I'm interested in pursuing studies of general problem solving.

I'm particularly interested in writing general programs which can write specific programs for specific problems.

I enjoy the idea that most if not all problems can be modeled as follows: There is some thing, defined as a list of predicates. It is either in a state that is wholly, partially, or not at all known. There are some list of transformations that change what predicates are in a state of accept or decline, or that change the vector value of these predicates, via some ruleset that is either partially, wholly, or not at all known. The goal of the problem is to take the thing from its state to some other state(either wholly or partially defined, or with maximization in mind) using some combination of transformations.

Is OR a good field for me given my interests and preference for general modeling and solving? If not, is there a field you can recommend.

asked 12 May '15, 14:25

battlefrisk's gravatar image

battlefrisk
3914
accept rate: 0%

I don't see how OR can be simply "right" or "wrong" for anyone. Do you have a decision you are trying to make? If you can ask a detailed question about a specific decision, you will be more likely to get some help.

(15 May '15, 11:55) 4er

I'm trying to decide between majors. Lets say I wanted to write a chess program without any prior experience with the game. Would I be better off if I had experience in OR OR computer science?

(15 May '15, 13:27) battlefrisk
1

Your interests sound a lot more like some branch of computer science than anything common to OR. But in any case, for choosing a major it is a good idea to look closely at the experience that you would actually get from your major at the college or university where you are actually studying.

(15 May '15, 22:16) 4er

12next »

If you like and are (or can get) good at mathematical modeling and problem solving, and have an analytical bent, OR might be a good field for you.

I don't understand all your predicate jazz, so I don't know what it does or could amount to in practice. But perhaps you'd be bringing a fresh perspective and tools to OR, which might have potential for making advances in the field.

link

answered 12 May '15, 17:20

Mark%20L%20Stone's gravatar image

Mark L Stone
447310
accept rate: 15%

Is the work mostly in the business sector or do OR work in social engineering or physics based engineering at all?

(12 May '15, 18:20) battlefrisk

With an OR education , you could certainly work in any of those areas. You might find the work to be most exciting and interesting if it's not on well-trodden ground, i.e., boldly go where no (or not many) man (or woman) with your background and abilities has gone before. As an OR analyst, you can work as part of a team with subject matter experts in other areas, such as physics based engineering, and help to structure, formulate, and solve problems, based in part on information you're able to elicit from them, but they don't know how to use as profitably themselves as with your help.

(12 May '15, 20:04) Mark L Stone

What you describe to me sounds like Machine Learning, but the wrinkle of using predicates rather than simpler inference mechanisms is going to make your ideas very difficult to analyze using the tools from that domain.

You will certainly find the machinery of optimization and simulation useful. Depending on what is available at your university, OR may be a good major, but you may also simply add the "Nonlinear Optimization for Poets" class, i.e., (Convexity, KKT conditions, Lagrangean Duality) to your plans, and whatever else you need to be successful at that course.

The following will make a much bigger difference: find a faculty member at your institution who can help you put your ideas into a scholarly context, after listening to them in more detail than what is practical for a forum like this. They can then carve out a very small problem related to what you describe and provide you with guidance so that you can produce something (ideally a paper, but not necessarily) that their colleagues would recognize as a marker of quality work for your level.

Through the process of creating that work, you will better understand, and become able to better express, what you really want to do. You will also begin to learn who are the experts most likely to help you advance your ideas. You can then use the work you produced, and the reference from the faculty member, to introduce yourself to those experts, and differentiate yourself from the hundreds of applications they get every year.

Whatever area those experts (sometimes rather accidentally) are in is the area you will do your graduate work in. Your specific undergraduate major will help, but less than what you seem to think. Any shortcomings in your background will be addressed in your first year of graduate school.

link

answered 17 May '15, 00:38

Leo's gravatar image

Leo
1.1k17
accept rate: 8%

1

Alright thanks Leo. I suppose for now then it just makes sense to get an introductory OR book and give it a go. I'm certainly not at a position where I can declare my major yet, and if it's true that it's not that important if I intend to go to graduate school, then I think my time will best be served studying instead of worrying. Thanks everyone from your replies. I will be in touch with this community once I begin my study!

