Norm averaging in a free group algebra
Written by GPT-6.1 Sol (OpenAI), Ultra, September 2026. Self-checked by the writing AI. New original text: public domain (CC0).
The free group factor gap controls distance from the scalars in the trace -norm. Here we prove a different statement in operator norm: finite averages of group conjugates can bring any element of the reduced group -algebra arbitrarily close to its scalar trace. This proves simplicity and uniqueness of the tracial state.
The prerequisites are Hilbert-space operator norms, orthogonal projections, reduced words and the canonical faithful group trace constructed in the preceding lesson. The proof uses finite words and finite operator averages. It works for a free group on any set of at least two generators; separability is unnecessary.
Fix distinct generators , put , and let
The normalized faithful trace is . The same proof of the trace using the commuting right regular representation works for an arbitrary indexing set ; finite linear combinations of basis vectors remain dense.
1. Orthogonal ranges give a norm estimate
Lemma 1.1. Let be an orthogonal projection on a Hilbert space, and unitaries such that
If , then
Proof. If , the operator has range in . These subspaces are mutually orthogonal by (2). Writing , we therefore have for , and
Now decompose
Both have range in , and both norms are at most . Apply (4) to each and use the triangle inequality, obtaining a bound before division by .
The equivalent orientation uses , with . Replacing each by its adjoint in (2)–(3) gives exactly that version.
More generally, if and , then and
Equality in the last inequality need not hold. For example, on , let , , be the coordinate flip and . Condition (2) holds, but
Thus an equality cannot be substituted for that intermediate norm inequality. The claimed estimate (3) remains valid.
2. A word partition with separated translates
Let be the nonidentity reduced words whose initial block is a nonzero power of , or whose initial -block is exactly . A word starting with a different generator is outside . Let , including the identity, and set
Lemma 2.1. For every nonzero integer ,
In particular the subsets , , are pairwise disjoint.
Proof. A word in starts with no -letter, and its initial -block, if present, has exponent different from . If , ends in , so concatenation with that word has no cancellation at the join. It starts with the -block , hence lies in .
If , . Its final cannot disappear completely against an initial -block of exponent different from : if that exponent is positive, the remaining block has exponent one less; if it is negative, the blocks join. A word starting with another generator causes no cancellation. The initial therefore survives, so the result lies in . This includes the empty word. If two translates intersected, multiplying on the left by the inverse of one power would contradict .
Lemma 2.2. If and , where is reduced word length, then
is either a nonzero power of , or starts with and ends with .
Proof. Strip all possible initial blocks and terminal blocks from the reduced word. Thus
where the remaining word is empty, or has neither the indicated initial nor terminal block. If is empty, , with nonzero exponent, and conjugation leaves it unchanged.
Otherwise (10) is
Both powers have positive length. At the left join, at most the terminal can cancel: cancellation of the following would require to start with , which has been excluded. At the right join, at most the initial can cancel, because a further cancellation would require to end with . If has length one and is or , one join removes it; the two surviving adjacent letters at the new join are or , so they do not cancel. If both joins cancel for a longer word, they cannot remove all of , since a reduced word cannot consist of . The first and final in (12) survive.
Put
Lemma 2.3. For every nonidentity word and every ,
Proof. Write . If starts with and ends with , neither join with or cancels. Thus starts with and ends with . If , , its last cancels one letter of the adjacent , leaving a ; if , its first cancels one letter of the adjacent , leaving a . There are no further cancellations across that join. In both cases the outer initial and terminal survive.
Multiplying this reduced word by a word in either causes no cancellation at its terminal , or cancels exactly one , since the input's initial -block is . The remaining terminal prevents any deeper cancellation. The product still starts with , which puts it in .
A more general choice is insufficient: it allows any in place of in . In rank at least three choose , , , and an input word . The product
still starts with the block , so belongs to , contradicting the proposed inclusion. The fixed choice (13) works in every rank at least two.
3. Simultaneous norm averaging
Theorem 3.1. Let have zero identity coefficient. If exceeds the length of every word in its support, then, for every ,
Proof. Let be the projection onto , so projects onto . Lemma 2.1 gives
Conjugate by . Every supporting word then maps into by Lemma 2.3, so the whole conjugated operator satisfies
Apply Lemma 1.1 to and . Its norm is , and this gives (16).
The estimate is for the whole polynomial, rather than a separate estimate summed over its coefficients. One value of works for its entire finite support.
Corollary 3.2. For every , is in the operator-norm closure of the convex hull of .
Proof. Given , choose a polynomial with , and put . Then
The average from Theorem 3.1 is a unital contraction and fixes scalars. Consequently
First choose small, then large. Each is an actual finite convex combination of the required conjugates.
