base=inference_params.base,
                   nLevels=inference_params.nLevels, tess='I',
                   zero_v_across_bdry=[False,False],
                   sigma_signal=inference_params.sigma_signal ,                   
                   scale_spatial=inference_params.scale_spatial,
                   scale_value=inference_params.scale_value,
                   wlp=inference_params.wlp,                   
                   ll_type=inference_params.ll_type,
                   only_local=False,
                   valid_outside=infernece_params.valid_outside)  
                 
    reg.set_dense(domain_start=0,domain_end=nPts)
                        
    reg.set_data(x=reg.tw.pts_src_dense,
                 signal_src=img1,signal_dst=img2,
                 isbinary=inference_params.isbinary)
    
     
    print reg
    
    
#    reg.set_run_lengths([50000]*reg.nLevels)
    reg.set_run_lengths([infernece_params.MCMCniters_per_level]*reg.nLevels)
        
#    if inside_spyder():
#        reg.set_run_lengths([100]*reg.nLevels)
#    #        reg.set_run_lengths([50000]*reg.nLevels)
#    else:        
#        reg.set_run_lengths(run_lengths)
    
    theta_est,inference_record = reg.fit(use_prior=infernece_params.use_prior,
                                     use_local=infernece_params.use_local,
                                     dispOn=dispOn,
                                     #interp_type_for_ll=cv2.INTER_LANCZOS4,
                                     interp_type_for_ll= 'gpu_linear',
                                     interp_type_during_visualization=cv2.INTER_LANCZOS4)
     
    if dispOn:
        plt.figure(1000)
        plt.clf()
        reg.plot_inference_summary(inference_record)    
    
    
    reg.transformed.gpu2cpu()
    reg.signal.transformed.gpu2cpu()
    
   
    inference_record.transformed = reg.transformed.cpu
    inference_record.src = reg.src.cpu
    inference_record.signal = Bunch()
    inference_record.signal.src = reg.signal.src.cpu
    inference_record.signal.dst = reg.signal.dst.cpu
    inference_record.signal.transformed = reg.signal.transformed.cpu
    

    
#    if TF.dump_results:
#    Pkl.dump(results_filename,inference_record,create_dir_if_needed=1,override=1)

     
    return  reg,inference_record,theta_est
Esempio n. 16
0
            transformed.gpu2cpu()
            v_dense.gpu2cpu()
           
            if 0:
                plt.figure(17)
                for c,A in enumerate(As):        
                    _x = np.ones((2,100))
                    m=cpa_space.cells_verts[c,0,0]
                    M=cpa_space.cells_verts[c,1,0]
                    _x[0] = np.linspace(m,M,100)
                    _v = A.dot(_x).flatten()
                    plt.plot(_x[0],_v) 
                      
            if 1:
#                plt.figure()
                plt.figure(1);plt.clf()
                plt.subplot(231)                 
                plt.plot(interval,src.cpu)   
                plt.title('src')
                plt.subplot(232)         
#                plt.plot(interval[1:],np.diff(src)/(interval[1]-interval[0]))
                dx=interval[1]-interval[0]
                plt.plot(interval[1:],np.diff(src.cpu.ravel())/dx)
                plt.title(" d/dx src")  
                plt.ylim(0,.5)
                plt.subplot(233)
                plt.plot(np.linspace(cpa_space.XMINS[0],cpa_space.XMAXS[0],
                                     interval.size),v_dense.cpu.ravel())
                plt.ylim(-1,1)
                plt.title('velocity')           
                plt.subplot(234)
Esempio n. 17
0
    def fit(self,use_prior=False,proposal_scale=0.001,use_local=True,dispOn=True,
            interp_type_for_ll=None,
            interp_type_during_visualization=None,
            scale_local_proposal=None):
        nLevels=self.nLevels
        tw = self.tw 
        sigma_signal = self.sigma_signal
        self.use_local = use_local
        
        inference_record = Bunch()        

        inference_record.tw_args = tw.args        
        inference_record.steps = []
        inference_record.use_prior = use_prior
        inference_record.proposal_scale= proposal_scale
        inference_record.sigma_signal = sigma_signal

        
        
        
        try:
            run_lengths = self.run_lengths
        except AttributeError:
            self.set_run_lengths([500]*self.nLevels)
            run_lengths = self.run_lengths
#            raise Exception("self.set_run_lengths was not called yet")
        wlp=self.wlp
        for i in range(nLevels):   
#        for i in range(nLevels+5): 
        
            if i<nLevels:
                level=i
                if level == 0:
                    theta = tw.ms.L_cpa_space[level].get_zeros_theta()
                else:    
                    theta_fine = tw.ms.L_cpa_space[level].get_zeros_theta()
                    tw.ms.propogate_theta_coarse2fine(theta_coarse=theta,theta_fine=theta_fine)
                    theta = theta_fine
                _sigma_signal = sigma_signal    
            else:                
                _sigma_signal  *= 0.9
                                                         