(21 May '15, 18:20) battlefrisk

Just curious, what year (of college?) are you in? I would think most colleges would let you declare a major by the end of your freshman year and maybe anytime sophomore year. Or are you still in high school, in which case you sound awfully advanced?

(21 May '15, 19:54) Mark L Stone

I'm actually a drop out. Long story short, i picked a very difficult major for all the wrong reasons. When i couldn't skate through my upper divisions the way i did in my lower classes, i dropped out. After which i did some soul searching and figured out roughly what i actually want to do, but i haven't yet put a name to this "thing" which i seek to pursue. Hence why I'm inquiring about these different majors which seem to be related to my interest.

link

answered 24 May '15, 16:02

battlefrisk's gravatar image

battlefrisk
3914
accept rate: 0%

Keep on keeping on. It sounds like you've now attained a level of maturity which should serve you well in round 2.

Hint: Unless you are extraordinarily gifted, don't pursue a path of Bachelors to Ph.D. studies in a mathematically intense field if you want to just "skate through".

(24 May '15, 19:29) Mark L Stone

"To drop out" is a verb, not a noun. Leave it in the past tense. You dropped out, and now you are looking for the next step. There are faculty at your school who really care (and some who really don't - you'll learn to tell the difference) and who can help you more than a forum can. You are on a long journey, and you will make plenty more mistakes. Being in the ballpark as far as topic helps, but it matters much more to take every class seriously and impress people with quality work. That is what will create the 2nd and 3rd chances you need to maximize the utility of your outcomes.

(24 May '15, 19:37) Leo

I don't intend to skate through. I actually intend to be ahead of all my classes from day one, since I will have one to two years of study before I even begin my bachelor's program.

I actually don't mind using it as a noun to define myself (at this time). In a world where so many people miserably and thoughtlessly trudge through education because it's the beaten path, I think it's indicative of what makes me different in a good way.

link

answered 25 May '15, 12:33

battlefrisk's gravatar image

battlefrisk
3914
accept rate: 0%

edited 25 May '15, 12:34

I was impressed that a young man searching for ideas for their future reads discussions in this forum. In my opinion it is worth to hone their skills and exchange their views with people whose knowledge and experience is richer than ours. Unfortunately, this discussion requires a certain knowledge to be gotten in school. Good luck at learning.

Referring to your last sentence - I have just saw a documentary HBO by Andrew Rossi with 2014, which addresses the issue of shape and cost of higher education, shocking.

(25 May '15, 14:11) Slavko

Thanks slavko! Yes it seems the monetary cost to benefit ratio of education is, for the most part, progressively moving in a negative direction. But as more and more jobs become automated and more and more people get degrees what are we to do?

When I have a better understanding of the language of the field, I intend to utilize this site a lot.

For now, I'm taking book recommendations here (https://www.or-exchange.org/questions/12345/introductory-text) if any of you are interested.

link

answered 25 May '15, 15:54

battlefrisk's gravatar image

battlefrisk
3914
accept rate: 0%

If you're a good self-learner, with the wealth of high quality, legal, and free material available online nowadays (including some excellent O.R. related textbooks, such as http://stanford.edu/~boyd/cvxbook/bv_cvxbook.pdf and http://statweb.stanford.edu/~tibs/ElemStatLearn/printings/ESLII_print10.pdf ), all you need is an internet connection. So for the cost of a PC, internet connection (or go to the library), and electricity to run your PC, you can learn as much or more as at any school. How many people fall in this category? Not many.

(25 May '15, 17:12) Mark L Stone

By the way, although the books I linked in the preceding comment are excellent books, I think they will be too advanced for you at your current state. Nevertheless, go ahead and download them, and you can at least read the first several pages and skim through the rest to be aware of the fruits which await you.

(25 May '15, 17:16) Mark L Stone

The key word is a e-learning. The best universities offering interesting courses free of charge. As an example You can see: http://ocw.mit.edu/index.htm or http://web.mit.edu/15.053/www/AppliedMathematicalProgramming.pdf.

