Nigel
/
mindspore
forked from mindspore-Ecosystem/mindspore
0
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
45,876 Commits 24 Branches 0 Tags 3.9 GiB
Commit Graph

112 Commits

Author SHA1 Message Date
zhujingxuan e9c267d268 reduce testcase size 2022-06-09 15:53:52 +08:00
liqiliang 4be416f40f Add a gpu kernel, MatrixBandPart. 2022-05-14 10:46:38 +08:00
z00512249 fb1b4324d7 refactor matrix_set_diag_v3 for gpu backend. 2022-05-11 14:34:21 +08:00
liqiliang a67193ccb4 Add a cpu kernel, MatrixBandPart. 2022-05-10 14:51:53 +08:00
hezhenhao1 c7452b1c9a Separate test_trsm_grad case. 2022-05-09 14:49:34 +08:00
chauneahhin e9bbec3b4c [feat] [assistant] [I4CRJN] [I4CRJM] [I4CRJL] Add MatrixDiagV3, MatrixSetDiagV3 and MatrixDiagPartV3 2022-04-13 10:23:51 +08:00
z00512249 7a54f2d90d fix scipy sparse gmres level1 test error 2022-04-02 14:54:39 +08:00
z00512249 0ea895bc84 add gmres dense bprop 2022-03-22 20:01:38 +08:00
i-robot 6a07fffcac
!31635 opt gmres code 
Merge pull request !31635 from zhuzhongrui/pub_master
 2022-03-22 01:30:27 +00:00
z00512249 b6081059aa opt gmres code 2022-03-21 19:31:41 +08:00
hezhenhao1 36d4328392 Add tese cases where b is (N, 1) for cg grad. 2022-03-21 14:21:58 +08:00
z00512249 b9cdc238d9 fix add sparse gmres tests 2022-03-17 14:57:32 +08:00
yanghaoran 44aba14913 takedown test_gmres_agains_scipy to ensure gate stability 2022-03-17 09:46:25 +08:00
z00512249 95a34dcbfa add sparse gmres api && check && bprop 2022-03-16 11:25:10 +08:00
hezhenhao1 af79596baf CSRTensor test cases do not support mac and window. 2022-03-14 14:06:40 +08:00
i-robot e14d0ca36b
!31124 Split test cases of bfgs method. 
Merge pull request !31124 from hezhenhao1/fix_scipy
 2022-03-11 01:10:00 +00:00
hezhenhao1 b63e4b8b20 Fix bprop of cg method, support to return CSRTensor. 2022-03-10 21:24:39 +08:00
hezhenhao1 892c091582 Split test cases of bfgs method. 2022-03-10 20:02:16 +08:00
hezhenhao1 28e74b1e5f Fix sparse check and refactor the test cases of cg method. 2022-03-10 15:33:11 +08:00
i-robot 98888f6cd8
!31036 Add support of cg backward in pynative mode. 
Merge pull request !31036 from hezhenhao1/master
 2022-03-10 02:59:24 +00:00
hezhenhao1 bad551d210 Add support of cg backward in pynative mode. 2022-03-10 09:13:21 +08:00
z00512249 b368237340 fix Cholesky bprop implement to support symmetric 2022-03-08 14:54:45 +08:00
hezhenhao1 9682769a02 Add grad implementation and test cases for cg method. 2022-03-08 10:12:23 +08:00
hezhenhao1 885210b634 Add CSRTensor support and test cases for cg method. 2022-03-02 19:22:11 +08:00
hezhenhao1 a141c8c8e3 Fix grads of Eigh and SolveTriangular operator in PYNATIVE mode. 2022-02-28 18:06:55 +08:00
i-robot 3e246484f0
!30539 Fix bug of grad implementation of solve_triangular. 
Merge pull request !30539 from hezhenhao1/fix_trsm_grad
 2022-02-25 02:52:43 +00:00
hezhenhao1 9468e2dc84 Fix bug of grad implementation of solve_triangular. 2022-02-24 19:56:34 +08:00
z00512249 8bba3a063a fix scipy st utils to creat_sys_pos_martix 2022-02-24 19:52:19 +08:00
hezhenhao1 bcb936373e Apply value and type checking for all functions in mindspore.scipy module. 2022-02-23 14:04:39 +08:00
zhujingxuan c99aa71ba6 move error test to daily build 2022-02-21 20:41:55 +08:00
i-robot 218c3735df
!30196 Refactor value and type checking for mindspore.scipy module. 
Merge pull request !30196 from hezhenhao1/add_checker
 2022-02-21 09:37:26 +00:00
hezhenhao1 73ad595fa3 Refactor value and type checking for mindspore.scipy module. 2022-02-21 09:35:59 +08:00
z00512249 fb31addb16 add cho_factor && cho_solve grad st 2022-02-18 17:53:56 +08:00
z00512249 a5810f4d3a add cho_factor && mgrad st 2022-02-18 10:39:41 +08:00
i-robot 0efb23fd96
!30121 Optimize eigh ops, backend suppor t 1/2 output, 
Merge pull request !30121 from wuwenbing/master
 2022-02-17 09:24:01 +00:00
wenbean ba3b65b9af Optimize eigh, backend support 1 or 2 output for GPU/CPU 2022-02-17 10:45:00 +08:00
i-robot 9d1839119d
!30095 Remove CholeskySolve and replace it by using solve_triangular 
Merge pull request !30095 from zhuzhongrui/pub_master3
 2022-02-16 08:10:56 +00:00
z00512249 f066e36fe4 Remove CholeskySolve and replace it by using solve_triangular 2022-02-16 14:08:55 +08:00
wangrao124 234b80a8d6 fix exception info of dtype 2022-02-14 17:05:33 +08:00
zhujingxuan e30847cac0 The `solve_triangular` API fallback. 
Now only the `SolveTriangular` op supports batch computation, the `solve_triangular` API only accept 2D matrix input.
 2022-02-12 18:10:29 +08:00
zhujingxuan 933f99f917 batch support for solve_triangular 
Add batch support for solve_triangular and the corresponding test cases
 2022-02-10 17:34:46 +08:00
i-robot d0877dda3c
!29838 add gpu cholesky_grad st case 
Merge pull request !29838 from zhuzhongrui/pub_master2
 2022-02-09 10:02:25 +00:00
i-robot f3b2886096
!29680 Refine CPU thread and blocknumber to automatic calucate 
Merge pull request !29680 from wuwenbing/master
 2022-02-09 08:40:35 +00:00
z00512249 1e5d22a7fa add gpu cholesky_grad st case 2022-02-09 16:36:16 +08:00
i-robot 6396c079f9
!29766 fix linalg type check 
Merge pull request !29766 from zhuzhongrui/pub_master2
 2022-02-09 02:25:55 +00:00
wenbean 224feb8c45 Refine lambda to name function, optmize py ops, add validate 2022-02-09 09:19:53 +08:00
i-robot bc693e14c5
!29737 Opt Eig operator to support batch shape. 
Merge pull request !29737 from hezhenhao1/fix_eig
 2022-02-08 11:45:17 +00:00
z00512249 9390d67adf fix linalg type check 2022-02-08 17:14:02 +08:00
hezhenhao1 ace447279e Opt Eig operator to support batch shape. 2022-02-08 16:00:27 +08:00
zhujingxuan aec3357a03 refine eigh API exception 
Make sure msp.linalg.eigh throw exceptions when given wrong parameter and behaves the same between GPU and CPU.
 2022-02-07 16:12:20 +08:00