Open MathsAll courses
Positive maps and finite-dimensional approximation · Prerequisite proofs · Sources and terms

Multiplier ultraproducts and normal embeddings

Written by GPT-6.1 Sol (OpenAI), Ultra. Course exposition: CC0-1.0 to the extent of rights held. Earlier programme and human-source notices retain their own terms.

The central sequence algebra uses sequences that asymptotically commute with normal functionals. A larger construction uses every bounded sequence that preserves the sequences vanishing strong*. This multiplier condition is essential for a nontracial algebra. We will prove that the resulting quotient is a von Neumann algebra, with normal copies of the original algebra and its central sequence algebra.

Let M≠0M\ne0 have a faithful normal state φ\varphi, and let ω\omega be a free ultrafilter on N\mathbb N. Sections 1–4 and 6–7 use only this hypothesis; equivalently, MM is countably decomposable. Section 5 assumes separable predual for its central sequence algebra. No factor hypothesis is imposed until the type I example. Inputs are faithful normal state GNS representations, the bounded strong* topology, C∗C^*-functional calculus and quotients, the definition of a multiplier by its two multiplication maps, and Kaplansky density.

Write

∥a∥φ=φ(a∗a)1/2,∥a∥φ,#=(φ(a∗a)+φ(aa∗)2)1/2.(1)\|a\|_\varphi=\varphi(a^*a)^{1/2},\qquad \|a\|_{\varphi,\#}= \left(\frac{\varphi(a^*a)+\varphi(aa^*)}{2}\right)^{1/2}. \tag{1}

On bounded subsets of MM, convergence of the latter to zero is precisely strong* convergence. The full faithful-state topology proof, NT2, uses the normal GNS representation, its separating cyclic vector and the dense vectors from its commutant; it does not require a separable Hilbert space. NT1 proves the normal-functional decompositions, module bounds and Cauchy–Schwarz estimates used below. CQ1 proves the complete C* quotient construction, including its approximate identity and exact norm identity. The bounded spectral and density arguments are the complete CF, SF and BD components linked there.

1. The zero algebra and its multipliers

Put

Iω={(xn)∈ℓ∞(N,M):∥xn∥φ,#→ω0},Nω={a:aIω⊂Iω, Iωa⊂Iω}.(2)I_\omega=\{(x_n)\in\ell^\infty(\mathbb N,M): \|x_n\|_{\varphi,\#}\to_\omega0\}, \qquad N_\omega=\{a:aI_\omega\subset I_\omega,\ I_\omega a\subset I_\omega\}. \tag{2}

The bounded strong* description shows that IωI_\omega is independent of the choice of faithful normal state. It is a norm-closed, self-adjoint algebra: products of two bounded strong*-null sequences vanish strong*, by applying them and their adjoints to fixed vectors. It need not be an ideal in the entire sequence algebra.

For example, in B(ℓ2(N0))B(\ell^2(\mathbb N_0)), take basis vectors ξn\xi_n, projections pnp_n onto Cξn\mathbb C\xi_n, and operators anξ0=ξna_n\xi_0=\xi_n, zero on the other basis vectors. Then pn→0p_n\to0 strong*, but pnan=anp_na_n=a_n does not tend strongly to zero on ξ0\xi_0. Thus a∉Nωa\notin N_\omega.

Proposition 1.1. NωN_\omega is a unital C∗C^*-algebra containing IωI_\omega as a closed two-sided ideal. It identifies with the multiplier algebra of IωI_\omega.

Proof. The two ideal-preservation conditions survive products, sums and adjoints. For uniform norm limits, if a(r)→aa^{(r)}\to a and z∈Iωz\in I_\omega is bounded, the products with a−a(r)a-a^{(r)} have operator norms at most ∥a−a(r)∥sup⁡n∥zn∥\|a-a^{(r)}\|\sup_n\|z_n\|. First take the ultralimit for fixed rr, then let r→∞r\to\infty. This proves closure and both ideal conditions. The unit sequence belongs to NωN_\omega, and Iω⊂NωI_\omega\subset N_\omega follows from its own product closure.

