I am a Ph.D. student who decided to change the direction midway. At a high level, I work on combinatorial optimization under uncertainty. In my proposal, I had said that I would use a technique A and its variants to tackle my problem. But now I have become uninterested in this direction.

  1. I want to broaden my combinatorial optimization problem to P' so that I can consider a technique B which is totally unrelated to A, and since I broadened the problem I cannot directly compare the results from A.

  2. Furthermore, I want to make some more changes in the problem to P'' and then use a technique C.

Now I have problems P,P', P'' which are variants of same combinatorial optimization problem under uncertainty but the techniques A, B and C are completely different and the results cannot be compared directly. Though I will compare A, B and C with techniques from literature separately. The underlying link is that all these problems are characterized by uncertain information.

Can I rope this into a thesis? Please note that I have finished my work with P, A and am starting to think about P', B. My concern is that there is a gap between these sub-topics and my Ph.D. thesis might lack cohesion. Suggestions/advice requested.

  • "Uninterested", as far as I know, is the same as "not interested". But "disinterested", on the other hand, means "I ain't got no dog in this race." I.e. a judge who presides when the XYZ Corporation sues the ABC Corporation should not be a stockholder in the ABC Corporation; rather he is supposed to be "disinterested". Commented Sep 21, 2017 at 19:28
  • @MichaelHardy what???
    – user58480
    Commented Sep 21, 2017 at 20:46
  • This is a standard distinction between "uninterested" and "disinterested". Commented Sep 22, 2017 at 1:25
  • 1
    OK, Thanks for correction. I am uninterested in that direction :)
    – user58480
    Commented Sep 22, 2017 at 3:23

3 Answers 3


From your description there seems to already be a fair amount of cohesion in your research. No need to apologize for it not being even more cohesive. If you spin it the right way, it is likely that no one (but you) who reads the thesis will think that there is anything at all odd about it.

In the introduction you could perhaps say something like: "In this thesis we explore 3 methods for optimization under uncertainty. The methods are used to solve 3 variations of problem ...". In a summary chapter you could explain why A is appropriate for P, B is appropriate for P' and C is appropriate for P". That type of thing might be important to know. For example, it is useful to know that a greedy algorithm works for the generic minimum cost spanning tree problem but is not appropriate when there are constraints on the maximum degree of the nodes in the resulting tree (since the latter is NP-hard).

  • Your welcome -- your question reminded me of my own thesis. I never solved the problem that I set out to solve but instead got sidetracked on better understanding the tool that my advisor recommended I use. I was able to use this tool to solve some other problems which were only loosely related to the original problem. It wasn't what I was aiming for, but I was still happy with the result. Good luck. Commented Sep 21, 2017 at 19:09
  • But still, your thesis solved one particular problem. I have already solved one problem using a method and published it. Now I am looking at a related problem but a different solution.
    – user58480
    Commented Sep 21, 2017 at 20:47

Ask your advisor. In most places, they de facto alone decide whether or not a thesis satisfies the program' requirements. Even if, in your place, they don't, the can give your valuable information as it is their job to advise you.

  • Unfortunately I am between two different advisors. In other words I have left one and in the midst of finding the next one. Cannot afford to do nothing in this period as funding is limited.
    – user58480
    Commented Sep 20, 2017 at 17:25
  • 1
    @Damodar Then talk to your chair.
    – Fomite
    Commented Sep 20, 2017 at 21:27

In general, there is nothing wrong with having a thesis with disjunct topics. How well it is received does matter on the field. In theoretical computer science it is rather common. We just had a defense last week with 3 completely disjunct topics. And the guy got a summa cum laude.

  • Is it OK if 30% of my thesis is theoretical and the rest is based on simulation and experiments?
    – user58480
    Commented Sep 22, 2017 at 3:26

You must log in to answer this question.