A half-line of Schur idempotent norms

arXiv:2611.08419 · math.FA, math.CO, math.OA · Submitted 2026-11-05 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Mathematics. The day in mathematics: new theorems, new proofs, and what they open.

Ada: I'm Ada.

Emmy: And I'm Emmy, here with Kolya.

Ada: We're looking at the paper: "A half-line of Schur idempotent norms".

Emmy: Here's the gist of the paper. A family of bipartite graphs constructed by hanging identical rooted trees at every vertex of a highly symmetric graph shows that every sufficiently large real number is the norm of an idempotent Schur multiplier on the bounded operators on l2 (Page 1).

Ada: The Problem and Main Result The paper addresses whether the set N, which is the set of norms of idempotent Schur multipliers on B(l2), has non-empty interior and whether it contains a half-line

1: .

Emmy: The main result establishes that there is a real number t such that

t, ∞) ⊆ N, implying that N has non-empty interior and is uncountable (Page 2). [Ada: This contrasts with the Toeplitz case where the set of norms of idempotent Toeplitz Schur multipliers forms a countable set (Page 2).

Paper summary: Ada: So this paper "A half-line of Schur idempotent norms" talks about finding a specific type of number related to these Schur multipliers. It claims that there is a real number t such that any number bigger than t is the norm of an idempotent Schur multiplier on bounded operators on l two > <ref:2611.08419#pg1,is the norm of an idempotent Schur multiplier on>

Emmy: That’s the core idea, Ada. They are looking at the set N, which is just all those norms of idempotent Schur multipliers, and they’re showing that this set isn't just scattered around; it has a whole range starting from some point t > two > <ref:2611.08419#pg1>

Kolya: So what makes this different from other cases? Why can't we get this half-line for every kind of matrix or graph structure?

Ada: Well, the paper is focused on using a specific construction involving bipartite graphs built by hanging identical rooted trees at every vertex of a highly symmetric graph to show they can find that starting point t > two > <ref:2611.08419#pg1,by hanging identical rooted trees at every vertex of a highly symmetric>

Emmy: Exactly. They use this family of bipartite graphs called TC(m) and they establish three properties, P1, P2, and P3, which together prove that this construction yields an interval of norms that goes all the way to infinity >

Kolya: I mean what's the structure behind these three properties? What makes them work for these specific graphs TC(m)?

Ada: Property one involves a continuity property for a map called TC(m)• on the set of branching numbers when you use geometric decay of optimal weights as you go deeper into the trees >

Emmy: And property two is that this interval Lambda zero has some positive length, which sets up the next step for showing that things don't just terminate at a single point >

Kolya: So it's about showing that these norms grow in a specific way as you increase the size or depth of these trees, and it doesn't just stop at some finite value >

Ada: Right. And property three is where they link those intervals together, showing that the intervals Lambda p and Lambda p plus one overlap, which leads directly to their main result that there's a half-line

t, infinity) contained in N > [Emmy: So for someone just listening to this show, it means that if you're looking at these operators, you can find an infinite family of them whose norms are arbitrarily large >

Kolya: But what about the caveats? The paper mentions that they use a specific bipartite graph called C = (J3 - I3) tensor (J4 - I4) as their example for a productive core, which is quite concrete, but does that mean this result applies to every possible highly symmetric graph structure >

Ada: Not every structure, no. The paper establishes that for *this* specific bipartite graph C it works and it's called a productive core because hanging sufficiently large trees on it raises the norm strictly above both two and the norm of C itself >

Emmy: That’s interesting because if they can prove it for this specific product structure, maybe there's a way to generalize that idea to other complex graph geometries later on >

Kolya: I'm thinking about what this implies for applications. If we are dealing with systems where these norms matter, does finding a half-line mean that the behavior of those systems isn't bounded by some fixed constant like two >

Ada: That's exactly what it suggests. It means that you can find operators whose norms get bigger and bigger without running into some hard ceiling defined by the structure of the underlying operator space >

Emmy: And this contrasts sharply with the Toeplitz case they mention, where those norms are restricted to a countable set, which is a very different kind of restriction >

Kolya: So we have two regimes here. One where you get this continuous half-line behavior, and another regime, like the Toeplitz one, where it’s just a countable collection of points >

Ada: Right. The main difference is whether you can find an entire interval or just discrete points >

