如何增强 lpsolve R 优化解决方案以在 hadoop 集群上运行？

【问题标题】：How to augment lpsolve R optimization solution to run on a hadoop cluster?如何增强 lpsolve R 优化解决方案以在 hadoop 集群上运行？
【发布时间】：2014-05-17 03:05:15
【问题描述】：

我正在使用 R lpsolve 包来优化我的运输模型。我的代码运行良好，但运行需要很长时间，因为我有大量的节点和路径。我打算在 hadoop 集群上运行我的代码。

请指导我有关我需要对我的代码进行的更改。我认为在 hadoop 集群上运行优化可能是不可能的，因为我们最终可能会得到局部最小值而不是全局最小值。

我在互联网上搜索“lpsolve hadoop”之类的术语，但没有得到任何有用的信息。

请指导我查看我应该查看的材料或示例。 ======================================更新1=========== ============================ 我原来的问题是here。

我进一步简化了问题，我正在解决的当前问题如下。

附上我用excel构建的R代码和输入数据文件。在实际场景中，输入数据文件将使用 SQL 生成，长度超过 30,000 行。

我的输入excel文件：

startlink   endlink link_dsc    lnk_type    cons_type   cost    equality_const  fpc_const   max_const
"source","-","-","0"    "vmi1","MM1","VMI","1"  source_to_VMI   supply  equality supply 0   100 null    null
"vmi1","MM1","VMI","0"  "cust1","MM1","SHIP_CUST_1d_AIR","1"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","1"  "cust1","MM1","SHIP_CUST_1d_AIR","2"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","2"  "cust1","MM1","SHIP_CUST_1d_AIR","3"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","3"  "cust1","MM1","SHIP_CUST_1d_AIR","4"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","4"  "cust1","MM1","SHIP_CUST_1d_AIR","5"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","5"  "cust1","MM1","SHIP_CUST_1d_AIR","6"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","6"  "cust1","MM1","SHIP_CUST_1d_AIR","7"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","7"  "cust1","MM1","SHIP_CUST_1d_AIR","8"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","8"  "cust1","MM1","SHIP_CUST_1d_AIR","9"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","9"  "cust1","MM1","SHIP_CUST_1d_AIR","10"   vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","0"  "vmi1","MM1","VMI","1"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi1","MM1","VMI","1"  "vmi1","MM1","VMI","2"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi1","MM1","VMI","2"  "vmi1","MM1","VMI","3"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi1","MM1","VMI","3"  "vmi1","MM1","VMI","4"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi1","MM1","VMI","4"  "vmi1","MM1","VMI","5"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi1","MM1","VMI","5"  "vmi1","MM1","VMI","6"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi1","MM1","VMI","6"  "vmi1","MM1","VMI","7"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi1","MM1","VMI","7"  "vmi1","MM1","VMI","8"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi1","MM1","VMI","8"  "vmi1","MM1","VMI","9"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi1","MM1","VMI","9"  "vmi1","MM1","VMI","10" vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi1","MM1","VMI","10" "SINK","-","-","100"    vmi_to_sink esacpe  null    100 null    null    null
"cust1","MM1","SHIP_CUST_1d_AIR","1"    "cust1","MM1","CUST","1"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_1d_AIR","2"    "cust1","MM1","CUST","2"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_1d_AIR","3"    "cust1","MM1","CUST","3"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_1d_AIR","4"    "cust1","MM1","CUST","4"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_1d_AIR","5"    "cust1","MM1","CUST","5"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_1d_AIR","6"    "cust1","MM1","CUST","6"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_1d_AIR","7"    "cust1","MM1","CUST","7"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_1d_AIR","8"    "cust1","MM1","CUST","8"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_1d_AIR","9"    "cust1","MM1","CUST","9"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_1d_AIR","10"   "cust1","MM1","CUST","10"   shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","CUST","4"    "SINK","-","-","100"    cust_to_sink    flow    demand  0   null    null    50
"vmi1","MM1","VMI","0"  "cust1","MM1","SHIP_CUST_2d_AIR","2"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","1"  "cust1","MM1","SHIP_CUST_2d_AIR","3"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","2"  "cust1","MM1","SHIP_CUST_2d_AIR","4"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","3"  "cust1","MM1","SHIP_CUST_2d_AIR","5"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","4"  "cust1","MM1","SHIP_CUST_2d_AIR","6"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","5"  "cust1","MM1","SHIP_CUST_2d_AIR","7"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","6"  "cust1","MM1","SHIP_CUST_2d_AIR","8"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","7"  "cust1","MM1","SHIP_CUST_2d_AIR","9"    vmi_to_shipcust flow    null    5   null    null    null
