# Confusion about the Hamel Basis

Alright, so I'm reading a book on Hilbert spaces and functional analysis, and here it defines a "Hamel basis" to be a "maximal linearly independent set". I take this to mean that $S$ is a Hamel basis if $S$ is linearly independent, and there is no linearly independent set $T$ for which $|T|>|S|$.

On the other hand, this article defines (on page 6) a "Hamel basis" to be a basis in which "we do not allow infinite sums" (i.e. in which every element can be expressed as a finite linear combination of basis vectors).

Also, the distinction they make between an orthonormal basis and a Hamel basis makes me think that an orthonormal basis does allow infinite sums, but their definition of "basis" on the first page requires finite linear combinations as well.

I'm just really confused, I can't tell which source I should trust and can't seem to find a single resource that gives the "correct" answer (not sure if I want to trust Wikipedia, though they seem to make the distinction too when defining the Schauder basis). What's the correct definition of a Hamel basis? Does the general definition of a "basis" allow for infinite linear combinations, or only finite ones? What about orthonormal and Hamel? Why would the book I'm reading define a Hamel basis in the way it did? What's even the point of making the distinction between a Hamel basis and any other basis; is a Hamel basis also orthonormal?

To completely avoid ambiguity, these are the rigorous definitions I'm using:

Linearly Independent: A subset $S$ of a vector space $V$ over a field $K$ is linearly independent if for every finite subset $G\subseteq S$, the only sequence $\{\alpha_s\}_{s\in G}$ to $\sum_{s\in G}\alpha_ss = 0$ is $\{\alpha_s\} = 0$.

Basis: A subset $S$ of a vector space $V$ is a basis if it is linearly independent, and every vector $v\in V$ can be expressed as a finite linear combination of vectors in $S$.

Orthonormal Basis: A subset $S$ of a Hilbert space $\Bbb{H}$ is an orthonormal basis if it is pairwise orthogonal ($e_1,\ e_2\in S$ and $e_1\neq e_2\rightarrow e_1\perp e_2$), every element is a unit vector, and every element of $\Bbb{H}$ can be expressed as a linear combination (finite or infinite) of elements of $S$.

Hamel Basis Definition 1: A Hamel basis for a Hilbert space $\Bbb{H}$ is a set $S$ such that $S$ is linearly independent, and there is no linearly independent set $T$ with $|T|>|S|$.

Hamel Basis Definition 2: A Hamel basis for a Hilbert space is a (potentially orthonormal?) basis under finite linear combinations.

• Hilbert spaces almost never have orthonormal Hamel bases since convergent infinite sums like $\sum_{k=1}^{\infty} \frac{1}{k^2} e_k$ can be written down that have nonzero scalar product with infinitely many basis vectors $e_1,e_2,e_3,...$. Probably only exist when it's finite-dimensional. – user399601 Dec 28 '16 at 5:55
• @EricWofsey Yes I do, thanks! – user3002473 Dec 28 '16 at 5:59
• Maximal does not mean that the cardinal is maximal: it means that the Hamel basis is not properly contained in any other linearly independent set. – Mariano Suárez-Álvarez Dec 28 '16 at 6:00
• – Michael Hoppe Dec 28 '16 at 11:42

Your definitions of "linearly independent", "basis", and "orthonormal basis" are all correct. In particular, an orthonormal basis for an infinite-dimensional Hilbert space is not actually a basis (since you will need to use infinite linear combinations).

"Hamel basis" means exactly the same thing as "basis". The reason that it is given a different name is to emphasize that you are talking about a basis with respect to finite linear combinations, as opposed to some other kind of object that might be referred to using the word "basis" but which is not actually a basis (such as an orthonormal basis or a Schauder basis). Indeed, when talking about infinite-dimensional topological vector spaces, it is rare that you actually care about a basis as opposed to some related notion that allows for infinite linear combinations. So in most contexts if someone refers to a "basis" of such a space, it is actually more likely than not that they are abusing terminology and are using "basis" as an abbreviation for "orthonormal basis" or something similar. To make it clear that you literally mean just a basis, it is common to say "Hamel basis".

As for your book's definition, note that "maximal" means "cannot be enlarged to a superset", not "cannot be enlarged in cardinality". So a "maximal linearly independent set" is a linearly independent set $S$ such that there is no proper superset $T$ of $S$ which is linearly independent. This is equivalent to saying $S$ spans the whole space (using finite linear combinations). Indeed, if $S$ does not span the whole space, you can take any vector not in its span and add it to $S$ to get a larger linearly independent set. Conversely, if $S$ does span the whole space, any vector not in $S$ is a linear combination of elements of $S$ and thus would give a linearly dependent set if you added it to $S$.

• That certainly makes a lot more sense, thank you! Any insight on the book's definition of Hamel basis? It's still unclear to me how it might be equivalent to the Hamel basis you and the commenters described. – user3002473 Dec 28 '16 at 6:12
• I've added a final paragraph on that. – Eric Wofsey Dec 28 '16 at 6:17

I take this to mean that $S$ is a Hamel basis if $S$ is linearly independent, and there is no linearly independent set $T$ for which $|T|>|S|$. That is a flaw in your "Hamel basis definition 1".

That is wrong. Instead of $|T|>|S|$ you should say $T\supsetneq S.$ That would be equivalent if $S$ is finite, but you cannot assume that here.

On the other hand, this article defines (on page 6) a "Hamel basis" to be a basis in which "we do not allow infinite sums" (i.e. in which every element can be expressed as a finite linear combination of basis vectors).

Exactly. That's the definition of linear independence.

Also, the distinction they make between an orthonormal basis and a Hamel basis makes me think that an orthonormal basis does allow infinite sums, but their definition of "basis" on the first page requires finite linear combinations as well.

Right. An orthonormal basis of an infinite-dimensional Hilbert space is not a Hamel basis. It's not big enough to be a Hamel basis.

There are infinite-dimensional inner product spaces in which an orthonormal basis is a Hamel basis, but they are not Hilbert spaces because they are not complete. The set of all finite linear combinations of an orthonormal basis is such a space. The set of all trigonometric polynomials with the usual inner product is one concrete example.