Emmy: So to summarize the main point of "A half-line of Schur idempotent norms," they show that for the specific construction using a productive core C = (J3 - I3) tensor (J4 - I4), there is a real number t such that all numbers greater than t are norms of idempotent Schur multipliers >

Kolya: It's interesting because they mention numerical remarks suggesting this half-line starts at much smaller values of t than perhaps you might expect from the lower bounds they establish in Theorem four point eight(a) for b minus a > <ref:2611.08419#pg2>

Ada: That's a point that warrants more investigation. If t can be smaller, it means that the boundary between the countable set and this continuous half-line behavior is closer to two >

Emmy: And looking at the open questions they list, they are asking if this half-line actually starts at two, or if perhaps it passes between regimes near two >

Kolya: That’s a question about whether the boundary condition is sharp or just an approximation based on the current bounds of their proof structure >

Ada: It suggests that the structure of these norms is much richer than just being bounded by something simple like two, which really gives you room to explore beyond those basic bounds >

Conclusion: Ada: So we’re finishing up "A half-line of Schur idempotent norms" by Beke, Goh, Hatami, Jaffe and Naylor >

Emmy: Yeah, they are showing that for this specific setup with those bipartite graphs and the productive core C equals J3 minus I3 tensor J4 minus I4 >

Ada: Exactly. The authors have proven there is a real number t such that every number larger than t is the norm of an idempotent Schur multiplier on bounded operators on l two >

Emmy: It means that instead of just having a few isolated points for these norms, you get an entire continuous range starting from t > two >

Kolya: So what does this imply about the structure we're seeing in these operator spaces? Does it mean we can find operators with arbitrarily large norms without hitting some hard structural ceiling determined by the setup >

Ada: It suggests that for this particular family of graphs, you can definitely construct multipliers whose norm goes on forever >

Emmy: But they are careful about the context, right? They focus specifically on this product structure as their productive core >

Kolya: I mean, if we look at other graph structures, do we have to stick to this type of construction to find that half-line behavior >

Ada: Not necessarily. The work shows that *this* specific way of hanging trees on a graph creates the necessary structure for that continuous range >

Emmy: It’s about finding the right geometric decay in those optimal weights, which is what property one is all about >

Kolya: So if we want to find this half-line for a different graph, how much harder is it gonna be to prove that same continuity property >

Ada: That’s the next big question. The authors are hinting that it might be achievable with other productive cores, but they’re focusing on this one right now >

Emmy: And they've already given us some numerical hints suggesting the starting number t is actually smaller than you might think based on their initial bounds >

Kolya: That’s interesting because if t is smaller, it means that the boundary between the countable set of norms and this continuous half-line behavior is closer to two >

RUPERT H. LEVENE

math.FA, math.CO, math.OA

Submitted: 2026-11-05

Updated: 2026-11-05

License: http://creativecommons.org/licenses/by-sa/4.0/

Importance score: 83/100

The gist: A family of bipartite graphs constructed by hanging identical rooted trees at every vertex of a highly symmetric graph shows that every sufficiently large real number is the norm of an idempotent…

Key concepts

Idempotent Schur Multiplier Norms (N)
This set N consists of all possible norms resulting from applying idempotent Schur multipliers to bounded operators on l2 space. The paper explores whether this set has non-empty interior or contains an interval, moving beyond the countable sets found in related problems.
Productive Core (C)
A core is a specific bipartite graph structure used in the construction. A core is considered 'productive' if hanging sufficiently large rooted trees onto its vertices causes the resulting norm to strictly exceed both 2 and the product of the core's size and some parameter, ensuring growth.
Geometric Decay of Weights
This property describes how optimal weights in the graph construction decrease as you move deeper into the tree structure. This geometric decay is essential because it guarantees that a specific norm function, ||TC(m)||•, remains continuous over the set of branching numbers.

Terminology

Summary

A family of bipartite graphs constructed by hanging identical rooted trees at every vertex of a highly symmetric graph shows that every sufficiently large real number is the norm of an idempotent Schur multiplier on the bounded operators on l2 (Page 1). p. 0

Where the paper says it

LEVENE Abstract. Using a family of bipartite graphs constructed by hanging identical rooted trees at every vertex of a highly symmetric graph, we show that every sufficiently large real number is the norm of an idempotent Schur multiplier on the bounded operators on ℓ 2 .