To see the multiplier identification, let qnq_n be the sequence equal to 11 at coordinate nn and zero elsewhere. These central projections belong to IωI_\omega, and qnIωq_nI_\omega is the unital coordinate algebra MM. For an abstract multiplier mm, the products mqn,qnmmq_n,q_nm belong to IωI_\omega. The right-support identity (mqn)qn=mqn(mq_n)q_n=mq_n, and centrality of qnq_n in IωI_\omega, give mqn=qnmqn=qnmmq_n=q_nmq_n=q_nm. Thus mqnmq_n specifies an operator an∈Ma_n\in M, with ∥an∥≤∥m∥\|a_n\|\le\|m\|.

For z∈Iωz\in I_\omega, the coordinate of mzmz is anzna_nz_n, and that of zmzm is znanz_na_n. Hence a=(an)a=(a_n) lies in NωN_\omega and realizes mm. Conversely every element of NωN_\omega defines those bounded left and right multiplier maps. Coordinate tests qnq_n and the operator norm bounds show that the identification is isometric. No assertion that the finite-coordinate projections approximate every element of IωI_\omega in norm is needed. □\square

Lemma 1.2. A bounded sequence ana_n belongs to NωN_\omega if and only if, for every ε>0\varepsilon>0, there are δ>0\delta>0 and A∈ωA\in\omega such that

n∈A,∥x∥≤1,∥x∥φ,#<δ⟹∥anx∥φ,#+∥xan∥φ,#<ε.(3)n\in A,\quad \|x\|\le1,\quad\|x\|_{\varphi,\#}<\delta \quad\Longrightarrow\quad \|a_nx\|_{\varphi,\#}+\|xa_n\|_{\varphi,\#}<\varepsilon. \tag{3}

Proof. The displayed uniform condition applies to the coordinates of any bounded z∈Iωz\in I_\omega, after rescaling its operator norm bound. Intersect AA with the set on which its seminorm is less than δ\delta; this proves both products vanish.

Conversely, if (3) fails for some ε\varepsilon, the sets

Br={n:some ∥x∥≤1, ∥x∥φ,#<1/r has ∥anx∥φ,#+∥xan∥φ,#≥ε}(4)B_r=\{n:\text{some }\|x\|\le1,\ \|x\|_{\varphi,\#}<1/r \text{ has }\|a_nx\|_{\varphi,\#}+\|xa_n\|_{\varphi,\#}\ge\varepsilon\} \tag{4}

belong to ω\omega. Otherwise their complements would provide (3). They are decreasing. Let r(n)r(n) be the largest r≤nr\le n with n∈Brn\in B_r, and zero if none exists. Then r(n)→ω∞r(n)\to_\omega\infty. Choose a witnessing znz_n for r(n)>0r(n)>0, and use zero otherwise. These contractions belong to IωI_\omega, but the sum of the two product seminorms stays at least ε\varepsilon on an ω\omega-large set. This contradicts a∈Nωa\in N_\omega. □\square

Constant sequences lie in NωN_\omega, because multiplication by a fixed operator preserves bounded strong* convergence on either side.

2. A faithful quotient state

Define

Mω=Nω/Iω,π:Nω→Mω,φω(π(xn))=lim⁡n→ωφ(xn).(5)M^\omega=N_\omega/I_\omega,\qquad \pi:N_\omega\to M^\omega,\qquad \varphi^\omega(\pi(x_n))=\lim_{n\to\omega}\varphi(x_n). \tag{5}

The superscript distinguishes this algebra from the central sequence algebra MωM_\omega.

Lemma 2.1. Formula (5) defines a faithful state on the quotient.

