Serving the Quantitative Finance Community

Search found 24 matches

  • 1
  • 2
by richardlm
August 23rd, 2010, 7:07 pm
Forum: Numerical Methods Forum
Topic: PCG discussions
Replies: 13
Views: 30676

PCG discussions

<t>QuoteOriginally posted by: mblatt[...]BTW if it is not in the mkl, it does not mean that it is not there in other libraries. I am pretty sure that parallel ilu preconditioners exist, as the parallelization should be rather straight forward due to the limited and forseeable additional fill-in.I do...
by richardlm
August 19th, 2010, 5:39 pm
Forum: Numerical Methods Forum
Topic: PCG discussions
Replies: 13
Views: 30676

PCG discussions

QuoteOriginally posted by: zetaCentral to many precon schemes is incomplete LU which *isn't* easily parallelizable.I'm not familiar with any problems where ILU is the best preconditioner. Where would you use it?
by richardlm
August 7th, 2010, 10:40 am
Forum: Numerical Methods Forum
Topic: PCG discussions
Replies: 13
Views: 30676

PCG discussions

<t>Get yourself a copy of GreenbaumQuoteOriginally posted by: Dcole[...]However, it needs to be parallelized. I am having a bit of difficulty with this. Does anyone have any links to a good discussion of the PCG algorithm's parts, as well as what needs to be shared when doing it across multiple proc...
by richardlm
March 6th, 2010, 9:53 am
Forum: Numerical Methods Forum
Topic: MPI subarray problem
Replies: 7
Views: 33386

MPI subarray problem

<t>QuoteOriginally posted by: mblatt[...]Additionally, I would be carefull to use MPI_Type_vector with an array of arrays as it is only intended to be used with plain arrays. There might be additional padding between your matrix rows on some platforms.To avoid this you could create a struct type (MP...
by richardlm
March 6th, 2010, 9:17 am
Forum: Numerical Methods Forum
Topic: MPI subarray problem
Replies: 7
Views: 33386

MPI subarray problem

<t>When using MPI datatypes you need to commit them before you use them. See:MPI_Type_commitand clean up afterwards:MPI_Type_free.Your receive structure is not correct. 1) One send should be paired with one receive. 2) You have not allocated memory correctly for your receive. You are using the same ...
by richardlm
March 4th, 2010, 10:08 pm
Forum: Careers Forum
Topic: Newbie direct applications, am I the problem?
Replies: 28
Views: 35148

Newbie direct applications, am I the problem?

Don't give up on these quant-grad-schemes totally, they definitely do hire from them. From my experience, expect to wait between 1 and 4 months for first round interview invitations.
by richardlm
March 2nd, 2010, 10:40 am
Forum: Careers Forum
Topic: Credit Suisse QSI super day, what to expect
Replies: 6
Views: 34464

Credit Suisse QSI super day, what to expect

<t>I can't speak about Credit Suisse specifically, but you should probably expect several interviews (general stuff & motivation, brain-teasers, financial maths, algorithms, coding), group exercises, and tests (covering similar ground to the technical interview subjects).Just follow the usual qu...
by richardlm
February 24th, 2010, 3:26 pm
Forum: Careers Forum
Topic: What Do Quants Do?
Replies: 37
Views: 40457

What Do Quants Do?

On a related note, what is the difference between library quants and quant developers?
by richardlm
February 19th, 2010, 11:34 am
Forum: Programming and Software Forum
Topic: Database design
Replies: 38
Views: 43684

Database design

Does anyone have any experience of Berkeley DB?
by richardlm
February 17th, 2010, 8:57 pm
Forum: Programming and Software Forum
Topic: Efficient matrix multiplication
Replies: 13
Views: 38326

Efficient matrix multiplication

<t>QuoteOriginally posted by: DanWAre you sure you're not suck in a loop, this is what I'm using, think it multiplied a 50 x 50 matrix for about 1000 times in a few seconds[...]I think your problem is so small it fits in cache. Try again with this:QuoteOriginally posted by: bakaitI am trying to mult...
by richardlm
February 10th, 2010, 8:35 pm
Forum: Programming and Software Forum
Topic: Efficient matrix multiplication
Replies: 13
Views: 38326

Efficient matrix multiplication

<t>Strassen Algorithm requires fewer operations that standard matrix-matrix product implementations, but it works only works on square matrices of size 2^n. This can be fixed by adding empty rows and columns to make your matrices the correct size. To overcome the highly rectangular nature of your pr...
by richardlm
February 10th, 2010, 7:54 pm
Forum: Student Forum
Topic: MSc dissertation topic CUDA
Replies: 4
Views: 32987

MSc dissertation topic CUDA

Mike Giles (University of Oxford) works on this kind of stuff. You might want read his papers / citations / hyperlinks to get an idea of current research.
by richardlm
February 3rd, 2010, 7:29 pm
Forum: Programming and Software Forum
Topic: friendly latex editor
Replies: 28
Views: 43249

friendly latex editor

I use Kile, developed for KDE linux, although works elsewhere - including Windows apparently (still experimental though).
by richardlm
January 28th, 2010, 12:13 pm
Forum: Numerical Methods Forum
Topic: Dynamic ADI
Replies: 42
Views: 46600

Dynamic ADI

<t>Over lunch I started thinking about this, and in particular GS without synchronisation during the update sweep. This is as far as I have got:Consider a 1-D finite differenced Laplacian:Solve using GS: (Initial guess must be non-zero)Let n=8, a single sweep of GS is: for (int j = 1; j <= 8; ++j) {...
by richardlm
January 27th, 2010, 4:11 pm
Forum: Student Forum
Topic: Binomial Model
Replies: 11
Views: 33158

Binomial Model

QuoteOriginally posted by: richstudentukSo the values in the book are wrong??Don't think so, see "5.8 Other Choices for u, v and p"
  • 1
  • 2