The Problem and Main Result

The paper addresses whether the set N, which is the set of norms of idempotent Schur multipliers on B(l2), has non-empty interior and whether it contains a half-line [1]. The main result establishes that there is a real number t such that [t, ∞) ⊆ N, implying that N has non-empty interior and is uncountable (Page 2). p. 1

Where the paper says it

There is a real number t such that [t, ∞) ⊆ N . In particular, the set N has non-empty interior and is uncountable.

This contrasts with the Toeplitz case where the set of norms of idempotent Toeplitz Schur multipliers forms a countable set (Page 2). p. 1
Where the paper says it

Green and Sanders [12] proved a quantitative version of this theorem. The coset ring of Z is countable, so the norms of the idempotent Toeplitz Schur multipliers form a countable set.

Graph Construction and Norm Formula

The authors realize a nontrivial interval as the norms of an explicit family of bipartite graphs, denoted TC(m), built from a core C by hanging rooted trees at every vertex of C (Page 10).

Properties of Productive Cores

The existence of the half-line relies on establishing that a specific bipartite graph construction yields a productive core C. A core is productive if hanging sufficiently large trees in this way raises the norm strictly above both 2 and C• (Page 3). p. 1

Where the paper says it

5 and Figure 1. A core C is productive if hanging sufficiently large trees in this way raises the norm

The proof involves three properties:

  1. (P1) The map m -> TC(m)• is continuous on ΣN R with respect to the product topology (Page 3). p. 2

    Where the paper says it

    For a productive core C and sufficiently large R, the three properties are the following. (P1) The map m 7→ ∥TC(m)∥• is continuous on ΣN R with respect to the product topology.

  2. (P2) The interval Λ∅ has positive length (Page 3). p. 2

    Where the paper says it

    (P1) The map m 7→ ∥TC(m)∥• is continuous on ΣN R with respect to the product topology. (P2) The interval Λ∅ has positive length.

  3. (P3) The intervals Λ(p,d) and Λ(p,d+1) overlap, leading to the conclusion that [t, ∞) ⊆ N by Lemma 5. p. 2

    Where the paper says it

    (P2) The interval Λ∅ has positive length. (P3) The intervals Λ(p,d) and Λ(p,d+1) overlap, so that (1.

    2 (Page 4).

Decay of Weights and Continuity

The continuity property (P1) is proven using geometric decay of optimal weights. Proposition 4. p. 2

Where the paper says it

2) Λp = [ 2R d=R Λ(p,d) . These are established in Proposition 4.

4 shows that if C is productive with parameters R0 and h0, then the optimal weights decay geometrically with depth, specifically xk ≤ r2k0 for 0 ≤ k ≤ h (Page 15). p. 3
Where the paper says it

In Section 3, we apply the dual formula to TC(m) to write the norm as the maximum of an expression involving trace norms of certain symmetric tridiagonal matrices of size at most the length of m, over weights that depend only on depth. In Section 4, we define productive cores and show that optimal weights decay geometrically with depth, which yields the continuity (P1) in Proposition 4.

This decay is crucial for showing that the norm function TC(m)• is continuous on the set of branching numbers (Page 15). p. 9
Where the paper says it

It reduces the norm to a maximisation over h + 1 weights, one for each depth, of an explicit function of the singular values of the core and of the trace norms of chains. The size of these chains is bounded by tree height rather than vertex count, so the formula makes it feasible to estimate the norms of highly branching trees on a core numerically.

Conclusion and Open Questions

The final proof of Theorem 1.1 uses the fact that C = (J3 − I3) ⊗ (J4 − I4) is a productive core, leading to the explicit construction of t (Page 22). p. 3

Where the paper says it

1) is productive. In Section 6, we compute an explicit value of t, which is large.

Numerical remarks suggest that the set N contains a half-line, and open questions concern whether N passes between regimes near 2 (Page 25). p. 3
Where the paper says it

In Section 6, we compute an explicit value of t, which is large. We also compare the constants of the proof with numerical computations and record open questions.

