Changeset 101

Timestamp:

01/15/08 17:44:22 (1 year ago)

Author:

LaurentPerrinet

Message:

Finished examples of use of NeuroTools: retina / renamd SpikeTrain.mean_firing_rate to SpikeTrain.mean_rate for consistency with SpikeList.mean_rate / bugfix in spikes.py for the definition of a range in SpikeTrain.mean_rate

Files:

• trunk/benchmark.py (modified) (2 diffs)
• trunk/examples/retina/benchmark_linear.py (modified) (5 diffs)
• trunk/examples/retina/benchmark_noise.py (modified) (4 diffs)
• trunk/examples/retina/benchmark_retina.py (modified) (7 diffs)
• trunk/examples/retina/make_all.py (modified) (4 diffs)
• trunk/examples/retina/retina.py (modified) (6 diffs)
• trunk/spikes.py (modified) (8 diffs)

trunk/benchmark.py

@@ -203 +203 @@
-        return unfinished_experiments
-
-
+, type='full'
-        """ Removes all experiments files
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
-
-
@@ -230 +238 @@
-            params = self.get('params')
-            print "Running Experiment " + experiment + " ..."
+            print 'Changing :', self.get('experiments')[experiment]
-            # updates params with the changed parameers from experiment
-            params.update(self.get('experiments')[experiment])  # change some parameters

trunk/examples/retina/benchmark_linear.py

@@ -17 +17 @@
-import retina as model
-import NeuroTools.benchmark as benchmark
-
-
-
-
@@ -43 +41 @@
-
-    N, N_ret, simtime = params['N'], params['N_ret'], params['simtime']
-N_smooth = 100 # how many time bins
+t_smooth = 100. #  smooting time (ms)
-
-    for experiment in experiment_list:
@@ -50 +48 @@
-        except:
-            print "Loading data for experiment " , experiment
-
-
-
-
-
-
-
-
-
-
+
+
-            # storing
-            benchmark.put('firing_rate',{
-temporal_O
-##                    'temporal_OFF' : temporal_OFF
-lower_edges
+sum(out_ON.firing_rate(t_smooth))/
+                    #'temporal_OFF' : out_OFF.firing_rate(t_smooth)
+out_ON.time_axis(t_smooth)
-                    'firing_rate_ok' : True # a flag to do it once
-                      }, experiment )
@@ -89 +79 @@
-    #pylab.legend()
-
-ng
+df
-
-
-if __name__ == '__main__':
-
-    # not working tag= '07-02-16'
-    # tag= '07-02-14'
-    # tag= '07-02-23'
-    tag= '07-02-27'
-    tag = '07-03-08'
-    tag = '07-05-09'
-    tag = 'test'
-
@@ -106 +90 @@
-
-    ret = model.Retina(21)
-
-
-
-
-
+
+
+
+
+
+
-    B = benchmark.Benchmark(filename,ret,run)
-
-    B.run_simulations()
-    show(B)
-
+

trunk/examples/retina/benchmark_noise.py

@@ -35 +35 @@
-    noise_std, CRF = numpy.zeros(len(experiments)), numpy.zeros(len(experiments))
-    for count, experiment in enumerate(experiments):
-#.as_list_id_list_time()
+
-        noise_std[count] = benchmark.get('experiments')[experiment]['noise_std']
-#[0]) / N / N / simtime * 1000.0
+
-
-
@@ -48 +48 @@
-    #pylab.axes([Lmargin, dmargin , 1.0 - Rmargin- Lmargin,1.0-umargin-dmargin]) # [left, bottom, width, height]
-
-    #pylab.subplot(111)
-    idx_ = numpy.argsort(noise_std) # extract lines
-    pylab.plot(noise_std[idx_],CRF[idx_],'go-', label='line 1', linewidth=2)
@@ -65 +64 @@
-    # create a Retina Model object that contains a method to init, to run, ...
-    ret = model.Retina(10)
-    ret.params['simtime'] = 4000*0.1
-
-
-
-
-
-
-
-
+
-
-    name = 'results/' + tag + '_benchmark_noise'
-
-    # create the list of experiments as all possibilities accross parameter vectors
-
+
+
+
+
-
-    # create benchmark (or open it if it exists)
@@ -86 +81 @@
-
-    show(B)
-
+

