There is a well-known criterium that distinguishes Banach spaces into the following two classes: those Banach spaces that can be made into a Hilbert space $(X, \langle .,. \rangle)$ and those that cannot. A norm $\lVert . \rVert$ is induced by an inner product $\langle ., . \rangle$ iff $\lVert \cdot \rVert$ satisfies the parallelogram identity $\lVert x + y \rVert^2 + \lVert x - y \rVert^2 = 2(\lVert x \rVert^2 + \lVert y \rVert^2)$ in which case one can set $\langle x, y \rangle := \frac{1}{4}(\lVert x + y\rVert^2 - \lVert x - y \rVert^2)$.

Is there also a criterium that similarly distinguishes Banach spaces from Banach lattices? Let us split this question into two questions:

  1. Given a Banach space $(X, \lVert . \rVert)$, is there a partial order $\leq$ on $X$ such that $(X, \leq, \lVert . \rVert)$ is a Banach lattice meaning that (i) $(X, \leq)$ is a Riesz space and (ii) $\lVert . \rVert$ is a Riesz norm?

  2. If question 1 has a negative answer, under what additional (possibly sufficient and necessary) conditions on the Banach space $(X, \lVert . \rVert$) does question 1 has a positive answer?

Note that we start here with a Banach space structure and search for a compatible lattice structure. In contrast, if we start with a Riesz space structure $(X, \leq)$, then it is known (Fremlin, Measure Theory, 355Xb) that there is at most one norm $\lVert . \rVert$ (up to equivalence) that turns $(X, \leq, \lVert . \rVert)$ into a Banach lattice.

Added: Maybe this is also helpful to make an abstract Riesz space more concrete: any Riesz space $(X, \leq)$ can be represented as a subspace of a quotient of some $\mathbb{R}^X$ (Fremlin, Measure Theory, 352L). In more detail, there exists a set $Y$, a solid linear subspace $V \subseteq \mathbb{R}^Y$ and an injective Riesz homomorphism $\varphi : X \to \mathbb{R}^Y / V$.

  • 1
    $\begingroup$ One of the necessary conditions is the l.u.st. property $\endgroup$
    – Norbert
    Sep 9, 2016 at 19:45
  • $\begingroup$ @Norbert Thank you. Here is some more information with further references on the local unconditional structure property (l.u.st. property and the weaker GL-l.u.st. property (Gordon-Lewis)): "Handbook of the Geometry of Banach Spaces", pp. 59. Every Banach lattice has l.u.st. Examples for Banach spaces that do not have (GL-)l.u.st. are compact bounded operators on $l^2$, the disc algebra and James' space. $\endgroup$
    – yada
    Sep 10, 2016 at 6:32

1 Answer 1


This is a classical theorem of James.

Let $X$ be a Banach lattice. Then either $X$ is reflexive or it contains an isomorphic copy of $c_0$ or $\ell_1$.

Now, there are non-reflexive spaces that contain neither $c_0$ nor $\ell_1$ (for example, the James space) so they cannot be turned into Banach lattices.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.