"vmi1","MM1","VMI","8"  "cust1","MM1","SHIP_CUST_2d_AIR","10"   vmi_to_shipcust flow    null    5   null    null    null
"cust1","MM1","SHIP_CUST_2d_AIR","2"    "cust1","MM1","CUST","2"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_2d_AIR","3"    "cust1","MM1","CUST","3"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_2d_AIR","4"    "cust1","MM1","CUST","4"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_2d_AIR","5"    "cust1","MM1","CUST","5"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_2d_AIR","6"    "cust1","MM1","CUST","6"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_2d_AIR","7"    "cust1","MM1","CUST","7"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_2d_AIR","8"    "cust1","MM1","CUST","8"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_2d_AIR","9"    "cust1","MM1","CUST","9"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM1","SHIP_CUST_2d_AIR","10"   "cust1","MM1","CUST","10"   shipcust_to_cust    flow    null    0   null    null    null
"source","-","-","0"    "vmi2","MM2","VMI","2"  source_to_VMI   supply  equality supply 0   50  null    null
"vmi2","MM2","VMI","0"  "cust1","MM2","SHIP_CUST_1d_AIR","1"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","1"  "cust1","MM2","SHIP_CUST_1d_AIR","2"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","2"  "cust1","MM2","SHIP_CUST_1d_AIR","3"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","3"  "cust1","MM2","SHIP_CUST_1d_AIR","4"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","4"  "cust1","MM2","SHIP_CUST_1d_AIR","5"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","5"  "cust1","MM2","SHIP_CUST_1d_AIR","6"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","6"  "cust1","MM2","SHIP_CUST_1d_AIR","7"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","7"  "cust1","MM2","SHIP_CUST_1d_AIR","8"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","8"  "cust1","MM2","SHIP_CUST_1d_AIR","9"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","9"  "cust1","MM2","SHIP_CUST_1d_AIR","10"   vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","0"  "vmi2","MM2","VMI","1"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi2","MM2","VMI","1"  "vmi2","MM2","VMI","2"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi2","MM2","VMI","2"  "vmi2","MM2","VMI","3"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi2","MM2","VMI","3"  "vmi2","MM2","VMI","4"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi2","MM2","VMI","4"  "vmi2","MM2","VMI","5"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi2","MM2","VMI","5"  "vmi2","MM2","VMI","6"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi2","MM2","VMI","6"  "vmi2","MM2","VMI","7"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi2","MM2","VMI","7"  "vmi2","MM2","VMI","8"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi2","MM2","VMI","8"  "vmi2","MM2","VMI","9"  vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi2","MM2","VMI","9"  "vmi2","MM2","VMI","10" vmi_to_vmi_inv  flow    null    0   null    null    null
"vmi2","MM2","VMI","10" "SINK","-","-","100"    vmi_to_sink esacpe  null    100 null    null    null
"cust1","MM2","SHIP_CUST_1d_AIR","1"    "cust1","MM2","CUST","1"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_1d_AIR","2"    "cust1","MM2","CUST","2"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_1d_AIR","3"    "cust1","MM2","CUST","3"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_1d_AIR","4"    "cust1","MM2","CUST","4"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_1d_AIR","5"    "cust1","MM2","CUST","5"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_1d_AIR","6"    "cust1","MM2","CUST","6"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_1d_AIR","7"    "cust1","MM2","CUST","7"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_1d_AIR","8"    "cust1","MM2","CUST","8"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_1d_AIR","9"    "cust1","MM2","CUST","9"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_1d_AIR","10"   "cust1","MM2","CUST","10"   shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","CUST","9"    "SINK","-","-","100"    cust_to_sink    flow    demand  0   null    null    10
