Changeset 165 for trunk/examples/retina/benchmark_noise.py

Timestamp:

05/07/08 17:56:33 (4 years ago)

Author:

LaurentPerrinet

Message:

CodeJam? #2: clean-up of the NeuroTools package / removing benchmark.py and everything that depends on it (if you need it, find it in branches/working_concept2) / examples on single neuron are fine, still some work to do on retina

Files:

• trunk/examples/retina/benchmark_noise.py (modified) (1 diff)

trunk/examples/retina/benchmark_noise.py

@@ -15 +15 @@
 import os, datetime, numpy
 import retina as model
-import NeuroTools.benchmark as benchmark
+
+from NeuroTools.parameters import *
+
+# this is not mandatory but just a "easy_install progressbar" away
+# else remove all corresponding lines in this code...
+import progressbar # see http://projects.scipy.org/pipermail/scipy-dev/2008-January/008200.html
+
+p = ParameterSet({})
+N, N_exp_noise = 10, 22
+p.noise_std = ParameterRange(list(10.**(numpy.linspace(-.50,1.,N_exp_noise))))
+retina = model.Retina(N)
+retina.params['snr'] = 0
+
+results = shelve.open('results/benchmark_noise')
+try:
+    CRF = results['CRF']
+except:
+
+    # calculates the dimension of the parameter space
+    results_dim, results_label = p.parameter_space_dimension_labels()
+
+    # creates results array with size of parameter space dimension
+
+    CRF = numpy.empty(results_dim)
+
+    pbar=progressbar.ProgressBar(widgets=["calculating", " ", progressbar.Percentage(), ' ',
+            progressbar.Bar(), ' ', progressbar.ETA()], maxval=numpy.prod(results_dim))
+    for i_exp,experiment in enumerate(p.iter_inner()):
+        params = retina.params
+        params.update(experiment) # updates what changed in the dictionary
+        # simulate the experiment and get its data
+        data = retina.run(params,verbose=False)
+        # calculating the index in the parameter space
+        index = p.parameter_space_index(experiment)
+        # put the data at the right position in the results array
+        CRF[index] = data.mean_rate()#
+        pbar.update(i_exp)
+
+    results['CRF'] = CRF
+    pbar.finish()
+
+results.close()
+
+from NeuroTools.plotting import pylab_params
+
+""" Figure 1
+
+Prints to a figure the mean firing rate for the output (ON and OFF) as a function of the different parameter values. It's similar to a CRF function.
+
+TODO put standard deviation of activity, print CV
+
+"""
 
 
-def show(benchmark):
-    from NeuroTools.plotting import pylab_params
+params = retina.params
+N, simtime = params['N'], params['simtime']
 
-    """ Figure 1
+import pylab
 
-    Prints to a figure the mean firing rate for the output (ON and OFF) as a function of the different parameter values. It's similar to a CRF function.
+#pylab.rcParams.update(pylab_params(fig_width_pt =497.9/5, ratio = 1/1.2))
+pylab.figure(num = 1)#, dpi=200, facecolor='w', edgecolor='k')
 
-    TODO put standard deviation of activity, print CV
+#Lmargin, Rmargin, dmargin, umargin = 0.15, 0.05, 0.15,  0.05
+#pylab.axes([Lmargin, dmargin , 1.0 - Rmargin- Lmargin,1.0-umargin-dmargin]) # [left, bottom, width, height]
 
-    """
+idx_ = numpy.argsort(noise_std) # extract lines
+pylab.plot(noise_std[idx_],CRF[idx_],'go-', label='line 1', linewidth=2)
+pylab.ylabel('Firing Frequency (Hz)')
+pylab.xlabel('Noise amplitude')
+#pylab.axis('tight')
+pylab.savefig('results/benchmark_noise.pdf')#, dpi=300) #
+#command = 'epstopdf  "%s" "%s"'% (pdffile, psfile)
+#os.system(command)
 
-
-    params = benchmark.get('params')
-
-    N, simtime = params['N'], params['simtime']
-    experiments = benchmark.list_experiments()
-    noise_std, CRF = numpy.zeros(len(experiments)), numpy.zeros(len(experiments))
-    for count, experiment in enumerate(experiments):
-        out_ON_DATA = benchmark.get('out',experiment)['out_ON_DATA']
-        noise_std[count] = benchmark.get('experiments')[experiment]['noise_std']
-        CRF[count] = out_ON_DATA.mean_rate()
-
-
-    import pylab
-
-    #pylab.rcParams.update(pylab_params(fig_width_pt =497.9/5, ratio = 1/1.2))
-    pylab.figure(num = 1)#, dpi=200, facecolor='w', edgecolor='k')
-
-    #Lmargin, Rmargin, dmargin, umargin = 0.15, 0.05, 0.15,  0.05
-    #pylab.axes([Lmargin, dmargin , 1.0 - Rmargin- Lmargin,1.0-umargin-dmargin]) # [left, bottom, width, height]
-
-    idx_ = numpy.argsort(noise_std) # extract lines
-    pylab.plot(noise_std[idx_],CRF[idx_],'go-', label='line 1', linewidth=2)
-    pylab.ylabel('Firing Frequency (Hz)')
-    pylab.xlabel('Noise amplitude')
-    #pylab.axis('tight')
-    pylab.savefig(benchmark.directory + '/benchmark_noise.pdf')#, dpi=300) #
-    #command = 'epstopdf  "%s" "%s"'% (pdffile, psfile)
-    #os.system(command)
-
-    #TODO: make a plot showing it's the right point process with a histogram
-
-
-if __name__ == '__main__':
-
-    # create a Retina Model object that contains a method to init, to run, ...
-    ret = model.Retina(10)
-
-    tag = 'test'
-
-    name = 'results/' + tag + '_benchmark_noise'
-
-    # create the list of experiments as all possibilities accross parameter vectors
-    span, N_exp = 5, 2*7+1
-    #range = 10.**(numpy.linspace(-span,span,N_exp))
-    range = numpy.linspace(1/span,span,N_exp)
-    run = benchmark.get_experiment_dict({'noise_std': ret.params['noise_std']* range, 'snr' : [ 0.0 ]})
-
-    # create benchmark (or open it if it exists)
-    B = benchmark.Benchmark(name,ret,run)
-# there are many ways to get data from the benchmark. see benchmark.py
-    B.run_simulations()
-
-    show(B)
-    #B.reset_all() # erase experimental data
+#TODO: make a plot showing it's the right point process with a histogram