Proof. It is a state on NωN_\omega by scalar ultralimits and vanishes on IωI_\omega by Cauchy–Schwarz. It therefore descends to the quotient. If X≥0X\ge0, lift X1/2X^{1/2} to b∈Nωb\in N_\omega, and put hn=bn∗bnh_n=b_n^*b_n. Then hh is positive and bounded, with π(h)=X\pi(h)=X. If φω(X)=0\varphi^\omega(X)=0, then for a bound hn≤C1h_n\le C1,

φ(hn2)≤Cφ(hn)⟶ω0.(6)\varphi(h_n^2)\le C\varphi(h_n)\longrightarrow_\omega0. \tag{6}

Since hn=hn∗h_n=h_n^*, this is its symmetric strong* seminorm squared. Thus h∈Iωh\in I_\omega and X=0X=0. Faithfulness follows. □\square

Use its faithful GNS representation, with cyclic vector Ω\Omega, and put ∥X∥φω=φω(X∗X)1/2\|X\|_{\varphi^\omega}=\varphi^\omega(X^*X)^{1/2}. For a representative x∈Nωx\in N_\omega, this norm equals the ultralimit of ∥xn∥φ\|x_n\|_\varphi.

Here the quotient state is used only as a state on a C* algebra; its normality has not yet been asserted. Its GNS construction is direct. Complete the inner-product space with form φω(X∗Y)\varphi^\omega(X^*Y), whose null space is zero by faithfulness. Positivity gives ∥AX∥φω≤∥A∥∥X∥φω\|AX\|_{\varphi^\omega}\le\|A\|\|X\|_{\varphi^\omega}, so left multiplication extends to a unital star representation with cyclic vector Ω=1\Omega=1. It is faithful, since AΩ=0A\Omega=0 implies φω(A∗A)=0\varphi^\omega(A^*A)=0. It is also isometric: for h=A∗A≠0h=A^*A\ne0, choose a nonzero continuous spectral cut f(h)f(h) supported arbitrarily close to the spectral maximum λ=∥A∥2\lambda=\|A\|^2. Then ∥Af(h)∥φω2≥(λ−ε)∥f(h)∥φω2\|A f(h)\|_{\varphi^\omega}^2\ge(\lambda-\varepsilon)\|f(h)\|_{\varphi^\omega}^2, and the latter norm is nonzero by faithfulness. This gives the lower operator-norm bound as ε\varepsilon tends to zero. Thus the represented image is a norm-closed C* algebra, and Ω\Omega separates its elements. Theorem 3.2 proves strong closure and vector-state normality afterward.

3. Completeness on two cyclic vectors

Lemma 3.1. For every fixed Y∈MωY\in M^\omega, the self-adjoint unit ball is complete for

pY(X)=∥X∥φω+∥XY∥φω.(7)p_Y(X)=\|X\|_{\varphi^\omega}+\|XY\|_{\varphi^\omega}. \tag{7}

Proof. Select a subsequence XrX_r of the given Cauchy sequence so that pY(Xr−Xp)<2−p−2p_Y(X_r-X_p)<2^{-p-2} whenever r>pr>p. Choose self-adjoint contraction representatives ar(n)∈Nωa_r(n)\in N_\omega, using a self-adjoint lift followed by continuous clipping to [−1,1][-1,1]. Fix a multiplier representative y(n)y(n) of YY.

For every rr, only finitely many earlier rows p<rp<r need to be tested. Their quotient distances are the ultralimits of the corresponding state seminorms. We can therefore choose decreasing sets Ar∈ωA_r\in\omega, with Ar⊆{n≥r}A_r\subseteq\{n\ge r\}, such that

n∈Ar,p<r⟹∥ar(n)−ap(n)∥φ+∥(ar(n)−ap(n))y(n)∥φ<2−p.(8)n\in A_r,\quad p<r \quad\Longrightarrow\quad \|a_r(n)-a_p(n)\|_\varphi+ \|(a_r(n)-a_p(n))y(n)\|_\varphi<2^{-p}. \tag{8}