The set N is the union of sets NL where L is the least integer for which NL has non-empty interior, and this least L is bounded by 5 (Page 25). p. 8
Where the paper says it

A core is a finite bipartite graph C with at least one row vertex and at least one column vertex, such that Aut(C) acts transitively on the vertices of C. Since some automorphism of a core maps a row vertex to a column vertex, it exchanges the set of rows with the set of columns, so these sets have the same cardinality, which we write as n.

Numerical Remarks and Open Questions

The paper concludes with numerical experiments suggesting that in fact [t, ∞) ⊆ N for considerably smaller, albeit still large, values of t (Page 23). The size of t is mainly due to the lower bound δR in Theorem 4. p. 7

Where the paper says it

This is at most ∥Q(c)∥1 by the pinching inequality [5, (IV. 52)], which is the lower bound.

8(a) for b - a (Page 23). Open questions include whether a∗ ≤ 2 and whether [2, ∞) ⊆ N (Page 24). p. 1
Where the paper says it

9] whether N has non-empty interior, whether [t,∞) ⊆ N for some t, and whether N is countable. In this paper, we answer these questions by establishing the following result.

The set N is the union of sets NL by [3, Theorem 1. p. 0
Where the paper says it

This answers a question previously raised by the author. Introduction For index sets I and J and a complex matrix B = [bij ]i∈I,j∈J , the Schur norm of B is ∥B∥• = sup{∥B • X∥ : X ∈ B(ℓ 2 J , ℓ2 I ), ∥X∥ ≤ 1}, where ∥ · ∥ is the operator norm, ℓ 2 J is the Hilbert space ℓ 2 of the index set J, the space B(ℓ 2 J , ℓ2 I ) consists of the bounded operators from ℓ 2 J to ℓ 2 I , and B • X = [bijxij ] is the entrywise product of B and X = [xij ].

1], and NKNL ≤ NKL by Lemma 5. p. 2
Where the paper says it

)∥•, ∥TC(p, 2R, 2R, . Since ∥TC(m)∥• is non-decreasing in every mℓ by Lemma 2.

1 (Page 25).

References

[1] W. B. Arveson, Operator algebras and invariant subspaces, Ann. p. 8

Where the paper says it

A HALF-LINE OF SCHUR IDEMPOTENT NORMS 9 Each of these subspaces is invariant for T0 and each ιj , since for 0 ≤ j < h, we have T0fi,ℓ,s = δℓ,0δs,0λifi,0,0 and ιjfi,ℓ,s = δj,sfi,ℓ,s+1. Since K is the orthogonal direct sum of the subspaces Wi,ℓ, each one is reducing for T0 and ιj .

of Math., (2) 100 (1974), 433–532 (Page 1).

[2] I. Balla, L. Hambardzumyan and I. Tomon, Factorization norms and an inverse theorem for MaxCut, Math. p. 0

Where the paper says it

1 arXiv:2611. 08419v1 [math.

Ann., 394 (2026), Paper No. p. 0
Where the paper says it

08419v1 [math. FA] 5 Nov 2026

52 (Page 1).

[3] C. Beke, M. K. Goh, H. Hatami, S. Jaffe and D. p. 1

Where the paper says it

This has led to quantitative structural results for idempotent Schur multipliers, in the spirit of the theorem of Green and Sanders [13, 11, 2]. Recently Beke, Goh, Hatami, Jaffe and Naylor [3] showed that every idempotent Schur multiplier is a finite signed sum of contractive idempotent Schur multipliers, answering a question of Katavolos and Paulsen [17, Section 2].

Naylor, A characterization of idempotent Schur multipliers, preprint, arXiv:2607. p. 1
Where the paper says it

This has led to quantitative structural results for idempotent Schur multipliers, in the spirit of the theorem of Green and Sanders [13, 11, 2]. Recently Beke, Goh, Hatami, Jaffe and Naylor [3] showed that every idempotent Schur multiplier is a finite signed sum of contractive idempotent Schur multipliers, answering a question of Katavolos and Paulsen [17, Section 2].

14316, 2026 (Page 1). p. 0
Where the paper says it

08419v1 [math. FA] 5 Nov 2026

[4] G. Bennett, Schur multipliers, Duke Math. p. 0

Where the paper says it

We identify a 0–1 matrix B indexed by I ×J with the bipartite graph G = (I, J, E) whose edges are the pairs ij with bij = 1, and we write ∥G∥• = ∥B∥•. Let N be the set of norms of the idempotent Schur multipliers on B(ℓ 2 ), that is, N = {∥B∥• : B is a 0–1 Schur multiplier indexed by N × N}.

