1 страниц (4 вхождений)
Shortening calculation times for matrix multiplication operations - A functions of a sigle parameter is set-up starting from multiple matrix multiplications: how to sherten calculation times? - Сообщения
Dear all,
please find attached a worksheet I'm working on. SMath performs well but quite slowly (around 7 hours...).
Any idea on how to speed up calculations and, particularely, the plot of the functions Zφ2(ω), Zφad2(p_ad2) , Zφad1(p_ad1)?
For info:
1 The worksheet is not at the latest SMath version....I have not installed the last version fearing some regression issue...if someone can suggest me how to install the latest version without overwriting the old one (in order to verify non-regression) I'd be grateful to him
2 For all the plotted functions the option OPTIMISATION is set to SYMBOLIC
3 I'm setting up a version of the quoted functions (Zφ2(ω), Zφad2(p_ad2) , Zφad1(p_ad1)) where I'll use ONLY the terms of the first and last matrix necessary for the specific boundary condition. But it appears quite cumbersome....
Many thanks
Notched_Alu_Beam_vs_measures V6 adimensional out-of-plane bending cfr.sm (393,87 КиБ) скачан 514 раз(а).TB_TM algo_original.sm (99,51 КиБ) скачан 512 раз(а).TB_TM_adim_algo.sm (892,69 КиБ) скачан 513 раз(а).
please find attached a worksheet I'm working on. SMath performs well but quite slowly (around 7 hours...).
Any idea on how to speed up calculations and, particularely, the plot of the functions Zφ2(ω), Zφad2(p_ad2) , Zφad1(p_ad1)?
For info:
1 The worksheet is not at the latest SMath version....I have not installed the last version fearing some regression issue...if someone can suggest me how to install the latest version without overwriting the old one (in order to verify non-regression) I'd be grateful to him
2 For all the plotted functions the option OPTIMISATION is set to SYMBOLIC
3 I'm setting up a version of the quoted functions (Zφ2(ω), Zφad2(p_ad2) , Zφad1(p_ad1)) where I'll use ONLY the terms of the first and last matrix necessary for the specific boundary condition. But it appears quite cumbersome....
Many thanks
Notched_Alu_Beam_vs_measures V6 adimensional out-of-plane bending cfr.sm (393,87 КиБ) скачан 514 раз(а).TB_TM algo_original.sm (99,51 КиБ) скачан 512 раз(а).TB_TM_adim_algo.sm (892,69 КиБ) скачан 513 раз(а).
.....About my post "Shortening calculation times for matrix multiplication operations"
one more useful information. I stated around 7 hours to calculate (the plots mainly) the worksheet.
It's fair to specify on which machine the worksheet ran (see attached file).
One more question : is it possible multitasking in SMath as it's in MathCAD? You think it could help for my calculation?
THANKS
laptop.pdf (239,37 КиБ) скачан 507 раз(а).
one more useful information. I stated around 7 hours to calculate (the plots mainly) the worksheet.
It's fair to specify on which machine the worksheet ran (see attached file).
One more question : is it possible multitasking in SMath as it's in MathCAD? You think it could help for my calculation?
THANKS
laptop.pdf (239,37 КиБ) скачан 507 раз(а).
...sorry again: not multitasking BUT multiprocessors (calculations)....
Hi. Adding eval to each function significantly increases speed, although it might introduce some rounding error. This worked for the original algorithm, at least; I assume it will work for the dimensionless algorithm as well. Assigning optimization to Numeric can also improve sometimes the calculation speed.
TB_TM algo_original.sm (101,6 КиБ) скачан 509 раз(а).
Notched_Alu_Beam_vs_measures V6 adimensional out-of-plane bending cfr.sm (392,73 КиБ) скачан 512 раз(а).
Best regards.
Alvaro.
TB_TM algo_original.sm (101,6 КиБ) скачан 509 раз(а).
Notched_Alu_Beam_vs_measures V6 adimensional out-of-plane bending cfr.sm (392,73 КиБ) скачан 512 раз(а).
Best regards.
Alvaro.
1 страниц (4 вхождений)
- Новые сообщения
- Нет новых сообщений