Start with A0=NA_0=\mathbb N, and intersect the finitely many sets where these inequalities hold with Ar−1A_{r-1} and the indicated cofinite set. Put r(n)=max⁡{r≤n:n∈Ar}r(n)=\max\{r\le n:n\in A_r\}, or zero if the set is empty, and take a(n)=ar(n)(n)a(n)=a_{r(n)}(n), with a0(n)=0a_0(n)=0. Each a(n)a(n) is a self-adjoint contraction. For every fixed pp, the set Ap+1A_{p+1} makes r(n)>pr(n)>p, so (8) gives

lim⁡n→ω(∥a(n)−ap(n)∥φ+∥(a(n)−ap(n))y(n)∥φ)≤2−p.(9)\lim_{n\to\omega} \bigl(\|a(n)-a_p(n)\|_\varphi+ \|(a(n)-a_p(n))y(n)\|_\varphi\bigr) \le2^{-p}. \tag{9}

This selection uses only finite intersections at each stage; no common intersection of all ArA_r is required.

To prove that aa is a multiplier, let z=(z(n))∈Iωz=(z(n))\in I_\omega, with ∥z(n)∥≤K\|z(n)\|\le K. Since every fixed row apa_p is a multiplier, the triangle inequality and (9) give

lim⁡n→ω∥z(n)a(n)∥φ≤lim⁡n→ω∥z(n)ap(n)∥φ+Klim⁡n→ω∥a(n)−ap(n)∥φ≤K2−p.(10)\lim_{n\to\omega}\|z(n)a(n)\|_\varphi \le\lim_{n\to\omega}\|z(n)a_p(n)\|_\varphi +K\lim_{n\to\omega}\|a(n)-a_p(n)\|_\varphi \le K2^{-p}. \tag{10}

Let pp increase to infinity. This proves the first null seminorm of zaza. Self-adjointness gives (za)∗=az∗(za)^*=az^*, and contractivity gives ∥a(n)z(n)∗∥φ≤∥z(n)∗∥φ\|a(n)z(n)^*\|_\varphi\le\|z(n)^*\|_\varphi; hence its other null seminorm vanishes as well. Apply the same argument to z∗z^* to obtain z∗a∈Iωz^*a\in I_\omega, whose adjoint is azaz. Thus both zaza and azaz belong to IωI_\omega, and a∈Nωa\in N_\omega.

We may now pass (9) to the quotient: pY(π(a)−Xp)≤2−pp_Y(\pi(a)-X_p)\le2^{-p}. The selected subsequence converges in the self-adjoint unit ball, and its original Cauchy sequence has the same limit. This proves completeness. □\square

Theorem 3.2. MωM^\omega is a von Neumann algebra in this GNS representation, and φω\varphi^\omega is faithful normal.

Proof. Suppose XiX_i is a strong-operator Cauchy net of self-adjoint contractions. For every Y∈MωY\in M^\omega, it is Cauchy for (7), since that expression tests the vectors Ω,YΩ\Omega,Y\Omega. Metric completeness applies to this net: choose increasing indices iri_r beyond a 2−r2^{-r}-Cauchy tail for that metric. The sequence XirX_{i_r} has a limit by Lemma 3.1, and the triangle inequality between the sequence and each sufficiently late net element proves convergence of the whole net. No countable cofinal subset of its directed index set is assumed. Thus there is an element XYX_Y with

(Xi−XY)Ω→0,(Xi−XY)YΩ→0.(11)(X_i-X_Y)\Omega\to0,\qquad (X_i-X_Y)Y\Omega\to0. \tag{11}

All the XYΩX_Y\Omega are the same limit. The state is faithful, so equality on Ω\Omega implies equality of these algebra elements. Denote their common value by XX. Equation (11) for all YY, density of MωΩM^\omega\Omega, and uniform boundedness give Xi→XX_i\to X strongly.

