Matlab for Beginners

tristanreid · Post by **tristanreid** » Fri Aug 07, 2009 6:13 pm

if I had been introduced to Matlab in high school I would have paid more attention in Math rather than playing tic-tac-toe on the desk with the chick beside me

No, you probably wouldn't have. Wink

-t.

doreilly · Post by **doreilly** » Fri Aug 07, 2009 6:42 pm

-- if I had been introduced to Matlab in high school I would have paid more attention in Math rather than playing tic-tac-toe on the desk with the chick beside me.

They are not mutually exclusive

Tic-Tac-Toe

dgn2 · Post by **dgn2** » Fri Aug 07, 2009 7:48 pm

yeah t you are right!

hedgeQuant · Post by **hedgeQuant** » Thu Aug 13, 2009 6:45 am

@original poster:

Are you looking to trade equities with Matlab ? A lot depends on what exactly you want to do. But I do second the opinion about E. P. Chan's blog. Feel free to post questions about any particular issues you encounter.

To continue on the Matlab / R discussion.....

I have been using Matlab and C for the last 15 years. I have dabbled a little bit in R for the last 2 years. Learning LISP is something I have been trying; not successful so far.

First of all, Matlab wins when doing vector and, more importantly, matrix operations. A matrix is a very powerful data structure. A simple example is that one can represent complex numbers and quaternions using real 2 x 2 matrices. (This is a very useful feature in Signal Processing and Communication). In general matrix representations are very, very useful in analyzing many engineering problems. Matlab has a ton of matrix related functions that make it the best choice for engineering applications. Most of the signal proceesing, communication and control engineers I know love Matlab.

When it comes to statistics, I think I like R a little better. (But instead of saying a=3 I have to type in a <- 3. I hate the extra key stroke). Main reason is that many stat. text books have R code to illustrate examples.

When it comes to AI, I think Lisp is a natural choice. Again many AI books use LISP to illustrate the concepts. So it makes life easier if ones knows some LISP when learning AI.

As far as the expense of Matlab goes, I think it is a drop in the ocean when compared to rest of the costs in running a Hedge Fund operation.

As far as multicore goes, I think Matlab has some operations that use all the cores (within the same computer). For example when doing a big QR decomposition I have noticed that all the cores in my quad core server are used. I am running Matlab on Windows 64-bit machine; I do not have the parallel computing toolbox.

Edit: I checked the Mathworks site. Matlab has "implicit multiprocessing" for some functions (including QR). This does not require the Parallel Computing toolbox.

urnash · Post by **urnash** » Thu Aug 13, 2009 9:01 am

hedgQuant: But instead of saying a=3 I have to type in a <- 3. I hate the extra key stroke.

[quote]

Ehh, you don't need to, a=3 works as well. However one of these days you are going to type a==3 and that's something different (as is a < -3 Smiley )

[/quote]

[EDIT] prophet: Almost all these functions are overloaded such that they accept combinations of vectors, matrices and can selectively operate across specific dimensions of a multi-dim array. I've not ever seen a language that does all that.

[quote]

Most good R functions accept a vector whereever possible. There is always Vectorize() for those that don't -- but I must admit that Vectorized functions are slower than well optimised R functions.

[/quote]

hedgeQuant · Post by **hedgeQuant** » Thu Aug 13, 2009 4:18 pm

Ehh, you don't need to, a=3 works as well. However one of these days you are going to type a==3 and that's something different

Thanks Urnash. Looks like I need to hit the R books harder. But why do R books always use the a<-3 notation ?

My usual mistkae is:

if(a=10)

printf('a=10\n');

And I wonder why the statement works when a is nowhere close to 10 ${L_SMILIES_SMILE}$

urnash · Post by **urnash** » Thu Aug 13, 2009 4:35 pm

Ehh, no idea. Wasn't it Pascal where they used :=? But I like it. You can even do 3 -> a if you want....

rgj · Post by **rgj** » Thu Aug 13, 2009 5:01 pm

In the good old days of R the assigment operator was _ (I think). So, e.g., a_3 produced the desired result. Ahh the memories...

MrMagoo · Post by **MrMagoo** » Thu Sep 03, 2009 2:04 am

I love matlab, but i agree R is better at some things. The problem is i have no time or patience to learn the R syntax. So i only used R when strictly necessary, and with great pain. As i have gained access to insightful's S+ , my R use is rare now, since S+ libraries reduce the learning curve. Even now, i use 85% matlab and 15% R.

Dont know if you already seen this, but rapidminer may well compete with matlab or R in some specific datamining/neuralnet processes, and is extremely easy to use. Its free, and i did a couple of cool financial models with it.

rapidminer

Ice Viking · Post by **Ice Viking** » Fri Sep 25, 2009 12:03 pm

Matlab.webinars.( using the "." does that hide content from google searches? or am I a twat?prob both)

If you're like me, you'd like to be able to download the matlab.webinars for viewing at your preference. The site won't let you do this but there is a simple work around if anyone is interested. Email in profile