Implement dpnp.linalg.lu_solve() batch inputs (#2619)

This PR suggests extending `dpnp.linalg.lu_solve()`
#2575 for batch arrays and
adding new lapack `getrs_batch` extension

Author: vlad-perevezentsev

CHANGELOG.md

@@ -11,6 +11,7 @@ This release changes the license from `BSD-2-Clause` to `BSD-3-Clause`.
 ### Added
 
 * Added the docstrings to `dpnp.linalg.LinAlgError` exception [#2613](https://github.com/IntelPython/dpnp/pull/2613)
+* Added implementation of `dpnp.linalg.lu_solve` for batch inputs (SciPy-compatible) [#2619](https://github.com/IntelPython/dpnp/pull/2619)
 
 ### Changed
 

dpnp/backend/extensions/lapack/CMakeLists.txt

@@ -41,6 +41,7 @@ set(_module_src
     ${CMAKE_CURRENT_SOURCE_DIR}/getrf_batch.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/getri_batch.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/getrs.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/getrs_batch.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/heevd.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/heevd_batch.cpp
     ${CMAKE_CURRENT_SOURCE_DIR}/orgqr.cpp

dpnp/backend/extensions/lapack/getrs.cpp

@@ -51,14 +51,14 @@ namespace type_utils = dpctl::tensor::type_utils;
 using ext::common::init_dispatch_vector;
 
 typedef sycl::event (*getrs_impl_fn_ptr_t)(sycl::queue &,
-                                           oneapi::mkl::transpose,
+                                           const oneapi::mkl::transpose,
                                            const std::int64_t,
                                            const std::int64_t,
                                            char *,
-                                           std::int64_t,
+                                           const std::int64_t,
                                            std::int64_t *,
                                            char *,
-                                           std::int64_t,
+                                           const std::int64_t,
                                            std::vector<sycl::event> &,
                                            const std::vector<sycl::event> &);
 
@@ -70,10 +70,10 @@ static sycl::event getrs_impl(sycl::queue &exec_q,
                               const std::int64_t n,
                               const std::int64_t nrhs,
                               char *in_a,
-                              std::int64_t lda,
+                              const std::int64_t lda,
                               std::int64_t *ipiv,
                               char *in_b,
-                              std::int64_t ldb,
+                              const std::int64_t ldb,
                               std::vector<sycl::event> &host_task_events,
                               const std::vector<sycl::event> &depends)
 {
@@ -234,7 +234,7 @@ std::pair<sycl::event, sycl::event>
         throw py::value_error("The right-hand sides array "
                               "must be F-contiguous");
     }
-    if (!is_ipiv_array_c_contig || !is_ipiv_array_f_contig) {
+    if (!is_ipiv_array_c_contig && !is_ipiv_array_f_contig) {
         throw py::value_error("The array of pivot indices "
                               "must be contiguous");
     }

dpnp/backend/extensions/lapack/getrs.hpp

@@ -40,8 +40,23 @@ extern std::pair<sycl::event, sycl::event>
           const dpctl::tensor::usm_ndarray &a_array,
           const dpctl::tensor::usm_ndarray &ipiv_array,
           const dpctl::tensor::usm_ndarray &b_array,
-          oneapi::mkl::transpose trans,
+          const oneapi::mkl::transpose trans,
           const std::vector<sycl::event> &depends = {});
 
+extern std::pair<sycl::event, sycl::event>
+    getrs_batch(sycl::queue &exec_q,
+                const dpctl::tensor::usm_ndarray &a_array,
+                const dpctl::tensor::usm_ndarray &ipiv_array,
+                const dpctl::tensor::usm_ndarray &b_array,
+                const oneapi::mkl::transpose trans,
+                const std::int64_t n,
+                const std::int64_t nrhs,
+                const std::int64_t stride_a,
+                const std::int64_t stride_ipiv,
+                const std::int64_t stride_b,
+                const std::int64_t batch_size,
+                const std::vector<sycl::event> &depends = {});
+
 extern void init_getrs_dispatch_vector(void);
+extern void init_getrs_batch_dispatch_vector(void);
 } // namespace dpnp::extensions::lapack