Commit Graph

20 Commits

Author SHA1 Message Date
Gromiusz
fcfd95a1af feat: faster thread method 2025-01-01 18:50:11 +01:00
Gromiusz
fd94f597a5 feat: trying and testing different ways to implement threads 2024-12-28 13:24:58 +01:00
afd9e90a38 feat: add distributed arrays to processing type 2024-12-05 20:24:53 +01:00
3721418264 Reapply "Merge branch 'main' of https://github.com/kuhyx/PORR"
This reverts commit 87d6bb0d97.
2024-12-05 20:16:13 +01:00
87d6bb0d97 Revert "Merge branch 'main' of https://github.com/kuhyx/PORR"
This reverts commit a7123acced, reversing
changes made to 4512cf9305.
2024-11-28 17:48:11 +01:00
aleksandrasob
85a5f61142 add writinf/reading to/from file, remove rest of eigs calcs 2024-11-28 15:02:00 +01:00
aleksandrasob
c2263b9361 sing vals as eigenvals for nementh12 2024-11-17 17:45:23 +01:00
aleksandrasob
7c6482eed9 adding time measurements for processes, adding two more matrix types 2024-11-11 17:25:36 +01:00
Gromiusz
73a12d3859
Time measurement feature (#6)
* making new branch and importing time

* add time_measurement decorator

* add measurement functionality to solve()
2024-10-31 17:50:21 +01:00
Gromiusz
c938e63b7f
Complete all threads functions in Linear Algebra Utils, add new parameters to tests (#5)
* complete threads function (tests FAIL)

Test fail due to indivisibility of the problem into threads. Test is turned to ProcessingType.THREADS!. On ProcessingType.SEQUENTIAL all test PASS.
TODO: make sure that len(problem) is divisible by num of threads. It can be less threads for small problem.

* fixing bugs in linear_algebra_utils, properly adress chunks to threads

* the refactor of the ThreadsLinearAlgebraUtils

* set test to execute both sequential and threads processing types

* Delete unnecessary lines tests.py

* rename scalar_matrix_multiply to scalar_vector_multiply
2024-10-27 20:13:43 +01:00
Gromiusz
b02117c11b
Add Interface for Extending Parallel Processing Methods (#4)
* Add choice between sequential or parallel(threads) processing

The LinearAlgebraUtils class has been changed to abstract class.

* remove needless imports and move the functionality from constructor to the function

* add ProcessingType class for a more transparent selection of methods

* the refactor of the assign_method

* rename the assign_method method
2024-10-25 17:01:19 +02:00
6946035518 feat: compare norms calculated by richardson vs numpy 2024-10-20 19:34:12 +02:00
cb179a76ce chore: moved debug prints to separate function 2024-10-20 19:20:38 +02:00
d74da05da8 feat: made max iterations obligatory argument 2024-10-20 19:06:15 +02:00
9014c37064 feat: can run tests independently 2024-10-20 18:39:01 +02:00
838d5ed563 fix: removed adding "size" to self values 2024-10-20 18:32:12 +02:00
ea98dc9712 feat: changed to proper pytests 2024-10-20 18:27:51 +02:00
76ace2b76c feat: reworked convergence calculations 2024-10-20 18:09:06 +02:00
d3b0e8e0cc feat: convergence method suggested by wikipedia 2024-10-20 17:47:37 +02:00
aleksandrasob
52d6012092 Add sequential version of Richardson's algorithm and first tests 2024-10-20 16:30:42 +02:00