My Ph.D. dissertation was in "Decision Making Under Uncertainty". The topics in the OP are extensive with most of them looking good. E.g., I noticed interpolation -- an old idea but sometimes important. Apparently missing, however, was the old technique of <i>scenario aggregation</i>.<p>Before going very far on this subject, we should note there was a lot of attention to this subject at the ORFE department at Princeton.<p>Some of the related advanced math is pretty, but for practice here are some of the real issues:<p>(1) Far too easily, needed computer resources, both CPU cycles and memory sizes, especially main memory, are high beyond reason, feasibility, practicality. Without exaggeration, this subject is Godzilla for any super computer -- right away gobbles up and spits out -- of less than 100 million square feet of floor space.<p>(2) Too often this methodology is what would do -- provably best possible for any means of processing the input data -- if had a lot of data and assumptions about that data that don't have.<p>(3) So, in practice, the first obstacle is to find a problem that gets around (1) and (2) and is doable otherwise.<p>(4) In practice, the next obstacle is convincing the <i>suits</i>, who are nearly always wildly afraid of anything so new to them, to devote any time, money, data, or attention to the work. The <i>suits</i> too easily see any such project as a lose-lose for them: (A) If the project works and with quite valuable results, the project head might get promoted, maybe over the relevant <i>suits</i>. (B) If the project doesn't work really well, generally it will be seen as a waste with the <i>suits</i> having a black mark on their records, a black mark that makes them easy targets in the C-suite competition.<p>One possible path: If have a project that passes (1) and (2), in addition have one more attribute: Have the project one (A) can fund and do oneself, (B) have a good list of good candidate customers, and (C) do a corresponding startup. Due to (B), are just a vendor to the customer organization and, thus, not a threat to the C-suite <i>suits</i>. And to the customer, the project has no "waste" since if the customer is not happy the startup doesn't get paid.<p>Net, do a startup.