Unbundling C++ Quant Libraries on Amazon Web Services

Brian McCallion Amazon Web Services, Cloud Computing, Financial Markets, Lambda Leave a Comment

Unbundling Large Quant Libraries into the Compute Fabric of Amazon Web Services

I want to test some QuantLib https://en.m.wikipedia.org/wiki/QuantLib
functions today. I did some work with Emscripten   to transcompile QuantLib from C++ to JavaScript. I’m pretty close to getting that to work or learning of any hard limits. I was inspired when I learned 20MM lines of Doom C++ code have been compiled into a working Javascript version of the game. Bullet Physics has also been successfully transcompiles from C++ to Javascript. I’ve been able to transcompile the Boost C++ library to Javascript, and just need to do the next step of using Javascript Boost to compile javascript Quantlib. One team worked to rewrite QuantLib in Java. Another approach is to use SWIG. I haven’t tested using SWIG to run QuantLib wirh Java. From what I read there are some issues with QuantLib running multithreaded. I’d like to see how Lambda functions, called recursively would work and if maybe there’s a way to use this technique to price derivatives really quickly.

I’m actually far more interested in creating a set of useful primitive Lambda functions than I am in transcompiling code or the physics of the functions themselves. People with far more experience than I have can and so these calculations each day.

My vision is a core set of Lambda functions that will enable developers to quickly build financial applications on AWS. What I expect to build are the apis and the user interfaces to organize the data, the streams, the access control, the workflow, and the testing of such functions and applications composed of these functions.

alarm clock

Update December 2015