"vmi2","MM2","VMI","0"  "cust1","MM2","SHIP_CUST_2d_AIR","2"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","1"  "cust1","MM2","SHIP_CUST_2d_AIR","3"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","2"  "cust1","MM2","SHIP_CUST_2d_AIR","4"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","3"  "cust1","MM2","SHIP_CUST_2d_AIR","5"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","4"  "cust1","MM2","SHIP_CUST_2d_AIR","6"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","5"  "cust1","MM2","SHIP_CUST_2d_AIR","7"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","6"  "cust1","MM2","SHIP_CUST_2d_AIR","8"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","7"  "cust1","MM2","SHIP_CUST_2d_AIR","9"    vmi_to_shipcust flow    null    1.4 null    null    null
"vmi2","MM2","VMI","8"  "cust1","MM2","SHIP_CUST_2d_AIR","10"   vmi_to_shipcust flow    null    1.4 null    null    null
"cust1","MM2","SHIP_CUST_2d_AIR","2"    "cust1","MM2","CUST","2"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_2d_AIR","3"    "cust1","MM2","CUST","3"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_2d_AIR","4"    "cust1","MM2","CUST","4"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_2d_AIR","5"    "cust1","MM2","CUST","5"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_2d_AIR","6"    "cust1","MM2","CUST","6"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_2d_AIR","7"    "cust1","MM2","CUST","7"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_2d_AIR","8"    "cust1","MM2","CUST","8"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_2d_AIR","9"    "cust1","MM2","CUST","9"    shipcust_to_cust    flow    null    0   null    null    null
"cust1","MM2","SHIP_CUST_2d_AIR","10"   "cust1","MM2","CUST","10"   shipcust_to_cust    flow    null    0   null    null    null
"vmi1","MM1","VMI","0"  "cust3","MM1","SHIP_CUST_1d_AIR","1"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","1"  "cust3","MM1","SHIP_CUST_1d_AIR","2"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","2"  "cust3","MM1","SHIP_CUST_1d_AIR","3"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","3"  "cust3","MM1","SHIP_CUST_1d_AIR","4"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","4"  "cust3","MM1","SHIP_CUST_1d_AIR","5"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","5"  "cust3","MM1","SHIP_CUST_1d_AIR","6"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","6"  "cust3","MM1","SHIP_CUST_1d_AIR","7"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","7"  "cust3","MM1","SHIP_CUST_1d_AIR","8"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","8"  "cust3","MM1","SHIP_CUST_1d_AIR","9"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","9"  "cust3","MM1","SHIP_CUST_1d_AIR","10"   vmi_to_shipcust flow    null    15  null    null    null
"cust3","MM1","SHIP_CUST_1d_AIR","1"    "cust3","MM1","CUST","1"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_1d_AIR","2"    "cust3","MM1","CUST","2"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_1d_AIR","3"    "cust3","MM1","CUST","3"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_1d_AIR","4"    "cust3","MM1","CUST","4"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_1d_AIR","5"    "cust3","MM1","CUST","5"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_1d_AIR","6"    "cust3","MM1","CUST","6"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_1d_AIR","7"    "cust3","MM1","CUST","7"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_1d_AIR","8"    "cust3","MM1","CUST","8"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_1d_AIR","9"    "cust3","MM1","CUST","9"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_1d_AIR","10"   "cust3","MM1","CUST","10"   shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","CUST","6"    "SINK","-","-","100"    cust_to_sink    flow    demand  0   null    null    5
"vmi1","MM1","VMI","0"  "cust3","MM1","SHIP_CUST_2d_AIR","2"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","1"  "cust3","MM1","SHIP_CUST_2d_AIR","3"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","2"  "cust3","MM1","SHIP_CUST_2d_AIR","4"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","3"  "cust3","MM1","SHIP_CUST_2d_AIR","5"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","4"  "cust3","MM1","SHIP_CUST_2d_AIR","6"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","5"  "cust3","MM1","SHIP_CUST_2d_AIR","7"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","6"  "cust3","MM1","SHIP_CUST_2d_AIR","8"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","7"  "cust3","MM1","SHIP_CUST_2d_AIR","9"    vmi_to_shipcust flow    null    15  null    null    null
"vmi1","MM1","VMI","8"  "cust3","MM1","SHIP_CUST_2d_AIR","10"   vmi_to_shipcust flow    null    15  null    null    null