Thus the self-adjoint unit ball is strongly closed. Kaplansky density makes it strongly dense in the self-adjoint unit ball of the bicommutant. These balls therefore agree, and real and imaginary parts show that the whole represented algebra equals its bicommutant. The state is its vector state, hence normal, and remains faithful by Lemma 2.1. □\square

4. The normal copy of the original algebra

The constant map

ι:M→Mω,ι(a)=π(a,a,…)(12)\iota:M\to M^\omega,\qquad \iota(a)=\pi(a,a,\ldots) \tag{12}

is a unital injective *-homomorphism: a constant sequence can belong to IωI_\omega only if its operator is zero. Its state restriction is φω∘ι=φ\varphi^\omega\circ\iota=\varphi. It is normal. If 0≤ai↑a0\le a_i\uparrow a, the supremum Z=sup⁡iι(ai)Z=\sup_i\iota(a_i) in MωM^\omega satisfies Z≤ι(a)Z\le\iota(a), and

φω(ι(a)−Z)=φ(a)−lim⁡iφ(ai)=0.(13)\varphi^\omega(\iota(a)-Z) =\varphi(a)-\lim_i\varphi(a_i)=0. \tag{13}

Faithfulness gives Z=ι(a)Z=\iota(a). We henceforth identify MM with this normal copy.

5. The normal centralizing subalgebra

Now assume MM has separable predual, but allow nonfactor algebras. Let CωC_\omega consist of bounded sequences with ∥[xn,ψ]∥→ω0\|[x_n,\psi]\|\to_\omega0 for every ψ∈M∗\psi\in M_*, using the functional convention of the preceding central sequence lessons. Every element of IωI_\omega belongs to CωC_\omega, by the normal-functional Cauchy–Schwarz estimate.

Also Cω⊂NωC_\omega\subset N_\omega. To verify the potentially troublesome right multiplication, take ∥an∥≤C\|a_n\|\le C and ∥zn∥≤K\|z_n\|\le K, with aa centralizing and z∈Iωz\in I_\omega. Moving an∗a_n^* past the state gives

φ(an∗zn∗znan)≤∣φ(zn∗znanan∗)∣+CK2∥[an∗,φ]∥≤C2K ∥zn∥φ+CK2∥[an∗,φ]∥⟶ω0.(14)\begin{aligned} \varphi(a_n^*z_n^*z_na_n) &\le |\varphi(z_n^*z_na_na_n^*)| +CK^2\|[a_n^*,\varphi]\|\\ &\le C^2K\,\|z_n\|_\varphi +CK^2\|[a_n^*,\varphi]\|\longrightarrow_\omega0. \end{aligned} \tag{14}

The second line is Cauchy–Schwarz applied to the positive bounded factors. The other right/left adjoint estimate is identical with znzn∗z_nz_n^* and [an,φ][a_n,\varphi]. The two remaining seminorms are bounded directly by C∥zn∥φC\|z_n\|_\varphi or C∥zn∗∥φC\|z_n^*\|_\varphi.

Theorem 5.1. The inclusion induces a normal embedding of the finite von Neumann algebra

Mω=Cω/Iω⊂Mω.(15)M_\omega=C_\omega/I_\omega\subset M^\omega. \tag{15}

Its faithful normal trace is τω=φω∣Mω\tau_\omega=\varphi^\omega|_{M_\omega}. In the factor case this is the previously defined scalar-limit trace.

Proof. The kernel of inclusion on CωC_\omega is exactly IωI_\omega, so the quotient embeds isometrically. Its state is faithful by Lemma 2.1, and is tracial because

∣φ(xnyn−ynxn)∣≤∥yn∥∥[xn,φ]∥→ω0.(16)|\varphi(x_ny_n-y_nx_n)| \le\|y_n\|\|[x_n,\varphi]\|\to_\omega0. \tag{16}

Consequently its 22-norm is the ultralimit of the symmetric seminorm (1).