J., 44 (1977), 603–639 (Page 2).

[5] R. Bhatia, Matrix analysis, Graduate Texts in Mathematics 169, Springer-Verlag, New York, 1997 (Page 2). p. 0

Where the paper says it

231, that theorem states that N ∩ [0, η6] = n 0, 1, q 4 3 , 1 2 (1 + √ 2), 1 15q 169 + 38√ 19, q 3 2 , η6 o . 2020 Mathematics Subject Classification.

[6] R. Bhatia, M.-D. Choi and C. Davis, Comparing a matrix to its off-diagonal part, in: The Gohberg anniversary collection, Vol. p. 3

Where the paper says it

The following notation is used throughout the paper. For a vector v, we write Dv for the diagonal matrix with diagonal v, and √ v is taken entrywise.

I, Oper. Theory Adv. p. 1
Where the paper says it

1 shows that, unlike in the Toeplitz case, this structure theorem leaves room for a half-line of norms. Idempotent Schur multipliers are also the idempotent bimodule maps over the diagonal masa of B(ℓ 2 ), the maximal abelian self-adjoint subalgebra of diagonal operators [17, Section 1], and ranges of such bimodule projections are studied in the theory of reflexivity that goes back to Arveson [1, 17, 10, 9].

Appl., 40, Birkh¨auser, Basel (Page 2).

[7] P. J. Cohen, On a conjecture of Littlewood and idempotent measures, Amer. p. 1

Where the paper says it

Then B is a Schur multiplier if and only if there is a measure µ on the circle with cn = ˆµ(n) for all n ∈ Z, and then ∥B∥• = ∥µ∥, the total variation norm of µ [4, Theorem 8. If B is also a 0–1 matrix, then µ is idempotent under convolution, and the idempotent theorem of Helson and Cohen [15, 7] shows that {n ∈ Z : cn = 1} lies in the coset ring of Z, the Boolean algebra of subsets of Z generated by the cosets of subgroups.

J. Math., 82 (1960), 191–212 (Page 3).

[8] C. C. Cowen, P. A. Ferguson, D. K. Jackman, E. A. Sexauer, C Vogt and H J Woolf, Finding norms of Hadamard multipliers, Linear Algebra Appl. p. 0

Where the paper says it

We identify a 0–1 matrix B indexed by I ×J with the bipartite graph G = (I, J, E) whose edges are the pairs ij with bij = 1, and we write ∥G∥• = ∥B∥•. Let N be the set of norms of the idempotent Schur multipliers on B(ℓ 2 ), that is, N = {∥B∥• : B is a 0–1 Schur multiplier indexed by N × N}.

, 247 (1996), 217–235 (Page 5).

[9] G. K. Eleftherakis, R. H. Levene and I. p. 0

Where the paper says it

A HALF-LINE OF SCHUR IDEMPOTENT NORMS RUPERT H. LEVENE Abstract.

G. Todorov, Schur idempotents and hyperreflexivity, Israel J. p. 0
Where the paper says it

A HALF-LINE OF SCHUR IDEMPOTENT NORMS RUPERT H. LEVENE Abstract.

Math., 215 (2016), 317–337 (Page 5).

[10] G. K. Eleftherakis and I. G. Todorov, Ranges of bimodule projections and reflexivity, J. p. 1

Where the paper says it

1 shows that, unlike in the Toeplitz case, this structure theorem leaves room for a half-line of norms. Idempotent Schur multipliers are also the idempotent bimodule maps over the diagonal masa of B(ℓ 2 ), the maximal abelian self-adjoint subalgebra of diagonal operators [17, Section 1], and ranges of such bimodule projections are studied in the theory of reflexivity that goes back to Arveson [1, 17, 10, 9].

Funct. Anal., 262 (2012), 4891–4915 (Page 5).

[11] M. K. Goh and H Hatami, Block complexity and idempotent Schur multipliers, Int. p. 1

Where the paper says it

This has led to quantitative structural results for idempotent Schur multipliers, in the spirit of the theorem of Green and Sanders [13, 11, 2]. Recently Beke, Goh, Hatami, Jaffe and Naylor [3] showed that every idempotent Schur multiplier is a finite signed sum of contractive idempotent Schur multipliers, answering a question of Katavolos and Paulsen [17, Section 2].

