7f67c4ba6e955bef7d5645b8317a493281881b3d,dscribe/descriptors/ewaldsummatrix.py,EwaldSumMatrix,create,#EwaldSumMatrix#Any#Any#Any#Any#Any#Any#Any#Any#,65

Before Change

        // Here we precalculate the size for each job to preallocate memory.
        if self._flatten:
            k, m = divmod(n_samples, n_jobs)
            jobs = (inp[i * k + min(i, m):(i + 1) * k + min(i + 1, m)] for i in range(n_jobs))
            output_sizes = [len(job) for job in jobs]
        else:
            output_sizes = None

After Change

        inp = [(i_sys, i_accuracy, i_w, i_rcut, i_gcut, i_a) for i_sys, i_accuracy, i_w, i_rcut, i_gcut, i_a in zip(system, accuracy, w, rcut, gcut, a)]

        // Determine if the outputs have a fixed size 
        n_features = self.get_number_of_features()
        if self._flatten:
            static_size = [n_features]
        elif self.permutation == "eigenspectrum":
            static_size = [self.n_atoms_max]
        else:
            static_size = [self.n_atoms_max, self.n_atoms_max]

        // Create in parallel
        output = self.create_parallel(inp, self.create_single, n_jobs, static_size, verbose=verbose)

        return output

Instances