Talk:Multi-index notation

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Derivative Notation[edit]

The article uses the :D notation for derivatives. In most microlocal analysis literature :\part is used instead, and :D means something else. Specifically;

\part^{\alpha} := \part_{1}^{\alpha_{1}} \part_{2}^{\alpha_{2}} \ldots \part_{n}^{\alpha_{n}} where \part_{x_i}^{m}:=\part^{m} / \part x_{i}^{m}

And :D is defined by;

D_{x_i}^{m}:=(-i)^m \part^m / \part x_i^m

So

D^{\alpha} := D_{1}^{\alpha_{1}} D_{2}^{\alpha_{2}} \ldots D_{n}^{\alpha_{n}}:= (-i)^{|\alpha|} \part^\alpha

However, I'm not sure if this is how it is in other branches of mathematics. — Preceding unsigned comment added by 193.1.100.110 (talkcontribs) 09:34, 4 September 2006 (UTC)

Low-level pedantry[edit]

I think it's just worth noting that there is room for confusion between your use of integers and your use of the  x_{i} . If you stated that \part_{i}^{m}:=\part^{m} / \part x_{i}^{m} rather than that \part_{x_i}^{m}:=\part^{m} / \part x_{i}^{m}, this would remove the potential for confusion regarding your intended relation between the  i and the  x_{i}.

Similarly, less confusion might be caused by stating that D_{i}^{m}:=(-i)^m \part^m / \part x_i^m instead of stating that D_{x_i}^{m}:=(-i)^m \part^m / \part x_i^m. However, as long as everyone knows that you're using the i to index, say, the dimensions or variables that partial differentiation is occurring with respect to, then no real confusion arises.

AnInformedDude (talk) 20:59, 2 September 2012 (UTC)

Someone's opinion of multi-indices[edit]

Multi-index notation made simple formulas very complex, and very complex formulas no understandable, like tensors on lections. We see many times as mathematicians start confusion, if they use it. So, don’t use it if you can. — Preceding unsigned comment added by 65.95.80.51 (talkcontribs) 00:48, 16 May 2008 (UTC)

For serious manipulation of multi-dimensional calculus formulae, multi-indices are very useful. Hopefully the example proof on the page shows why this might be (although it's a pity that it isn't something proved without breaking down into components). Quietbritishjim (talk) 14:29, 20 June 2010 (UTC)

I have to agree, it's essential if you are deriving things in a dimension-independent way, it collapses the sum or product over dimensions, much like tensor notation. It can be confusing if you don't know it's being used, or an introductory article (like this one!) isn't linked to. — Preceding unsigned comment added by Drhansj (talkcontribs) 18:32, 30 December 2010 (UTC)

Significant room for confusion and mis-understanding with multi-indices[edit]

Granted, when properly explained in such a way that ALL of the conventions are spelled out, then multi-indices can make for compact notation (though whether they are actually useful in terms of problem-solving strategy is likely a different matter, barring the issue of the condensed notation - see postscript).

For example, in the Multi-binomial theorem example on this page, the idea of finding the n-tuple power of the sum of two n-tuples is not necessarily clear (until you make the observation that a component by component product of the component-wise exponents is what is intended by such notation). As it stands, you have to direct to another article (Multi-binomial theorem) to see the convention which is being employed in this case. So this article does not seem to be stand-alone, even though it goes a good way to explaining the multi-index notation.

PS - Multi-index notation is NOT a technique as such (though neither is Einstein summation convention - though this does save quite a bit of writing).

82.3.199.73 (talk) 21:24, 2 September 2012 (UTC)

It is clear what (x+y)α means because if x and y are vectors then obviously so is their sum, and the meaning of a vector to the power of a multi-index is defined further up the article.
The text added is redundant since α and β were already defined to be multi-indices near the top of that section. The fact that |α| is a natural number is clear from its definition just above, but maybe it could do with being mentioned again when k is defined. I'll move the definitions into their own paragraph to make them slightly harder to miss. Quietbritishjim (talk) 00:27, 3 September 2012 (UTC)
I suppose that, in the interests of clarity, it might be theoretically advisable for all the vectors AND multi-indices to be underlined so that they are a vector. Of course, the fact that the article doesn't do this seems to indicate that we are following an implicit convention - that everything's either a multi-index or a vector (rather than a scalar). Of course, you could always say that the notation works for multi-indices, vectors And scalars - but then the onus is on you to make EXPLICITLY clear which of the two you are using under the circumstance. As we are using this article to introduce a definition - how can one justify not making things absolutely clear, providing emphasis to enable more people to understand what is being done? I suppose that, within a specific numerical example, such as  ((1,1,1)^{T})^{(1,1,1)}, it would be clear. In the case of the introduction to the multi-nomial coefficient, you have already described k as being an integer BUT it does no harm to say that this distinguishes from your use of all the other variables (given the implicit convention).
AnInformedDude (talk) 14:04, 10 September 2012 (UTC)
Wow, this was more than two months ago? I'm so sorry AnInformedDude, I've been meaning to reply. Anyway, here's a reply, at last. I left a message on your talk page in case you're not subscribed here.
You definitely seem to think that the article isn't that clear at the moment, but I'm not sure I understand all specific points you make. Here are replies to two things that I think you're saying.
One is that vectors and maybe maybe multi-indices should be underlined (actually in typeset maths we use bold rather than underline). Writing vectors in the same font as scalars is pretty common in higher level maths, although it's often coupled with a convention that scalars are Greek, which isn't followed here. I'm happy if someone goes through the whole article making the vectors bold (and even multi-indices, although I don't think that would help). Alternatively, I wouldn't mind if scalars were changed to Greek so long as multi-indices were made into Roman letters (maybe early ones a,b,... leaving vectors as x,y,...). But I don't have the energy to do either of these myself.
I think another problem you're having is that the meaning of the letters is defined at the top of the section, rather than right next to where they're used. I can see that if you skim the article quickly that you might miss this, but I'm against putting the definition of each letter next to every use of it because I think you'd end up with a sea of repeated definitions with the real information buried inside, and then no one's going to be able to skim it. But I did add a note to the multi-binomial theorem, because I can see that that might be particularly confusing considering the binomial theorem just above it. Quietbritishjim (talk) 17:44, 19 November 2012 (UTC)

Why is the multi-index binomial theorem missing?[edit]

Seems obvious to me that the multi-index version of the binomial theorem:

(x + y)^{s} = \sum_{r \le s} {s \choose r} x^{s-r} y^r

should be included here and probably linked to from the binomial theorem, multinomial theorem and maybe a few other pages. — Preceding unsigned comment added by Drhansj (talkcontribs) 18:28, 30 December 2010 (UTC)

The reason it's missing is just because nobody has bothered to add it. You could have done, there's no need to ask on the talk page first. Be bold! Anyway, I went ahead and added it, thanks for the suggestion. I haven't yet updated the binomial theorem article, but I agree that should be done too (at the moment it doesn't even mention this form of the theorem). (By the way, don't forget to sign comments on talk pages by putting four tildes ~~~~ after them.) Quietbritishjim (talk) 23:36, 23 July 2011 (UTC)

Ok, updated binomial theorem article (see binomial theorem#multi-binomial). After I finished doing all this I realised that "multi-binomial theorem" was a title that I'd sort of made up myself, and doesn't show up in a Google search, although I'm sure I've heard it in lectures. If someone can think of a more appropriate name, please feel free to rename that section in binomial theorem and the heading here. Quietbritishjim (talk) 12:22, 24 July 2011 (UTC)