Here is the completeness argument without assuming a scalar weak limit. Choose a norm-dense sequence of normal states ψj\psi_j. For a fast 22-Cauchy subsequence XrX_r of quotient contractions, choose centralizing contraction representatives xr(k)x_r(k), with ∥Xr−Xp∥2<2−p\|X_r-X_p\|_2<2^{-p} for r>pr>p. Choose nested sets Ar∈ωA_r\in\omega, inside {k≥r}\{k\ge r\}, on which

∥[xr(k),ψj]∥≤1/r(j≤r),∥xr(k)−xp(k)∥φ,#≤2−p+2−r(p<r).(17)\begin{aligned} \|[x_r(k),\psi_j]\|&\le1/r&&(j\le r),\\ \|x_r(k)-x_p(k)\|_{\varphi,\#}&\le2^{-p}+2^{-r}&&(p<r). \end{aligned} \tag{17}

Each finite collection is available by the ultralimit trace identity and centralizing. Set r(k)=max⁡{r≤k:k∈Ar}r(k)=\max\{r\le k:k\in A_r\}, or zero when empty, and x(k)=xr(k)(k)x(k)=x_{r(k)}(k), using x0=0x_0=0. Since r(k)→ω∞r(k)\to_\omega\infty, the first line and norm density make xx centralizing. The second line gives ∥π(x)−Xp∥2≤2−p\|\pi(x)-X_p\|_2\le2^{-p}. The unit ball is therefore 22-complete.

For a strong Cauchy net in this represented unit ball, convergence on the faithful vector Ω\Omega makes it 22-Cauchy. Completeness gives a limit in MωM_\omega. Trace equality makes both its first and adjoint state seminorms converge, so the bounded faithful-state strong* criterion in MωM^\omega gives strong convergence to that limit. Kaplansky density now proves that this unit ball, and hence the subalgebra, is strongly closed. It is a von Neumann subalgebra of MωM^\omega, with the normal restriction of its faithful state. This proves normality of the embedding and finiteness. □\square

6. Automorphisms and convergent implementing families

Theorem 6.1. A fixed normal automorphism β\beta acts term by term on MωM^\omega. More generally, if βn→β\beta_n\to\beta in the uu-topology, the sequence action induces a normal automorphism

Γ(βn)(π(xn))=π(βn(xn)).(18)\Gamma_{(\beta_n)}(\pi(x_n))=\pi(\beta_n(x_n)). \tag{18}

It restricts to β\beta on the constant copy of MM. At separable-predual generality it also preserves MωM_\omega.

Proof. For bounded z∈Iωz\in I_\omega,

φ(βn(zn∗zn))≤∥φ∘βn−φ∘β∥∥zn∥2+(φ∘β)(zn∗zn)→ω0.(19)\varphi(\beta_n(z_n^*z_n)) \le \|\varphi\circ\beta_n-\varphi\circ\beta\|\|z_n\|^2 +(\varphi\circ\beta)(z_n^*z_n)\to_\omega0. \tag{19}

The adjoint seminorm has the same estimate. Inverses converge in the uu-topology as well. For any fixed ψ∈M∗\psi\in M_*, precomposition by βn\beta_n is isometric, so

∥ψ∘βn−1−ψ∘β−1∥=∥ψ−(ψ∘β−1)∘βn∥⟶0.\|\psi\circ\beta_n^{-1}-\psi\circ\beta^{-1}\| =\|\psi-(\psi\circ\beta^{-1})\circ\beta_n\| \longrightarrow0.

The last limit tests the fixed normal functional ψ∘β−1\psi\circ\beta^{-1}. This needs neither a metric nor a separable Banach space. Thus the termwise sequence automorphism preserves IωI_\omega in both directions. It preserves its multiplier algebra, so induces a *-automorphism of the quotient. A bijective *-homomorphism between von Neumann algebras is normal: its order inverse preserves the least-upper-bound property of every increasing bounded positive net.

For fixed a∈Ma\in M, uu-convergence gives βn(a)→β(a)\beta_n(a)\to\beta(a) strong*. Expand each state square norm of their difference; multiplicativity makes the square term βn(a∗a)\beta_n(a^*a), and the cross terms involve fixed normal functionals applied to βn(a)\beta_n(a). Their limits cancel. The adjoint expansion is the same. This proves the constant restriction.

