elpa_eigenvalues_float(3)

✨ NAME ✨

okay so elpa_eigenvalues_float — this icon literally computes all eigenvalues λ for a real symmetric eigenproblem: A*Q = λ*Q. no cap, that's the whole vibe. 💅

(from the Library Functions Manual, bestie 🫶)

---

🌸 SYNOPSIS 🌸

💻 FORTRAN INTERFACE

yeah so first up, the Fortran girlies do this:

use elpa
class(elpa_t), pointer :: elpa_handle

call elpa_handle%eigenvalues_float (a, ev, error)

and here's what all those variables mean, fr fr:

• class(elpa_t) :: elpa_handle — an instance of the ELPA object. she's the main character 🔥
• real(kind=c_float) :: a OR type(c_ptr) :: a — the local part of the host/device matrix A for which the eigenvalues should be computed. the dimensions of matrix a must be set BEFORE with the methods elpa_set(3) and elpa_setup(3). the global matrix has to be symmetric, and this is not checked by the routine. in case of a GPU build, a can be a pointer to the device memory. 👑
• real(kind=c_float) :: ev OR type(c_ptr) :: ev — the global array where the eigenvalues λ will be stored in ascending order. in case of a GPU build, ev can be a pointer to the device memory. 🌈
• integer, optional :: error — the return error code of the function. should be "ELPA_OK". the error code can be queried with the function elpa_strerr(3) 🫶

🦋 C/C++ INTERFACE

C/C++ besties, y'all do:

#include <elpa/elpa.h>
elpa_t elpa_handle;

void elpa_eigenvalues_float(elpa_t elpa_handle, float *a, float *ev, int
       *error);

and the deets:

• elpa_t elpa_handle; — the handle to the ELPA object. iconic. 👑
• float *a; — the local part of the host/device matrix A for which the eigenvalues should be computed. the dimensions of matrix a must be set BEFORE with the methods elpa_set(3) and elpa_setup(3). the global matrix A has to be symmetric, this is not checked by the routine. in case of a GPU build, a can be a pointer to the device memory. 💀 (not checked, so don't come for her if you fed her a non-symmetric mess)
• datatype_real *ev; — the array where the eigenvalues λ will be stored in ascending order. eigenvalues will be stored in ascending order. in case of a GPU build, ev can be a pointer to the device memory. yes, ascending twice, she's consistent 💅
• int *error; — the error code of the function. should be "ELPA_OK". the error codes can be queried with elpa_strerr(3) ✨

---

🎀 DESCRIPTION 🎀

so what does she do? she computes the eigenvalues of a float precision real symmetric matrix. periodt.

but listen: elpa_init(3), elpa_allocate(3), elpa_set(3), and elpa_setup(3) must be called BEFORE elpa_eigenvalues_float can be called. like, do the pregame first, bestie. no shortcuts. 😭

---

🌟 SEE ALSO 🌟

the whole friend group, lowkey essential:

• elpa2_print_kernels(1)
• elpa_init(3)
• elpa_allocate(3)
• elpa_set(3)
• elpa_setup(3)
• elpa_strerr(3)
• elpa_skew_eigenvalues(3)
• elpa_eigenvectors(3)
• elpa_skew_eigenvectors(3)
• elpa_cholesky(3)
• elpa_invert_triangular(3)
• elpa_solve_tridiagonal(3)
• elpa_eigenvalues(3)
• elpa_uninit(3)
• elpa_deallocate(3)

---

xoxo, ELPA 🌸 — Thu Nov 28 2024 — elpa_eigenvalues_float(3) out 💖