Math Res. Not. IMRN, 2025 (2025), no. 24, Paper No. p. 1
Where the paper says it

9] whether N has non-empty interior, whether [t,∞) ⊆ N for some t, and whether N is countable. In this paper, we answer these questions by establishing the following result.

rnaf356 (Page 7).

[12] B Green and T Sanders, A quantitative version of the idempotent theorem in harmonic analysis, Ann. p. 1

Where the paper says it

If B is also a 0–1 matrix, then µ is idempotent under convolution, and the idempotent theorem of Helson and Cohen [15, 7] shows that {n ∈ Z : cn = 1} lies in the coset ring of Z, the Boolean algebra of subsets of Z generated by the cosets of subgroups. Green and Sanders [12] proved a quantitative version of this theorem.

of Math., (2) 168 (2008), 1025–1054 (Page 3).

[13] L. Hambardzumyan, H. Hatami and P. p. 1

Where the paper says it

This has led to quantitative structural results for idempotent Schur multipliers, in the spirit of the theorem of Green and Sanders [13, 11, 2]. Recently Beke, Goh, Hatami, Jaffe and Naylor [3] showed that every idempotent Schur multiplier is a finite signed sum of contractive idempotent Schur multipliers, answering a question of Katavolos and Paulsen [17, Section 2].

Hatami, Dimension-free bounds and structural results in communication complexity, Israel J. p. 1
Where the paper says it

1], and this norm is central in communication complexity. This has led to quantitative structural results for idempotent Schur multipliers, in the spirit of the theorem of Green and Sanders [13, 11, 2].

Math., 253 (2023), 555–616 (Page 7).

[14] P de la Harpe, Spectral multiplicity functions of adjacency operators of graphs and cospectral infinite graphs, Algebra Discrete Math., 37 (2024), 44–84 (Page 8).

[15] H. Helson, Note on harmonic functions, Proc. p. 1

Where the paper says it

Then B is a Schur multiplier if and only if there is a measure µ on the circle with cn = ˆµ(n) for all n ∈ Z, and then ∥B∥• = ∥µ∥, the total variation norm of µ [4, Theorem 8. If B is also a 0–1 matrix, then µ is idempotent under convolution, and the idempotent theorem of Helson and Cohen [15, 7] shows that {n ∈ Z : cn = 1} lies in the coset ring of Z, the Boolean algebra of subsets of Z generated by the cosets of subgroups.

Amer. Math., 4 (1953), 686–691 (Page 2).

[16] A J Hoffman, On limit points of spectral radii of non-negative symmetric integral matrices, in: Graph theory and applications, Lecture Notes in Math. p. 0

Where the paper says it

1 arXiv:2611. 08419v1 [math.

p. 3
Where the paper says it

1 turns it into [t,∞) for some t by Lemma 5. We remark that this digit-by-digit choice of m resembles Shearer’s solution [28] of Hoffman’s problem [16, Sections 1 and 6] on limit points of spectral radii.

303, Springer, Berlin, 1972, 165–172 (Page 2).

[17] A Katavolos and V. p. 0

Where the paper says it

Livshits [20] showed that N ∩ (1, √ 2 3 ) = ∅. Katavolos and Paulsen [17, Remark 5] recorded an observation of Davidson that N is closed under multiplication and under suprema of bounded sets, and noted that 2 is a limit point of N by [6].

I. Paulsen, On the ranges of bimodule projections, Canad. p. 1
Where the paper says it

1 shows that, unlike in the Toeplitz case, this structure theorem leaves room for a half-line of norms. Idempotent Schur multipliers are also the idempotent bimodule maps over the diagonal masa of B(ℓ 2 ), the maximal abelian self-adjoint subalgebra of diagonal operators [17, Section 1], and ranges of such bimodule projections are studied in the theory of reflexivity that goes back to Arveson [1, 17, 10, 9].

Math. Bull., 48 (2005), 97–111 (Page 5).