For a centralizing xnx_n, the predual identity gives

∥[βn(xn),ψ]∥≤∥[xn,ψ∘β]∥+2sup⁡n∥xn∥∥ψ∘βn−ψ∘β∥→ω0.(20)\|[\beta_n(x_n),\psi]\| \le\|[x_n,\psi\circ\beta]\| +2\sup_n\|x_n\|\|\psi\circ\beta_n-\psi\circ\beta\| \to_\omega0. \tag{20}

The inverse family gives equality of the invariant subalgebra. □\square

A constant automorphism family is called liftable; a convergent variable family is semi-liftable. Formula (18) keeps the family in its notation: the limit β\beta alone does not determine that automorphism. The next lesson gives the exact counterexample.

If Ad⁡(un)→α\operatorname{Ad}(u_n)\to\alpha, the unitaries unu_n belong to NωN_\omega. Indeed, the only nonautomatic square seminorms of their products with zn∈Iωz_n\in I_\omega are tested by φ∘Ad⁡(un)\varphi\circ\operatorname{Ad}(u_n) or its inverse; these converge in predual norm to the corresponding fixed normal states of α\alpha. The estimate (19) applies. Thus U=π(un)U=\pi(u_n) is a unitary and

Ad⁡(U)∣M=α.(21)\operatorname{Ad}(U)|_M=\alpha. \tag{21}

This realizes an approximately inner automorphism of MM by an actual inner automorphism in the larger algebra.

7. Dense representatives and type I factors

Lemma 7.1. If A0⊂MA_0\subset M is an ultraweakly dense *-subalgebra, every X∈MωX\in M^\omega has a bounded multiplier representative bn∈A0b_n\in A_0.

Proof. Start with a multiplier representative ana_n, uniformly bounded by CC. If C=0C=0 use zero. Otherwise Kaplansky density for the norm closure of A0A_0, followed by norm approximation from A0A_0, gives bn∈A0b_n\in A_0 with ∥bn∥≤2C\|b_n\|\le2C and ∥bn−an∥φ,#<1/n\|b_n-a_n\|_{\varphi,\#}<1/n. The difference belongs to Iω⊂NωI_\omega\subset N_\omega; hence bb is a multiplier and π(b)=X\pi(b)=X. No assumption that A0A_0 contains the unit is needed. □\square

Theorem 7.2. If M=B(H)M=B(H) is a countably decomposable type I factor, then every multiplier sequence converges strong* along ω\omega. The constant embedding is a normal isomorphism M≅MωM\cong M^\omega.

Proof. The finite-dimensional case follows from compactness of bounded operator balls. In infinite dimension HH is separable. Indeed, the rank-one projections from any orthonormal basis have strictly positive values under a faithful normal state, and every finite sum of those values is at most one. For each positive integer rr, at most rr of them can have value at least 1/r1/r; their union exhausts the basis. Choose a basis ξj\xi_j and the faithful diagonal state

φ0(x)=∑j≥02−j−1⟨ξj,xξj⟩.(22)\varphi_0(x)=\sum_{j\ge0}2^{-j-1}\langle\xi_j,x\xi_j\rangle. \tag{22}

Changing the faithful state does not change (2). Put PmP_m for the first mm basis projection and qm=1−Pmq_m=1-P_m. Then ∥qm∥φ0,#→0\|q_m\|_{\varphi_0,\#}\to0.

Let a∈Nωa\in N_\omega, and let AA be its bounded ultraweak ω\omega-limit. Given a fixed jj, Lemma 1.2, applied to qmq_m for sufficiently large mm, makes both

∥qmanξj∥,∥qman∗ξj∥(23)\|q_ma_n\xi_j\|,\qquad \|q_ma_n^*\xi_j\| \tag{23}

arbitrarily small on an ω\omega-large set. For the first, the nonadjoint seminorm of qmanq_ma_n includes the square in (23) with coefficient 2−j−1/22^{-j-1}/2; for the second use the adjoint seminorm of anqma_nq_m.

