[optimal, gradient_V, init, time] = cmb_estimation(network, q_info, bounds, data, prior, preposterior, pp, V, cmb_options) gradient_V Gradient of enzyme posterior loss with respect to fluxes (currently not computed) -------------------------------------------------------- Initial guesses for X and q (where preposterior q means: using prior (and possibly data) for q only X_init = argmin_X sum(sum( [ [X - preposterior.X.mean] ./ preposterior.X.std ] .^2)); WITHIN allowed polytopes! q_init = argmin_X sum(sum( [ [q - preposterior.q.mean] ./ preposterior.q.std ] .^2); WITHIN allowed ranges! -------------------------------------------------------- choice of initial values depends on cmb_options.initial_values_variant: 'average_sample', 'preposterior_mode', 'random' 'true_values' 'given_values'