"cust3","MM1","SHIP_CUST_2d_AIR","2"    "cust3","MM1","CUST","2"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_2d_AIR","3"    "cust3","MM1","CUST","3"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_2d_AIR","4"    "cust3","MM1","CUST","4"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_2d_AIR","5"    "cust3","MM1","CUST","5"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_2d_AIR","6"    "cust3","MM1","CUST","6"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_2d_AIR","7"    "cust3","MM1","CUST","7"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_2d_AIR","8"    "cust3","MM1","CUST","8"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_2d_AIR","9"    "cust3","MM1","CUST","9"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM1","SHIP_CUST_2d_AIR","10"   "cust3","MM1","CUST","10"   shipcust_to_cust    flow    null    0   null    null    null
"vmi2","MM2","VMI","0"  "cust3","MM2","SHIP_CUST_1d_AIR","1"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","1"  "cust3","MM2","SHIP_CUST_1d_AIR","2"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","2"  "cust3","MM2","SHIP_CUST_1d_AIR","3"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","3"  "cust3","MM2","SHIP_CUST_1d_AIR","4"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","4"  "cust3","MM2","SHIP_CUST_1d_AIR","5"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","5"  "cust3","MM2","SHIP_CUST_1d_AIR","6"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","6"  "cust3","MM2","SHIP_CUST_1d_AIR","7"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","7"  "cust3","MM2","SHIP_CUST_1d_AIR","8"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","8"  "cust3","MM2","SHIP_CUST_1d_AIR","9"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","9"  "cust3","MM2","SHIP_CUST_1d_AIR","10"   vmi_to_shipcust flow    null    1.8 null    null    null
"cust3","MM2","SHIP_CUST_1d_AIR","1"    "cust3","MM2","CUST","1"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_1d_AIR","2"    "cust3","MM2","CUST","2"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_1d_AIR","3"    "cust3","MM2","CUST","3"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_1d_AIR","4"    "cust3","MM2","CUST","4"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_1d_AIR","5"    "cust3","MM2","CUST","5"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_1d_AIR","6"    "cust3","MM2","CUST","6"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_1d_AIR","7"    "cust3","MM2","CUST","7"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_1d_AIR","8"    "cust3","MM2","CUST","8"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_1d_AIR","9"    "cust3","MM2","CUST","9"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_1d_AIR","10"   "cust3","MM2","CUST","10"   shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","CUST","8"    "SINK","-","-","100"    cust_to_sink    flow    demand  0   null    null    7
"vmi2","MM2","VMI","0"  "cust3","MM2","SHIP_CUST_2d_AIR","2"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","1"  "cust3","MM2","SHIP_CUST_2d_AIR","3"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","2"  "cust3","MM2","SHIP_CUST_2d_AIR","4"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","3"  "cust3","MM2","SHIP_CUST_2d_AIR","5"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","4"  "cust3","MM2","SHIP_CUST_2d_AIR","6"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","5"  "cust3","MM2","SHIP_CUST_2d_AIR","7"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","6"  "cust3","MM2","SHIP_CUST_2d_AIR","8"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","7"  "cust3","MM2","SHIP_CUST_2d_AIR","9"    vmi_to_shipcust flow    null    1.8 null    null    null
"vmi2","MM2","VMI","8"  "cust3","MM2","SHIP_CUST_2d_AIR","10"   vmi_to_shipcust flow    null    1.8 null    null    null
"cust3","MM2","SHIP_CUST_2d_AIR","2"    "cust3","MM2","CUST","2"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_2d_AIR","3"    "cust3","MM2","CUST","3"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_2d_AIR","4"    "cust3","MM2","CUST","4"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_2d_AIR","5"    "cust3","MM2","CUST","5"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_2d_AIR","6"    "cust3","MM2","CUST","6"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_2d_AIR","7"    "cust3","MM2","CUST","7"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_2d_AIR","8"    "cust3","MM2","CUST","8"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_2d_AIR","9"    "cust3","MM2","CUST","9"    shipcust_to_cust    flow    null    0   null    null    null
"cust3","MM2","SHIP_CUST_2d_AIR","10"   "cust3","MM2","CUST","10"   shipcust_to_cust    flow    null    0   null    null    null
"SINK","-","-","100"    "source","-","-","0"    closed_loop flow    null    0   null    null    null

