optimization question

[azhar.abdul-quarder]

i have an optimization problem that i am trying to solve here and i want
your advice on what tools to use (GA's) and what environment to develop the
model.

Essentially the problem is to develop a portfolio of securities (mainly
fixed income and mortgage backed securities) based on user defined
constraints that would represent a CDO.

Now there are various constraints to optimize over - such as
      - overall size of portfolio
      - max size of single issuance
      - ratings stratification (for example user will define what % are AAA
vs AA vs noninvestment grade)
      - industry concentrations (MAX and MIN limits that user defines for a
group of industries)
      - weighted average spread minimum and weighted aver rating factor

each issuance has an associated spread (bps over libor) associated with it.

I've thought about this issue and here's how i think i'm going to tackle
this:
      - user input to be done in excel + vba
      - this will pass the constraints to a java engine or a c++ engine
(any comments on this?)
      -  the engine will use some sort of heuristic approach to finding the
optimal solution (or solutions)...
      - the java/c++ engine will pass back to excel the optimal portfolio
based on user constraints

now my question is:  rather than me trying to build this thing from
scratch, is there objects/code  i can use for the actual optimization
portion of this problem?   In general is there literature or example code
that i can base myself off and then customize to my own solution.

any comments would be much appreciated.

sincerely,

Azhar



*************************************************************************
This message and any attachments (the "message") are confidential and
intended solely for the addressees.
Any unauthorised use or dissemination is prohibited.
E-mails are susceptible to alteration.  
Neither SOCIETE GENERALE nor any of its subsidiaries or affiliates
shall be liable for the message if altered, changed or falsified.

*************************************************************************

[Joseph Wang]

The fun thing about the optimization problem is that the question of
what technique works best in what conditions is still a major research
question.  It's not like PDE's where for the most part, you know that in
this situation you do this.

      - this will pass the constraints to a java engine or a c++ engine

Yes.  I'd use C++ for this.  The minor problem is that java tends to be
slower for things that are heavily numerical.
The major problem is that C++ tends to be much easier to interface.  I'm
sure that the hot new language of 2020
is going to have a way of interfacing with C/C++.

    now my question is:  rather than me trying to build this thing from
   scratch, is there objects/code  i can use for the actual optimization
   portion of this problem?  

I don't know of any open source library (not so say there aren't any).  
The proprietary engines I know about tend to be tuned to a particular
problem.

   In general is there literature or example code that i can base myself
off and then customize to my own solution.

Tons of literature (one magic google keywork is "stochastic
optimization" .  Very little open source code.