什么是“m桥技术”？用于分区二叉树以进行并行处理？
它是如何运作的？请用英语或伪代码详细解释，以便我可以用任何语言实现。
本文提到并简要描述了这一点： http://citeseerx. ist.psu.edu/viewdoc/download?doi=10.1.1.4.3643&rep=rep1&type=pdf
但那里没有足够的细节来实现我自己。 （图 2b 中的权重似乎是错误的？并且不清楚他们如何决定在图 2c 中进行切割的位置。）
我还查找了原始源论文（http://www-2.cs.cmu.edu/~glmiller/Publications/Papers/ReMiMo93.pdf)但也无法从那里弄清楚。
是否有更好的算法可以满足相同的需求？具体来说，有什么可以保证分区树“几乎相同大小”（但更多）？该论文建议 m-bridge 保证分区树不会大于 4n/p，如果您只有 4 个处理器，则这并不能保证！
How does it work? Please explain in enough detail in English or pseudocode so that I can implement in any language.
It is mentioned and briefly described in this paper:
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.4.3643&rep=rep1&type=pdf
but there isn't enough detail there to implement myself. (the weights in Fig2b seem wrong? and it's not clear how they decided where to make the cuts in Fig 2c.)
I also looked up the original source paper (http://www-2.cs.cmu.edu/~glmiller/Publications/Papers/ReMiMo93.pdf) but couldn't figure it out from there either.
Are there better algorithms out there that fill the same need? Specifically, anything that can guarantee partitioned trees that are "almost the same size" (but more so)? The paper suggests m-bridge guarantees no partitioned tree to be bigger than 4n/p, which is not much of a guarantee if you only have 4 processors!
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（1）
要进行一次切割：
要进行多次砍伐，请反复砍伐剩余的最大的树。
To make one cut:
To make many cuts, repeatedly cut the largest tree remaining.