Matlab mldivide 和 arrayfire 求解

问：

我正在尝试使用 arrayfire 将一些 Matlab 代码移植到 c++。在一个地方，Matlab代码是这样的： tmp = （R\v0）; 其中 R 是 9x9 复数双精度值，v0 是 9x1 复数双精度值。 在我的 ArrayFire 中，编写的代码如下： af：：array tmp = af：：solve（R， v0_ndx， AF_MAT_NONE）;v0-ndx 是 Arrayfire 案例中的 141 个条目之一。 在这两种情况下，R 数组都是：

(1.0000,0.0000)   (0.4330,0.5572)   (-0.4989,0.3674)  (0.6645,-0.1705)  (-0.5022,-0.3560) (0.0136,-0.5150)  (0.4447,0.2106)   (-0.5152,0.1762)  (0.7409,0.2369) 
(0.4330,-0.5572)  (1.0000,0.0000)   (0.2444,0.6848)   (0.3885,-0.7665)  (-0.5994,0.3623)  (-0.4474,-0.4617) (0.5153,-0.2913)  (-0.1954,0.6379)  (0.7954,-0.4984) 
(-0.4989,-0.3674) (0.2444,-0.6848)  (1.0000,0.0000)   (-0.4839,-0.1905) (0.4046,0.7260)   (-0.2262,0.3428)  (-0.1181,-0.5652) (0.6709,0.2600)   (-0.1537,-0.6409) 
(0.6645,0.1705)   (0.3885,0.7665)   (-0.4839,0.1905)  (1.0000,0.0000)   (-0.4047,-0.5696) (0.3418,-0.7446)  (0.2774,0.3234)   (-0.6889,-0.1392) (0.7275,0.4112) 
(-0.5022,0.3560)  (-0.5994,-0.3623) (0.4046,-0.7260)  (-0.4047,0.5696)  (1.0000,0.0000)   (0.3853,0.5248)   (-0.5786,0.0954)  (0.5647,-0.4580)  (-0.6453,0.0559) 
(0.0136,0.5150)   (-0.4474,0.4617)  (-0.2262,-0.3428) (0.3418,0.7446)   (0.3853,-0.5248)  (1.0000,0.0000)   (-0.2308,0.1934)  (-0.0831,-0.7353) (-0.1197,0.6637) 
(0.4447,-0.2106)  (0.5153,0.2913)   (-0.1181,0.5652)  (0.2774,-0.3234)  (-0.5786,-0.0954) (-0.2308,-0.1934) (1.0000,0.0000)   (-0.0868,0.3612)  (0.5367,-0.0571) 
(-0.5152,-0.1762) (-0.1954,-0.6379) (0.6709,-0.2600)  (-0.6889,0.1392)  (0.5647,0.4580)   (-0.0831,0.7353)  (-0.0868,-0.3612) (1.0000,0.0000)   (-0.4853,-0.3956) 
(0.7409,-0.2369)  (0.7954,0.4984)   (-0.1537,0.6409)  (0.7275,-0.4112)  (-0.6453,-0.0559) (-0.1197,-0.6637) (0.5367,0.0571)   (-0.4853,0.3956)  (1.0000,0.0000) 

v0 或 （v0_ndx） 数组为：

         (1.0000,0.0000) 
         (0.8827,0.4699) 
         (0.1784,-0.9840) 
         (-0.9616,0.2745) 
         (-0.5396,0.8419) 
         (0.9908,-0.1353) 
         (-0.5675,-0.8234) 
         (0.8795,-0.4758) 
         (0.9408,0.3389) 

Matlab leftdivide 返回

6.83220363990185 - 10.9619257689701i
40.3356738405241 + 25.0998371535142i
4.83385229841040 - 37.2045139160788i
-44.4504672789456 + 4.68752510309329i
-18.7242610414281 + 10.0126185742854i
13.2133070644273 - 1.29385372118540i
-2.14180173734413 + 12.1355411649223i
-4.16564154531280 - 6.04196278680202i
-2.43069655994337 + 6.59335067976014i

arrayfire 求解返回（使用 AF_MAT_NONE 参数;AF_MAT_LOWER 和 AF_MAT_UPPER 产生更差的结果）

(6.1867,-9.3346) 
(31.5626,-5.3650) 
(16.7719,-11.4021) 
(-31.0820,-10.4292) 
(-20.8629,35.2594) 
(3.3563,-6.7437) 
(-21.9321,-8.4480) 
(3.5929,5.8619) 
(-2.2422,9.7816)