trunk/examples/retina/benchmark_retina.py

@@ -27 +27 @@
-def show(benchmark):
-    from NeuroTools.plotting import pylab_params
+    from numpy import zeros, where, arange
-
-
@@ -32 +33 @@
-
-    N, N_ret, simtime, dt = params['N'], params['N_ret'], params['simtime'], params['dt']
-
-
-
-
-
-
-
+
+
+
+
+
+
-    r2 = x**2 + y**2
-
+
+
+
-    #kernel = amplitude * (amplitude > 0)
-    #kernel /=  numpy.sum(numpy.sum(kernel))
@@ -52 +54 @@
-            out_ON = benchmark.get('out',experiment)['out_ON_DATA']
-            out_OFF = benchmark.get('out',experiment)['out_OFF_DATA']
-            out_ON_DATA = out_ON.firing_rate(t_smooth)
-            out_OFF_DATA = out_OFF.firing_rate(t_smooth)
-
-            ##temporal_ON= numpy.sum(numpy.sum(out_ON_spikematrix, axis=0), axis=0)
-            temporal_ON , lower_edges = numpy.histogram(out_ON_DATA[1],
-                                bins=N_smooth, range=(out_ON.t_start,out_ON.t_stop))
-            ##temporal_OFF= numpy.sum(numpy.sum(out_OFF_spikematrix, axis=0), axis=0)
-            temporal_OFF , lower_edges = numpy.histogram(out_OFF_DATA[1],
-                                bins=N_smooth, range=(out_ON.t_start,out_ON.t_stop))
-
-
-            # TODO the range argument may serve to select a time window
-            map_spatial_ON , dump = numpy.histogram(out_ON_DATA[0],bins=N**2, range=(0,N**2))
-            map_spatial_OFF, dump = numpy.histogram(out_OFF_DATA[0],bins=N**2, range=(0,N**2))
-
-            benchmark.put('firing_rate',{
-    'temporal_ON' : temporal_O
-    'temporal_OFF' : temporal_OFF
-    'lower_edges' : lower_edges
-    'map_spatial_ON' : numpy.reshape(map_spatial_ON,(N,N)
-    'map_spatial_OFF' : numpy.reshape(map_spatial_OFF,(N,N)
-    
+'temporal_ON' : sum(out_ON.firing_rate(t_smooth))/
+'temporal_OFF' : sum(out_OFF.firing_rate(t_smooth))/N
+'lower_edges' : out_OFF.time_axis(t_smooth)
+'map_spatial_ON' : out_ON.mean_rates(t_start=simtime/4.,t_stop=3*simtime/4.
+'map_spatial_OFF' : out_OFF.mean_rates(t_start=simtime/4.,t_stop=3*simtime/4.
+
-
-    pylab.close('all')
@@ -91 +79 @@
-        # mean activity accross kernelseeds as a function of SNR
-        snr_list = benchmark.list_param('snr')
-
-        for snr in snr_list:
- 0, 0, 0, 0
+zeros(N_smooth), zeros(N_smooth), zeros(N), zeros(N)
-            seeds_list = benchmark.list_where('snr',snr)
+            n_seeds = len(seeds_list)
-            for experiment in seeds_list: # TODO sum on dictionaries?
-                run = benchmark.get('firing_rate',experiment)
-
-
-
-
-
-
+
+
+
+
+
-            pylab.subplot(221)
-
+/ n_seeds
-            pylab.subplot(223)
-
+/ n_seeds
-            pylab.subplot(222)
-space ,  numpy.sum(map_spatial_ON[N/2-N_ret/2:N/2+N_ret/2,:], axis=0) * N / N_ret
+r , map_spatial_ON ,'bo'
-            pylab.subplot(224)
-space ,  numpy.sum(map_spatial_OFF[N/2-N_ret/2:N/2+N_ret/2,:], axis=0) * N / N_ret
+r ,  map_spatial_OFF , 'ro'
-
-        pylab.subplot(221)
@@ -131 +118 @@
-        pylab.subplot(224)
-        #pylab.title('spatial distribution OFF',fontsize ='small')
-space
+distance from center
-
-        pylab.savefig(name) #
@@ -176 +163 @@
-        pylab.figure(num = 3, dpi=dpi, facecolor='w', edgecolor='k')
-
-
-
-
-matshow(benchmark.get('params')['amplitude']
-
-
-
+#
+#
+
+scatter(x,y,c=benchmark.get('params')['amplitude'],faceted=False
+
+#
+#
-        pylab.savefig(name)
-
@@ -201 +188 @@
-    # create a Retina Model object that contains a method to init, to run, ...
-    ret = model.Retina(10)
-
-
-
+
-    ret.params['description']=  'Testing Benchmark Retina with ' + ret.params['description'],
-
-    # create the list of experiments as all possibilities accross parameter vectors
-    N_exp_snr, N_exp_seeds = 5 , 4
-10.**(numpy.linspace(-.50,.75,N_exp_snr)
+numpy.linspace(0.1,2.0,N_exp_snr
-    seeds =  [43210987 + k for k in range(N_exp_seeds)]
-    run = benchmark.get_experiment_dict({'snr':snr,'kernelseed':seeds})

