070003f70129fd8dd88364df6d2ae64c1d2a35f8,spynnaker/pyNN/models/neural_projections/connectors/fixed_number_post_connector.py,FixedNumberPostConnector,_get_post_neurons,#FixedNumberPostConnector#,28

Before Change

        return self._get_delay_variance(self._delays, None)

    def _get_post_neurons(self):
        if self._post_neurons is None:
            n = 0
            while (n < self._post_n):
                permutation = numpy.arange(self._n_post_neurons)
                for i in range(0, self._n_post_neurons - 1):
                    j = int(self._rng.next(
                        n=1, distribution="uniform",
                        parameters=[0, self._n_post_neurons]))
                    (permutation[i], permutation[j]) = (
                        permutation[j], permutation[i])
                n += self._n_post_neurons
                if self._post_neurons is None:
                    self._post_neurons = permutation
                else:
                    self._post_neurons = numpy.append(
                        self._post_neurons, permutation)
            self._post_neurons = self._post_neurons[:self._post_n]
            self._post_neurons.sort()
        return self._post_neurons

    def _post_neurons_in_slice(self, post_vertex_slice):
        post_neurons = self._get_post_neurons()

After Change

        // Loop over all the pre neurons
        for m in range(0, self._n_pre_neurons):
            if self._post_neurons[m] is None:
                self._post_neurons[m] = numpy.random.choice(
                    self._n_post_neurons, self._n_post, False)
                self._post_neurons[m].sort()

                // This looks nice but it doesn"t work with PyNN 0.9 ?
//                 n = 0

Instances