[c00lkent]

Hey everyone. I have a distributed memory assignment that I would like some input on. I would talk to the professor but I believe he does not do much communication via email.

What I have done so far is create six matrices, did matrix addition on all of them to create the distributed memory vector. I found the dot product of all sets of pairs of the original matrices (15 total). I then created two "dirty" versions with 20% damage of each of the original six matrices.

Now do I take the dot product of the distributed memory vector with each of the clean versions and their associated dirty versions?

I am hoping someone has done something like this before and can offer some input. I am also not sure if I created my distributed memory vector properly.

Thanks,
c00lkent

The matrices labeled Dsomething are the "dirty" matrices and the rows on the third page that have C in front of them are the dot products of the distributed memory vector with the clean vectory.

Homework
Work Page 1
Work Page 2 Updated
Work Page 3

This post has been edited by c00lkent: 21 September 2014 - 02:14 PM

[macosxnerd101]

Moved to Computer Science.

Quote

Now do I take the dot product of the distributed memory vector with each of the clean versions and their associated dirty versions?

This really isn't clear from the instructions if you only take a dot product of the original vector with its damaged vectors, or if you take all 6 original vectors and take the dot product with each dirty vector. Perhaps a classmate or a TA will have more insight into this.

Quote

I am also not sure if I created my distributed memory vector properly.

I checked some of the first vector. It looks to me like you are doing things correctly on this count.

[c00lkent]

Quote

Quote

Now do I take the dot product of the distributed memory vector with each of the clean versions and their associated dirty versions?

This really isn't clear from the instructions if you only take a dot product of the original vector with its damaged vectors, or if you take all 6 original vectors and take the dot product with each dirty vector. Perhaps a classmate or a TA will have more insight into this.

I have talked with another student and we both believe that you are supposed to take the dot product of the distributed memory vector with the clean matrices and their dirty versions.

The problem is I am not seeing a clear answer at the end; the numbers don't seem to indicate anything that I am aware of at the moment. My friend is using zeros and ones and is also normalizing so our answers won't be that similar. He said he's redoing without normalizing though so maybe that will help.

We don't have TAs at our school, at least not in the CS department.

But thanks for the help, I've uploaded my final results following what I believe to be the correct process though I don't know what to take away from it.

This post has been edited by c00lkent: 21 September 2014 - 02:14 PM