Try Matlab, now that you can hook in GPU acceleration , makes it compelling in my opinion

<t>Some random thoughts I think the more important question is which language is more suitable to parallel or distributed computing as that will be the dominant theme in the next few years. C++ is the established candidate I believe with packages like MPI etc. Going forward I think Java would be mor...

<t>95% of jobs do quant dev work in C++ , 5% in Java , if you want to get a job you are better with C++ , having Java (10 yrs plus and M.Sc Applied Math ) hasnt worked for me trying to get a quant dev job. Sad fact of life. Sorry. Going forward CUDA looks like an interesting development , keep it on...

<t>Have been programming in Java for 10 years, IMHO Eclipse is the industry standard and 90% of work places will use it as the de facto development environment. I like it. IntelliJ used to be my favourite but almost all the places I worked were too cheap to buy it. I would advise using Eclipse , it ...

It seems that all the serious number crunching in investment banks is done using c++. Model may be protoyped in Matlab and then put into production using C++ . IMHO C++ (which says a lot as I am a Java Programmer )

<t>You can use java webstart , this will give you a fully fleged deskop application that you can distributevia the web. It does not have the restrictions of applets. It does require the host to have java 1.5 installed (or java 1.4 with the web start module added on ) . You could also use Macromedia ...

<t>If you want to do quant related work i.e number crunching I would recommend C++ VB and Excel If you want do enterprise trading applications I would recommend Java for everything but communicating with a quant library ( and thats because Java has not got its foot in the numerical computing door) ....

<t>Has anyone seen this book yet , there are different publication dates Java Methods for Financial EngineeringApplications in Finance and Investment by Philip Barker ISBN: 1852338326Who is Philip Barker , does he have a webpage , I would like to take a look at the code .Is the book any good ? df </t>

I have converted about 85-90 of this library to Java. If you are interestedI will send it on if you are interestedhttp://finance-old.bi.no/~bernt/gcc_prog/dfmax

apache has a ligthweight library , http://jakarta.apache.org/commons/math/

<r>a google search using "finite difference" as the exact phrasehttp://<URL url="http://www.amath.washington.edu/~rjl/pubs/am58X/amath58X05.pdfhttp://www.caam.rice.edu/~caam452/Numerical"><LINK_TEXT text="www.amath.washington.edu/~rjl/pubs/am58 ... /Numerical">www.amath.washington.edu/~rjl/pubs/am58...

Thanks for all the pointers , but a novice question , how do I calculate the coeffiecients of an AR(1) method and an AR(2) method ? similarly for an MA(1) and MA(2) etc given a set of raw observations x1,x2, x3, x4 etc , Sorry but the penny is not dropping ..dfmax

<t>I have been reading up on basic time series AR, MA , ARMA, ARCH , GARCH etcI found that most books and notes just lay out the formulas with scant attention to the implementing the method on a sample data set, i.e how do I kick off the basic calculation of the for AR etc, Does anyone know of any b...

I liked this one : C++ for Scientests, Engineers and Mathematicans. by D. M Capper

http://web.mit.edu/2.151/www/Handouts/C ... .pdfbetter still I think