如何解决Iris Builder生成的GCS非全连接问题？

我使用Iris Builder和部分种子状态创建广义构型空间（GCS）时，有时会得到非全连接的图。我当前的解决方案是找出断开的节点，为每个断开节点遍历图中已连接部分的节点，计算该节点与已连接节点中心的几何均值，将其作为下一区域的种子，重复此过程直至图全连接：

def bridge(self):
        max_bridge_per_region = 3
        disconnected_regions = set()
        give_up_regions = set()
        while disconnected_regions != (
                disconnected_regions := (set(self._get_disconnected_regions()) - give_up_regions)):
            log.info(f"disconnnected regions are {len(disconnected_regions)} many")
            if not disconnected_regions:
                log.info(f"bridging is done, had to give up on {give_up_regions}")
                return
            region = list(disconnected_regions)[0]
            connected_regions = set(self.iris_regions.values()) - (disconnected_regions | give_up_regions)
            bridge_count = 0
            for cr in connected_regions:
                region_q_sample = region.MaybeGetFeasiblePoint()
                cr_q_sample = cr.MaybeGetFeasiblePoint()
                bridge_seed = np.mean([region_q_sample, cr_q_sample], axis=0)
                Log.info(f"bridge seed is {bridge_seed}")
                try:
                    self.GenerateRegion(f"bridge_{uuid.uuid4().hex}", bridge_seed)
                    bridge_count += 1
                    if bridge_count > max_bridge_per_region:
                        break
                except Exception as e:
                    log.error(f"error in generate: {e}")
                    continue
            if bridge_count == 0:
                log.info(f"giving up on region {region}")
                give_up_regions.add(region)

这种朴素方法虽能生效，但速度极慢！请问是否有更标准的方法，可在现有区域无交集导致图断开时创建这类“桥接”区域？

内容的提问来源于stack exchange，提问作者frouhi