我的 R 代码如下：你必须把上面的 csv 数据粘贴到文件 C:/dumy_network.csv 中。

我的 R 代码可能效率不高，但可以达到目的！

library("lpSolve", lib.loc="C:/Users/njog/Documents/R/win-library/3.0")

#get the data from CSV file
mydata <- read.csv("C:/dumy_network.csv", header=TRUE)

#build list of nodes (no repetition)
nodes=unique(c(as.character(mydata$startlink),as.character(mydata$endlink)))
#list of all links
links=mydata[,1:2]
#cost of moving material on each link - optimization problem objective coefficients
cost=mydata$cost

#decision variable is flow on each link. Objective is to minimize product of (cost on each link*flow on the link). Therefore, count of decision variable is equal to count of links.

#constraints matrix: for each node in nodes, incoming quantity should be equal to outgoing quantity. 

constraints=matrix(0,sum(mydata$max_const!='null')+sum(mydata$equality_const!='null')+length(nodes),length(mydata$endlink))

for (i in 1:length(nodes) ) {
  constraints[i,]=t(1*(nodes[i]==links[,1])-1*(nodes[i]==links[,2]))
}

#get constraints for equality constraints- in some cases we have to ship material exactly same as this quanity.
constraint1=matrix(mydata$equality_const,1,length(mydata$equality_const))
constraint1[constraint1=="null"]=0
constraint1=as.numeric(constraint1)
constraint1_length=which(constraint1!=0)

constraint1_final=matrix(0,length(constraint1_length),length(mydata$equality_const))

for (i in 1:length(constraint1_length) ) {
  constraint1_final[i,constraint1_length[i]]=1
}
start=length(nodes)+1
end=length(nodes)+length(constraint1_length)
constraints[start:end,]=constraint1_final

#get constraints for maxconstraints - in some cases we cannot ship material exceeding this quanity.
constraint2=matrix(mydata$max_const,1,length(mydata$max_const))
constraint2[constraint2=="null"]=0
constraint2=as.numeric(constraint2)
constraint2_length=which(constraint2!=0)

constraint2_final=matrix(0,length(constraint2_length),length(mydata$max_const))

for (i in 1:length(constraint2_length) ) {
  constraint2_final[i,constraint2_length[i]]=1
}
start=end+1
end=end+length(constraint2_length)
constraints[start:end,]=constraint2_final


#building direction of constraints
direction=c(rep("=",length(nodes)),rep("=",sum(mydata$equality_const!='null')),rep("<=",sum(mydata$max_const!='null')))

#building right hand side of constraints
b1=as.numeric(as.character(mydata$equality_const[constraint1_length]))
b2=as.numeric(as.character(mydata$max_const[constraint2_length]))
b=c(rep(0,length(nodes)),b1,b2)

res = lpSolve::lp('min', cost, constraints, direction, b,  all.int = TRUE)
res$solution

answers1=data.frame(res$solution)
mydata=cbind(mydata,answers1)

=======================================更新2======== =============================== 没有得到任何答案，所以试图简化我的问题：

page 的示例部分给出了一个简单的问题。有谁知道如何使用 Mapreduce 解决它？我的意思是说我有一个类似的问题，但是有大量的变量和约束，那么有没有办法实现更快的处理？

【问题讨论】：

  • 通常分布式优化涉及某种分解方法（例如 Benders 分解或 Dantzig-Wolfe 分解）。我们需要有关您的模型的更多详细信息来帮助您。
  • 我的问题的简化版本是stackoverflow.com/questions/22736107/…。我的实际模型在 2 个城市之间有许多不同的节点和旅行选择......
  • 请编辑您的问题以添加您的数据和您当前正在运行的代码的最小可重现示例。
  • @josilber 我已根据您的要求添加了信息。我希望这是有道理的。期待您的意见！
  • 似乎有人提出了一种解决运输问题的并行方法：dtic.mil/dtic/tr/fulltext/u2/a196238.pdf。不过，实施这种方法将非常复杂。

