116   State current_state(num_parameters, num_effective_parameters);  in Minimize()  local
134   if (!Evaluate(evaluator, x, &current_state, &summary->message)) {  in Minimize()
142   summary->initial_cost = current_state.cost + summary->fixed_cost;  in Minimize()
143   iteration_summary.cost = current_state.cost + summary->fixed_cost;  in Minimize()
145   iteration_summary.gradient_max_norm = current_state.gradient_max_norm;  in Minimize()
146   iteration_summary.gradient_norm = sqrt(current_state.gradient_squared_norm);  in Minimize()
238       current_state.search_direction = -current_state.gradient;  in Minimize()
242           current_state,  in Minimize()
243           &current_state.search_direction);  in Minimize()
278       current_state.search_direction = -current_state.gradient;  in Minimize()
281     line_search_function.Init(x, current_state.search_direction);  in Minimize()
282     current_state.directional_derivative =  in Minimize()
283         current_state.gradient.dot(current_state.search_direction);  in Minimize()
293         ? min(1.0, 1.0 / current_state.gradient_max_norm)  in Minimize()
294         : min(1.0, 2.0 * (current_state.cost - previous_state.cost) /  in Minimize()
295               current_state.directional_derivative);  in Minimize()
304                        initial_step_size, current_state.directional_derivative,  in Minimize()
305                        (current_state.cost - previous_state.cost));  in Minimize()
312                         current_state.cost,  in Minimize()
313                         current_state.directional_derivative,  in Minimize()
321                        initial_step_size, current_state.cost,  in Minimize()
322                        current_state.directional_derivative);  in Minimize()
328     current_state.step_size = line_search_summary.optimal_step_size;  in Minimize()
329     delta = current_state.step_size * current_state.search_direction;  in Minimize()
331     previous_state = current_state;  in Minimize()
344                          &current_state,  in Minimize()
357     iteration_summary.gradient_max_norm = current_state.gradient_max_norm;  in Minimize()
358     iteration_summary.gradient_norm = sqrt(current_state.gradient_squared_norm);  in Minimize()
359     iteration_summary.cost_change = previous_state.cost - current_state.cost;  in Minimize()
360     iteration_summary.cost = current_state.cost + summary->fixed_cost;  in Minimize()
365     iteration_summary.step_size =  current_state.step_size;  in Minimize()