trunk/examples/retina/make_all.py

@@ -5 +5 @@
-===========
-
-0
+1
-
-Laurent Perrinet, INCM, CNRS
@@ -18 +18 @@
-
-# tag list
-tag = '07-02-21'
-tag = '07-03-08' # worked fine for D25
-tag = '07-04-12'
-tag = '07-04-08' # new with multiple seeds for CNS07
-tag = '07-05-14' # no
-tag= '07-04-08' # this worked nice and will be a reference for D25
-tag = '07-05-17' # going OO
-tag = '07-07-12' # testing the new shelve interface
-tag = '08-01-04' # demo for Debrecen
-
@@ -31 +23 @@
-from NeuroTools.benchmark import *
-from numpy import linspace, ones
+
+
-### testing the noise response
-import benchmark_noise
@@ -42 +36 @@
-B.run_simulations()
-benchmark_noise.show(B)
-
+
-### testing the linear response
-import benchmark_linear

trunk/examples/retina/retina.py

@@ -104 +104 @@
-            'simtime'    : 40000*0.1,      # float; (ms)
-            'syn_delay'  : 1.0,         # float; (ms)
-4
-3
+5
+2
-            'weight'     : 1.0, #
-            #'threads' : 2, 'kernelseeds' : [43210987, 394780234],      # array with one element per thread
@@ -116 +116 @@
-
-        # retinal neurons parameters
-
+
+
-        'C':289.5, 'V_reversal_E': 0.0, 'V_reversal_I': -75.0, 'TauSyn_E':1.5,
-        'TauSyn_I': 10.0, 'V_reversal_sfa': -70.0, 'q_sfa': 0., #14.48,
@@ -152 +153 @@
-        sim.Timer.start() # start timer on construction
-        sim.setup(timestep=params['dt'],max_delay=params['syn_delay'])
-        #sim.pynest.setDict([0],{'threads' : params['threads']})
-        sim.setRNGseeds([params['kernelseed']])
-
@@ -159 +159 @@
-        phr_ON = sim.Population(N,'dc_generator')
-        phr_OFF  = sim.Population(N,'dc_generator')
-
+
+
-            phr.set({ 'start' : params['simtime']/4, 'stop' : params['simtime']/4*3})
-
@@ -166 +167 @@
-        noise_OFF = sim.Population(N,'noise_generator',{'mean':0.,'std':params['noise_std']})
-
-        # TODO LUP : here introduce a tranformation of the image (on a grid) by a DOG and then on the specific location of the neuron
-
-
-        phr_ON.tset('amplitude', params['amplitude'] *  params['snr'])
-        phr_OFF.tset('amplitude', - params['amplitude']  * params['snr'])
-
-        # target ON and OFF populations (what about a tridimensional Population?)
@@ -191 +187 @@
-        ##    >>> p1.randomInit(vinit_distr)
-
-
-
+
+
+
+
+
-        retina_proj_ON = sim.Projection(phr_ON, out_ON, 'oneToOne')
-        retina_proj_ON.setWeights(params['weight'])#

trunk/spikes.py

@@ -24 +24 @@
-    >>> s1.isi()
-    array([ 0.1,  0.1,  0.3])
-firing_
+
-    8.0
-    >>> s1.cv_isi()
@@ -145 +145 @@
-
-    # Returns the mean firing rate of the SpikeTrain
-firing_
+
-        """ Mean firing rate between t_start and t_stop in Hz
-
@@ -151 +151 @@
-        """
-        if (t_start==None) & (t_start==None):
+            t_start=self.t_start
+            t_stop=self.t_stop
-            idx = self.spike_times
-        else:
-            if t_start==None: t_start=self.t_start
-art
+op
-            idx = numpy.where((self.spike_times >= t_start) & (self.spike_times <= t_stop))[0]
-        return 1000.*len(idx)/(t_stop-t_start)
@@ -219 +221 @@
-        are calculated from `self.t_start` and `self.t_stop`.
-        If `normalized` is True, the bin values are scaled so as to represent
-
+ it's the number of spikes per bin
-        """
-        nbins = self.time_axis(time_bin)
@@ -481 +483 @@
-
-
-mean_firing
+SpikeTrain.mean
-        """
-        rates = []
-        for id in self.id_list:
-firing_
+
-
-        return rates
@@ -492 +494 @@
-    # If display is True, then it plots the distribution of the rates
-    def rate_distribution(self, nbins=25, normalize=True, display=False):
-
-
-
+
+
+
+
-        if not display:
-            return rates
@@ -560 +563 @@
-        if isinstance(dims, tuple) or isinstance(dims, list):
-            activity_map = numpy.zeros(dims,float)
+            rates = self.mean_rates()
-            for id in self.id_list:
-                position = self.id2position(id, dims)
-self.spiketrains[id].mean_firing_rate()
+rates[id]
-            if not display:
-                return activity_map
@@ -575 +579 @@
-            if not len(self.id_list) == len(dims[0]):
-                raise Exception("Error, the number of positions does not match the number of cells in the SpikeList")
-[]
-
-
+self.mean_rates()
+#
+#
-            if not display:
-                return rates