4. Simplicity and the only trace
For the arbitrary generator set in this lesson, here is the full trace check on . On group polynomials , ,
Both sums are finite. Norm approximation by polynomials and continuity of the vector state extend this identity to every . For faithfulness, the right translations commute with all of . If and , then . Commutation gives for every . These vectors include every basis vector and have dense linear span in , so and . The argument requires no countability of the basis.
Theorem 4.1. The reduced -algebra of a free group with at least two generators is simple, and its canonical trace is its unique tracial state.
Proof. Every tracial state is invariant under conjugation by the group unitaries. It has the same value on and every convex average in Corollary 3.2, so continuity gives
Let be a nonzero closed two-sided ideal. Choose ; for example take for nonzero . Faithfulness gives . After rescaling, assume . Corollary 3.2 provides an element with , because every conjugate of remains in . The Neumann series makes invertible in , hence . Therefore .
The distinction between the two norms matters. A trace -norm approximation to need not produce an invertible element; the argument above uses operator norm.
5. Exercises with complete solutions
Exercise 1. Give a norm-equality counterexample with orthogonal ranges satisfying (2).
Solution. Take the two projections onto the coordinate axes of . Their ranges are orthogonal; their sum is the identity. Its squared norm is , while the sum of their squared norms is . They are the two conjugates of under and the coordinate flip, and . This verifies (7) under the full hypotheses.
Exercise 2. Prove the coefficient in the estimate for cannot be replaced by .
Solution. On a Hilbert space with orthonormal , let project onto and . Choose fixing and carrying to . Their conjugates of send to the orthogonal unit vectors . Thus the sum has norm , and its average has norm . The projections are orthogonal, so (2) holds.
Exercise 3. Which part of the proof of (9) requires exclusion of an initial -block of exponent ?
Solution. For negative , the terminal of can cancel against an initial positive -block. Exponent would remove that block completely and could expose an -letter which cancels further into the prefix. Excluding exponent leaves a nonzero -block, preserving the initial . For example ; the input is in , not in .
Exercise 4. Apply the stripping construction (11) to , where is a third generator.
Solution. Here , and . For , the conjugate is . No join cancels, so it starts with and ends with , as required.
Exercise 5. Explain why contains rather than , and test the weaker printed condition with .
Solution. The initial puts the result outside ; the final leaves one after multiplying by an input whose initial block is . Replacing by satisfies but begins with exponent of . With , the conjugated word and its product with retain that initial block, giving (15) in . Thus the weaker condition fails in rank three.
Exercise 6. For , explain how a single averaging family controls the whole polynomial.
Solution. The identity coefficient is zero, and the largest support length is . Choose , , and the conjugators . Every transformed supporting word sends into , so the compression of the whole transformed polynomial to is zero. Formula (16) gives , without replacing by a sum of coefficient magnitudes.
Exercise 7. Does one averaging family work for several polynomials at once?
Solution. Yes. Choose exceeding the lengths in the union of their finite supports after removing identity coefficients. The sets , the word and the powers then work for every polynomial. The bound for the -th polynomial is . Approximating a finite set of arbitrary elements by polynomials and applying (17) gives simultaneous norm approximation as well.
Exercise 8. Why does the simplicity proof need a positive element and a faithful trace?
Solution. A nonzero element can have trace zero, so rescaling it to trace may be impossible. If lies in the ideal, is a nonzero positive element in it. Faithfulness then ensures . Rescaling and norm averaging produce an invertible ideal element and force the ideal to contain .
Exercise 9. Give a finite-dimensional example showing that closeness to in normalized trace -norm does not imply invertibility.
Solution. In , let project onto the first coordinate and set . It is singular, but , tending to zero as increases. Its operator-norm distance from is , so it never satisfies the Neumann criterion.
Exercise 10. For the free group on one generator, show that the conclusion of Theorem 4.1 fails.
Solution. Let on . The algebra generated by is commutative. For , the unit vectors satisfy . Their vector states on Laurent polynomials tend to , giving bounded characters of the norm closure. The character at has a proper kernel containing the nonzero element ; the two characters are distinct tracial states. Commutativity also makes every conjugate of equal to , preventing scalar averaging for a nonscalar element.
Reading and attribution
Pierre de la Harpe, On simplicity of reduced C*-algebras of groups, arXiv:math/0509450v1, Section 3, Definition 9 and full Theorem 14 proof, pp.7 and 10–11, supplies the Powers partition and orthogonal-range averaging method. The exact estimate and invertible-ideal step were actually read. Appendix IX, p.19, discusses scalar convex averaging by reference; the complete scalar averaging and unique-trace arguments are proved here.
Sections 1–4 give the complete finite-word construction, correct range inequality, fixed , simultaneous finite-support estimate and ideal proof. The local two-range decomposition works for non-self-adjoint inputs, and invertibility uses the Neumann series in . The argument applies to every free generator set of size at least two, with no countability hypothesis or reliance on a boundary-action theorem. Exact transitive Hilbert-space, group trace and C*-algebra foundations remain pending; no source expression was imported.