(26 May '15, 03:08) Slavko

Thanks Mark! I'll give it a shot. I noticed the introductory section talking about convex sets. I take it an introductory understanding of set theory would be beneficial?

link

answered 25 May '15, 21:33

battlefrisk's gravatar image

battlefrisk
3914
accept rate: 0%

I don't think you need to have studied set theory. I think your linear algebra background is quite weak for the Boyd book. Also, it will be tough sledding without having taken a course in Real Analysis (see below). If you see equations, you might not really understand them. Still, there's nothing wrong with giving it a try. If you get to a tough section, trudge though, and you may get to an easier one. The acid test is whether you can work the problems. They are probably more difficult than you're used to. Some are quite difficult. (continued)

(25 May '15, 23:29) Mark L Stone

You should probably start with Appendix A - I think the chances that you understand the material in that appendix to be essentially nil. Your life will be easier if you are a MASTER of the Strang book up through eigenvnalues and eigenvectors. Hint: Strang and its problems is many times easier than Boyd.

You probably also need to study some elementary probability theory. You need to know about random variables, independence, expected value, variance, probability density function, and (cumulative) distribution function. You should probably take a course in probability from your school.

(25 May '15, 23:29) Mark L Stone

It will also help if you take a course in Real Analysis (redoing calculus correctly, not with symbol manipluation and hand waving). That will help you in all your future endeavors (optimization, probability, statistics, and much more).

(25 May '15, 23:31) Mark L Stone

After having taken hard core math classes as an undergrad, I just dived into Ph.D. level O.R. courses as an upper class undergrad then grad student. I skipped the introductory O.R. classes. So I never even read more than a few pages of the Hillier and Lieberman book, for instance, despite taking a research seminar course from Hillier. But in your case, if you want to get a flavor for the field, it may be the way to go for you.

(25 May '15, 23:46) Mark L Stone

alright so then your recommendations are linear algebra, analysis, and probability theory. Sounds like a good start. and I'm definitely gonna check out that Hillier and Lieberman book. Thanks guys, I think I'm all set now.

link

answered 26 May '15, 13:05

battlefrisk's gravatar image

battlefrisk
3914
accept rate: 0%

edited 26 May '15, 13:07

Well, that's my view, anyway. Good luck.

(26 May '15, 14:47) Mark L Stone

Your original description of the type of problems you are interested is very similar to what is being done in the field of Automated Planning. In fact most of your terms map on a similar concept in Automated Planning:

predicates                                   => state variables 
state                                        => state 
state wholly, partially, or not at all known => (partial) observability, contingent planning
transformations                              => actions 
ruleset                                      => action conditions and effects  
goal is to take the thing from its state to some other state => goal 
combination of transformation                => plan

And for sure, Automated Planning is using OR techniques

link

answered 27 May '15, 11:38

Philippe%20Laborie's gravatar image

Philippe Lab...
623
accept rate: 9%

Constraint Programming anyone? https://www.ibm.com/developerworks/community/forums/html/category?id=33333333-0000-0000-0000-000000000268

Note: this does not constitute a recommendation one way or another for the product which is the subject of the forum in the link.

(27 May '15, 14:44) Mark L Stone

wow thank you so much philllip i think this is in large what I'm looking for. Is there a way you would recommend getting into the field of study?

link

answered 01 Jun '15, 19:05

battlefrisk's gravatar image

battlefrisk
3914
accept rate: 0%

A classical textbook on Artificial Intelligence Planning is :

Automated Planning: Theory & Practice (The Morgan Kaufmann Series in Artificial Intelligence) by M. Ghallab, D. Nau, and P. Traverso (Elsevier, ISBN 1-55860-856-7) 2004.

There are some online courses on the topic. For instance this one, based on the above mentioned book: https://www.coursera.org/course/aiplan

link

answered 02 Jun '15, 08:10

Philippe%20Laborie's gravatar image

Philippe Lab...
623
accept rate: 9%

Your answer
toggle preview

Follow this question

By Email:

Once you sign in you will be able to subscribe for any updates here

By RSS:

Answers

Answers and Comments

Markdown Basics

  • *italic* or _italic_
  • **bold** or __bold__
  • link:[text](http://url.com/ "Title")
  • image?![alt text](/path/img.jpg "Title")
  • numbered list: 1. Foo 2. Bar
  • to add a line break simply add two spaces to where you would like the new line to be.
  • basic HTML tags are also supported

Tags:

×28
×9

Asked: 12 May '15, 14:25

Seen: 1,712 times

Last updated: 16 Jul '15, 11:17

OR-Exchange! Your site for questions, answers, and announcements about operations research.