标签： r hadoop mathematical-optimization lpsolve


【解决方案1】：

简而言之，您想要进行大规模的线性规划 (lp)，但您对求解器的性能不满意。

方法

我建议以下方法。

求解器

您需要使用此求解器吗？以下是一些替代方案：
- Gurobi
- Mosek
- CPlex

并行执行

实现并行处理的一个弱方法是并行运行进程。您可以一次启动多个优化吗？ （在这里，您可以在 Hadoop 中编写一个简单的 exectuion 任务）

改进的输入

如果您相应地提交数据，求解器的性能会更好（稀疏矩阵，...）

在您的代码中，您使用lpSolve::lp。根据CRAN page of lpSolver 5.6.7及其PDF，有一种特殊的传输模式lpSolve:lp.transport。 （我没有使用过 R 并且对语法不熟悉。我是 Matlab 的“爱好者”。）

最新研究

也许有些精力花在了学术上。您可能会在 sciencedirect.com 上找到一篇论文（花费一些美元）。

硬件

CPU 越快，您获得解决方案的速度就越快（答案）。您介意向我们介绍一下您的环境（操作系统、CPU）吗？

最后的想法

您是否尝试在quant.stackexchange.com 中提出您的问题，因为数值优化与数学密切相关并且“不涉及太多编程”。

并行模式

我们使用CLP。根据一些同事的说法，您可以设置一个标志或使用不同的实现来并行运行求解器（1 个问题在许多内核上执行）。看看 SYMPHONY。




随时向我们发布消息，您是否介意向我们提供有关您的环境（操作系统、CPU）的一些见解？

我的赏金

在 Quora 上，您会发现 Hadoop 擅长数据处理，I/O 速度快，不是专门为优化或计算而设计的：Quora

如果你不想给我赏金，请告诉我。

缺少 +100 的赏金

即使我进行了一些研究，我也没有收到任何赏金...这个帖子到了一个地步，您没有向我们提供必要的细节，我们应该为您解决它 - 这永远不会成功！

【讨论】：

  • solver：只想坚持 lpsolve 求解器。但我不认为更改求解器会产生巨大的影响，并行执行：不可能 - 整个网络必须在单个 pc 上运行 - 一个节点可以提供给超过 2 个站点，1 个站点可以从多个站点接收节点，改进的输入：会试一试 - 但不认为它会产生巨大影响，最新研究：将尝试试一试，硬件：我正在使用我可用的最快配置，最终想法：会尝试跨度>
  • 最近，我们对大型顶级保险进行了 ALM 优化，我们花了几天时间找出什么是加快计算速度的最佳求解器。对于另一个项目，我们购买了一台新的功能强大的机器来检查求解器如何在并行模式和特定操作系统下扩展...... CPU、求解器和操作系统的关系比（我们）想象的更密切。
  • 您提到了并行模式下的求解器缩放...并行模式是什么意思？能否提供一些见解？
  • 你提供了很多信息......让我读一下，然后就赏金问题回复你
  • 有什么办法可以分配给你一半的赏金吗？我感谢您的投入，并希望奖励您。同时，由于我没有得到解决方案，我认为我不应该给予全部赏金
猜你喜欢
  • 2020-08-09
  • 1970-01-01
  • 1970-01-01
  • 1970-01-01
  • 2015-03-24
  • 1970-01-01
  • 1970-01-01
  • 2012-07-09
  • 1970-01-01
相关资源
最近更新 更多
热门标签
Java Python linux javascript Mysql C# Docker 算法 前端 SpringBoot Redis Vue spring 设计模式 .net core .net kubernetes c++ 数据库 数据结构 大数据 js 机器学习 微服务 Android Go 程序员 面试 JVM ASP.net core 云原生 人工智能 后端 PHP git CSS golang k8s Nginx Django mybatis 深度学习 多线程 React 架构 devops 爬虫 云计算 Spring Boot LeetCode