[18] S Kwapie´n and A Pelczy´nski, The main triangle projection in matrix spaces and its applications, Studia Math. p. 3

Where the paper says it

The trace norm of a matrix is ∥ · ∥1, and for a finite vector v we write ∥v∥1 = P i |vi | for its ℓ 1 norm, which is the trace norm of Dv. The Hilbert–Schmidt norm is ∥ · ∥2, the matrix unit Eij has entry 1 at position (i, j) and 0 elsewhere, with its size clear from the context, and a matrix with countable index sets is identified with an operator between the corresponding ℓ 2 spaces whenever it defines a bounded operator.

, 34 (1970), 43–68 (Page 5).

[19] R. H. Levene, Norms of idempotent Schur multipliers, New York J. p. 0

Where the paper says it

A HALF-LINE OF SCHUR IDEMPOTENT NORMS RUPERT H. LEVENE Abstract.

Math., 20 (2014), 325–352 (Page 7).

[20] L Livshits, A note on 0–1 Schur multipliers, Linear Algebra Appl. p. 0

Where the paper says it

Let N be the set of norms of the idempotent Schur multipliers on B(ℓ 2 ), that is, N = {∥B∥• : B is a 0–1 Schur multiplier indexed by N × N}. Livshits [20] showed that N ∩ (1, √ 2 3 ) = ∅.

, 222 (1995), 15–22 (Page 4).

[21] R. Mathias, The Hadamard operator norm of a circulant and applications, SIAM J. p. 0

Where the paper says it

This answers a question previously raised by the author. Introduction For index sets I and J and a complex matrix B = [bij ]i∈I,j∈J , the Schur norm of B is ∥B∥• = sup{∥B • X∥ : X ∈ B(ℓ 2 J , ℓ2 I ), ∥X∥ ≤ 1}, where ∥ · ∥ is the operator norm, ℓ 2 J is the Hilbert space ℓ 2 of the index set J, the space B(ℓ 2 J , ℓ2 I ) consists of the bounded operators from ℓ 2 J to ℓ 2 I , and B • X = [bijxij ] is the entrywise product of B and X = [xij ].

Matrix Anal. p. 0
Where the paper says it

This answers a question previously raised by the author. Introduction For index sets I and J and a complex matrix B = [bij ]i∈I,j∈J , the Schur norm of B is ∥B∥• = sup{∥B • X∥ : X ∈ B(ℓ 2 J , ℓ2 I ), ∥X∥ ≤ 1}, where ∥ · ∥ is the operator norm, ℓ 2 J is the Hilbert space ℓ 2 of the index set J, the space B(ℓ 2 J , ℓ2 I ) consists of the bounded operators from ℓ 2 J to ℓ 2 I , and B • X = [bijxij ] is the entrywise product of B and X = [xij ].

Appl., 14 (1993), 1152–1167 (Page 5).

[22] J Mudge and H L Pham, Idempotents with small norms, J. p. 1

Where the paper says it

The coset ring of Z is countable, so the norms of the idempotent Toeplitz Schur multipliers form a countable set. Saeki [26, 27] and Mudge and Pham [22] determined the smallest elements of this set by classifying the idempotent measures of small norm.

Funct. Anal., 270 (2016), 4597–4603 (Page 8).

[23] J Parcet, M de la Salle and E Tablate, The local geometry of idempotent Schur multipliers, Forum Math. p. 1

Where the paper says it

Idempotent Schur multipliers are also the idempotent bimodule maps over the diagonal masa of B(ℓ 2 ), the maximal abelian self-adjoint subalgebra of diagonal operators [17, Section 1], and ranges of such bimodule projections are studied in the theory of reflexivity that goes back to Arveson [1, 17, 10, 9]. Parcet, de la Salle and Tablate [23] characterise, locally and under a transversality condition, the idempotent Schur multipliers whose symbols are indicator functions of smooth domains and that are bounded on the Schatten p-classes for 1 < p < ∞ with p ̸= 2.

Pi 13 (2025), e14 (Page 7).

[24] G Pisier, Similarity problems and completely bounded maps, 2nd, expanded ed. p. 1

Where the paper says it

Idempotent Schur multipliers are also the idempotent bimodule maps over the diagonal masa of B(ℓ 2 ), the maximal abelian self-adjoint subalgebra of diagonal operators [17, Section 1], and ranges of such bimodule projections are studied in the theory of reflexivity that goes back to Arveson [1, 17, 10, 9]. Parcet, de la Salle and Tablate [23] characterise, locally and under a transversality condition, the idempotent Schur multipliers whose symbols are indicator functions of smooth domains and that are bounded on the Schatten p-classes for 1 < p < ∞ with p ̸= 2.