             
            print '-'*10 ,'level',level,  '-'*10
            cpa_space = tw.ms.L_cpa_space[level]
            print cpa_space
            
           
            data = {'src':self.src,'transformed':self.transformed,
                    'signal':self.signal}
            
            
            if use_prior:
                lp_func = LP(ms=tw.ms,msp=tw.msp,level=level,SDLP=SDLP,                                          
                                                   required={})
                                                                  
            else:
                lp_func = None         
                                       
            sampler = Metropolis(ll_func= LL(
                                 ms=tw.ms,level=level,SDLL=self.SDLL,
                                 data=data,
                                 required={'sigma_signal':_sigma_signal,
                                           'params_flow_int':tw.params_flow_int_coarse,
                                           'interp_type_for_ll':interp_type_for_ll} ),
                                 proposal=Proposal(ms=tw.ms,msp=tw.msp,level=level,
                                                   scale=proposal_scale,
                                                   use_local=use_local,
                                                   scale_local_proposal=scale_local_proposal),
#                                 proposal=ProposalSphericalGaussian(ms=tw.ms,
#                                                                    level=level,
#                                                   scale=0.1 / (1.2**level) 
#                                                   scale=0.01,
        #                                          scale=0.1 / (1.1**level)                   
        #                                           scale=0.1 
#                                                   ) ,
                                                   lp_func=lp_func,
                                        wlp=wlp
                                        )
               
            sampler.set_theta(theta)
            run_length = run_lengths[level]
            sampler.run(run_length)
            theta = sampler.theta_current # prepare for next iteration
            
            inference_record.steps.append(sampler.get_record())
            
            if dispOn:
                if i >= nLevels-1 or 1:
                    
                    plt.figure(i+1)
                    plt.clf()
                    self.disp(sampler=sampler,
                             interp_type_during_visualization=interp_type_during_visualization)               

                 
         
        inference_record.theta = theta.copy()  
        steps = inference_record.steps
        nAccepted =  [ step.nAccepted  for step in steps]
        run_lengths =  [ step.N  for step in steps]
        times = [ step.runs[0].time for step in steps]
        total_time = sum(times) 

        print "run_lengths:",run_lengths
        print "nAccepted:",nAccepted
        
        print 'times:'
        print_iterable(times)
        print 'time total:',total_time  
        
         




        return theta.copy(),inference_record
Esempio n. 18
0
def example(base=[5],
            scale_spatial=100,
            nLevels=2,
            zero_v_across_bdry=[1],
            use_local_basis=True):
    nPtsDense = 10000
    tw = TransformWrapper(nCols=100,
                          nLevels=nLevels,
                          base=base,
                          scale_spatial=scale_spatial,
                          nPtsDense=nPtsDense,
                          zero_v_across_bdry=zero_v_across_bdry)

    print_iterable(tw.ms.L_cpa_space)

    seed = 0
    np.random.seed(seed)

    for level in range(tw.ms.nLevels):
        cpa_space = tw.ms.L_cpa_space[level]
        Avees = cpa_space.Avees
        velTess = cpa_space.zeros_velTess()

        if use_local_basis:
            if 0:
                tw.sample_gaussian_velTess(level, Avees, velTess, mu=None)
                Avees *= 0.001
                velTess *= 0.001
            else:

                if not zero_v_across_bdry[0]:
                    velTess[:] = 10 * np.random.standard_normal(velTess.shape)
                    cpa_space.velTess2Avees(velTess=velTess, Avees=Avees)
                else:
                    velTess[1:-1] = 10 * np.random.standard_normal(
                        velTess[1:-1].shape)
                    cpa_space.velTess2Avees(velTess=velTess, Avees=Avees)

            cpa_space.velTess2Avees(velTess=velTess, Avees=Avees)

        else:
            theta = cpa_space.get_zeros_theta()
            tw.sample_gaussian(level, Avees, theta, mu=None)
            #            theta/=10
            cpa_space.theta2Avees(theta=theta, Avees=Avees)

        # This step is important and must be done
        # before are trying to "use" the new values of
        # the (vectorized) A's.
        tw.update_pat_from_Avees(Avees, level)

        pts_src = tw.x_dense

        tw.calc_v(level=level, pts=pts_src, v=tw.v_dense)
        tw.v_dense.gpu2cpu()

        pts_fwd = CpuGpuArray.zeros_like(
            pts_src)  # Create a buffer for the output

        tw.calc_T_fwd(pts_src, pts_fwd, level=level)
        pts_fwd.gpu2cpu()

        pts_inv = CpuGpuArray.zeros_like(
            pts_src)  # Create a buffer for the output
        tw.calc_T_inv(pts_src, pts_inv, level=level)
        pts_inv.gpu2cpu()

        plt.figure(level)
        plt.clf()
        interval = pts_src.cpu  # interval doesn't have to be pts_src.cpu
        Visualize.simple(tw.x_dense,
                         tw.v_dense,
                         interval,
                         pts_src,
                         transformed_fwd=pts_fwd,
                         transformed_inv=pts_inv,
                         cpa_space=cpa_space)

    return tw