I was trying to solve a question on math.SE. I arrived at a solution whose validity I am not sure of. I do not wish to post it as an answer, even with caveats, as I believe it is ethically wrong to post a solution I am not even sure of. However, should I

1. post a new question (risk of duplicacy)

2. post my attempt with the caveats and let the community point out the dubious stuff in it? (downvotes, not posting "solution" but what I think is a solution)

3. post comment and expect a response?

For Further Details: The linked question is a definite integral. I tried solving by using the "differentiating under the integral sign" method. The thing is, I read that it is avoided in most texts because it is less of a method and more of a trick with various counter-examples and one had to tread carefully. Moreover, my answer seemed quiet different from what I get from mathematica, (i.e, when I do the indefinite integral and apply the funamental theorem of calculus manually. It sometimes work when mathematica wont budge over the definite integration). So I wish to know about the applicability of the method to this case, which would be my math.SE question.

• What about posting it as a community wiki answer. If this is not possible anymore you could flag for moderator attention and ask for community wikification. – Rasmus Mar 13 '11 at 9:37
• @Approximist: I think you misunderstood Rasmus. Rasmus suggests that you ask the question about differentiating under integrals signs as a community wiki. Rasmus is not suggesting that you make this meta discussion a community wiki. :-) If you are just testing to see if the mechanism works: yes, the mods saw your CW request. We will also see it if you ask a question on Main and flag it. – Willie Wong Mar 13 '11 at 15:37
• @Willie Lol! I got it now. – Please Delete Account Mar 14 '11 at 21:19

1. I don't see the problem with posting a solution you're not sure of as long as you indicate, somewhere in your answer, that you're not sure of it.

2. What's the difference between a method and a trick? There is a standard theorem which tells you a broad range of conditions under which differentiation under the integral sign works.

3. I think it would be fine to ask a question about which circumstances it's valid to use differentiation under the integral sign and then mention, by way of motivation, that you want to use it to answer another question (link to it).

• Re 2: the difference is whether the user is aware of the "standard theorem" :-). – Willie Wong Mar 13 '11 at 14:06
• @Willie: yes, so this is a perfect opportunity to ask a question about it! Actually, that might be a duplicate; I remember stating that theorem on math.SE before... – Qiaochu Yuan Mar 13 '11 at 14:09
• @Qiaochu can you link this theorem, thanks. – Please Delete Account Mar 13 '11 at 15:41
• Hum, do you remember what you wrote? Searching user:232 differentiating under integral sign does not turn up anything useful. – Willie Wong Mar 13 '11 at 15:41
• Ah, found it: math.stackexchange.com/questions/12909/… The third condition is the important one to check. – Qiaochu Yuan Mar 13 '11 at 15:46
• A related question has an example where it doesn't hold (from Counterexamples in analysis): math.stackexchange.com/questions/8711/… – Jonas Meyer Mar 13 '11 at 18:07
• @Qiaochu @Jonas Thanks. That clears it up. – Please Delete Account Mar 14 '11 at 21:18