https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Explicit-SVD-vs-SVD-with-eigen-solver/m-p/977993#M17211 <P>Hi,&nbsp;</P> <P>could you clarify two following questions.</P> <OL> <LI>Consider following mathematical problem.&nbsp;<STRONG style="font-size: 1em; line-height: 1.5;">Given general full rank square non symmetric matrix A of 13 000 x 13 000 size I want to find its SVD with all singular values and all right/left eigen vectors</STRONG><SPAN style="font-size: 1em; line-height: 1.5;">. But when I solve it with some driver SVD routine (e.g. LAPACKE_sgesdd) it takes about 2 x slower then I solve two eigen decomposition problems at a time: for A*A' (getting left eigen vectors of A) and A'*A&nbsp;</SPAN><SPAN style="font-size: 1em; line-height: 1.5;">(getting right eigen vectors of A)</SPAN><SPAN style="font-size: 1em; line-height: 1.5;"> matrices.</SPAN><BR /> <SPAN style="font-size: 1em; line-height: 1.5;">Is it normal behavior or I miss something and there is more proper/fast way to find SVD for given matrix type explicitly (via SVD driver routine)?</SPAN></LI> <LI>Considering the same problem, is there any way in MKL <STRONG>to find small subset</STRONG> (let it be 1 left and right eigen vectors for example) <STRONG>of all SVD right/left eigen vectors</STRONG> with the biggest singular values, <STRONG>saving "considerable" amount of time</STRONG> (at least 30%) ?</LI> </OL> <P>Thank you in advance.</P> Fri, 11 Apr 2014 05:04:58 GMT VICTOR_K_Intel 2014-04-11T05:04:58Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Explicit-SVD-vs-SVD-with-eigen-solver/m-p/977993#M17211 <P>Hi,&nbsp;</P> <P>could you clarify two following questions.</P> <OL> <LI>Consider following mathematical problem.&nbsp;<STRONG style="font-size: 1em; line-height: 1.5;">Given general full rank square non symmetric matrix A of 13 000 x 13 000 size I want to find its SVD with all singular values and all right/left eigen vectors</STRONG><SPAN style="font-size: 1em; line-height: 1.5;">. But when I solve it with some driver SVD routine (e.g. LAPACKE_sgesdd) it takes about 2 x slower then I solve two eigen decomposition problems at a time: for A*A' (getting left eigen vectors of A) and A'*A&nbsp;</SPAN><SPAN style="font-size: 1em; line-height: 1.5;">(getting right eigen vectors of A)</SPAN><SPAN style="font-size: 1em; line-height: 1.5;"> matrices.</SPAN><BR /> <SPAN style="font-size: 1em; line-height: 1.5;">Is it normal behavior or I miss something and there is more proper/fast way to find SVD for given matrix type explicitly (via SVD driver routine)?</SPAN></LI> <LI>Considering the same problem, is there any way in MKL <STRONG>to find small subset</STRONG> (let it be 1 left and right eigen vectors for example) <STRONG>of all SVD right/left eigen vectors</STRONG> with the biggest singular values, <STRONG>saving "considerable" amount of time</STRONG> (at least 30%) ?</LI> </OL> <P>Thank you in advance.</P> Fri, 11 Apr 2014 05:04:58 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Explicit-SVD-vs-SVD-with-eigen-solver/m-p/977993#M17211 VICTOR_K_Intel 2014-04-11T05:04:58Z