, Lecture Notes in Math. 1618, Springer, Berlin, 2001 (Page 5).

[25] O Rojo and L Medina, Spectra of weighted compound graphs of generalized Bethe trees, Electron. p. 7

Where the paper says it

The decomposition that underlies this formula is standard, see Rojo and Medina [25] for weighted graphs with a tree attached at every vertex and de la Harpe [14, Proposition 4. 6] for spherically symmetric trees.

J. Linear Algebra 18 (2009), 30–57 (Page 8). p. 1
Where the paper says it

Then B is a Schur multiplier if and only if there is a measure µ on the circle with cn = ˆµ(n) for all n ∈ Z, and then ∥B∥• = ∥µ∥, the total variation norm of µ [4, Theorem 8. If B is also a 0–1 matrix, then µ is idempotent under convolution, and the idempotent theorem of Helson and Cohen [15, 7] shows that {n ∈ Z : cn = 1} lies in the coset ring of Z, the Boolean algebra of subsets of Z generated by the cosets of subgroups.

[26] S Saeki, On norms of idempotent measures, Proc. p. 1

Where the paper says it

The coset ring of Z is countable, so the norms of the idempotent Toeplitz Schur multipliers form a countable set. Saeki [26, 27] and Mudge and Pham [22] determined the smallest elements of this set by classifying the idempotent measures of small norm.

Amer. Math., 19 (1968), 600–602 (Page 4).

[27] S Saeki, On norms of idempotent measures. p. 1

Where the paper says it

The coset ring of Z is countable, so the norms of the idempotent Toeplitz Schur multipliers form a countable set. Saeki [26, 27] and Mudge and Pham [22] determined the smallest elements of this set by classifying the idempotent measures of small norm.

II, Proc. Amer. Math., 19 (1968), 367–371 (Page 4).

[28] J B Shearer, On the distribution of the maximum eigenvalue of graphs, Linear Algebra Appl., 114/115 (1989), 17–20 (Page 5).

B A HALF-LINE OF SCHUR IDEMPOTENT NORMS. p. 0

Where the paper says it

A HALF-LINE OF SCHUR IDEMPOTENT NORMS RUPERT H. LEVENE Abstract.

N is the page number counting from 1 — the same N as that page's --- Page N --- line. The sentence's own full stop stays AFTER the closing >. A token with nothing after the comma is wrong. (Page 2).

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved system can prove that every sufficiently large real number is a norm of an idempotent Schur multiplier on bounded operators on l2 by constructing specific bipartite graphs from a highly symmetric graph, as shown in Theorem 1.1.

  2. The system can determine whether the set N of norms of idempotent Schur multipliers has non-empty interior and is uncountable by establishing that there is a real number t such that [t, infinity) is contained in N.

  3. The improved system can contrast the structure of idempotent Schur multipliers with the Toeplitz case, understanding that for Toeplitz matrices, the norms are related to measures on the circle, whereas for general bipartite graphs, a half-line of norms exists.

  4. The system can apply structural results from operator algebras and functional analysis to characterize idempotent Schur multipliers as idempotent bimodule maps over the diagonal masa of B(l2).

  5. The system can compute an explicit value for the constant t in Theorem 1.1, using numerical maximizers and bounds derived from productive cores, specifically finding t = (∥C∥• + 2)K where K is determined by log calculations involving constants R1 and deltaR.

  6. The system can perform numerical experiments to estimate the exact values of norms for specific graphs, such as showing that for m = (40, 40, 40, 40, 40), the norm of TC(m) is approximately 2.08 using a quasi-Newton method.

  7. The system can investigate open questions regarding the structure of N by comparing the infimum of limit points a star to the supremum t star and analyzing whether [2, infinity) is contained in N or if [2, 2 + epsilon] is contained in N for some epsilon > 0.

Abstract

Using a family of bipartite graphs constructed by hanging identical rooted trees at every vertex of a highly symmetric graph, we show that every sufficiently large real number is the norm of an idempotent Schur multiplier on the bounded operators on squared. This answers a question previously raised by the author.

Sources

Related papers