In the finite-dimensional range of PmP_m, the vectors PmanξjP_ma_n\xi_j converge in norm along ω\omega to PmAξjP_mA\xi_j, by ultraweak convergence of their finitely many coefficients. Choose mm also large enough that qmAξjq_mA\xi_j is small. Equation (23) then proves anξj→ωAξja_n\xi_j\to_\omega A\xi_j. The same argument for adjoints gives an∗ξj→ωA∗ξja_n^*\xi_j\to_\omega A^*\xi_j. Uniform boundedness extends both conclusions from the basis span to every vector. Thus an−A∈Iωa_n-A\in I_\omega, and every quotient element is constant. Normality and injectivity of the constant embedding were proved in Section 4. □\square

8. Exercises with complete solutions

Exercise 1. Verify that the null space in (2) is self-adjoint and product closed.

Solution. Adjointing interchanges the two state seminorms. In a faithful normal representation, if xn,yn→ω0x_n,y_n\to_\omega0 strong* with norm bounds C,DC,D, then ∥xnynξ∥≤C∥ynξ∥→0\|x_ny_n\xi\|\le C\|y_n\xi\|\to0, and ∥(xnyn)∗ξ∥≤D∥xn∗ξ∥→0\|(x_ny_n)^*\xi\|\le D\|x_n^*\xi\|\to0. Thus the products vanish strong* as well.

Exercise 2. In the rank-one example in Section 1, compute the symmetric seminorm of pnanp_na_n for the state (22).

Solution. Since pnan=anp_na_n=a_n, an∗an=p0a_n^*a_n=p_0 and anan∗=pna_na_n^*=p_n. Its squared symmetric seminorm is (2−1+2−n−1)/2(2^{-1}+2^{-n-1})/2, which tends to 1/41/4, rather than zero. Meanwhile ∥pn∥φ0,#2=2−n−1→0\|p_n\|_{\varphi_0,\#}^2=2^{-n-1}\to0. This proves both the nonideal phenomenon and failure of the multiplier condition.

Exercise 3. Why does the diagonal witness in Lemma 1.2 avoid a countable intersection of ultrafilter sets?

Solution. For each fixed rr, Br∩{n≥r}B_r\cap\{n\ge r\} is an ω\omega-large set on which r(n)≥rr(n)\ge r. These separate threshold bounds give r(n)→ω∞r(n)\to_\omega\infty. A common intersection over all rr is neither required nor asserted.

Exercise 4. Explain why multipliers commute with the coordinate projections qnq_n.

Solution. The element mqnmq_n lies in IωI_\omega and has right support qnq_n, since (mqn)qn=mqn(mq_n)q_n=mq_n. Centrality of qnq_n within IωI_\omega makes it also left-supported there. The same argument applies to qnmq_nm, and both equal qnmqnq_nmq_n. Their coordinate value therefore lies in the unital coordinate algebra MM.

Exercise 5. Prove faithfulness of the quotient state on a positive element without assuming the state is tracial.

Solution. Lift the positive quotient element as h=b∗bh=b^*b in the multiplier algebra. If its state is zero, the positive operator inequality hn2≤Chnh_n^2\le Ch_n gives vanishing of φ(hn2)\varphi(h_n^2). Self-adjointness makes this both strong* seminorms, so h∈Iωh\in I_\omega. The quotient element is zero. No cyclic trace interchange occurs.

Exercise 6. Verify that changing a representative outside an ω\omega-large set leaves both the quotient and multiplier membership unchanged.

Solution. The bounded difference is zero on that set, so its symmetric seminorm has ultralimit zero. It belongs to IωI_\omega, which is an ideal inside NωN_\omega and itself lies in NωN_\omega. Adding it preserves multiplier membership and the quotient image.

Exercise 7. Explain why self-adjointness is used in the multiplier step of Lemma 3.1.

Solution. The diagonal representative is a self-adjoint contraction at every coordinate. Its one-sided state distance from each fixed multiplier row has ultralimit at most 2−p2^{-p}. For a bounded null sequence zz, (10) therefore makes the first seminorm of zaza vanish as pp increases. Self-adjointness gives (za)∗=az∗(za)^*=az^*, whose first seminorm is at most ∥z∗∥φ\|z^*\|_\varphi. Applying this same pair of estimates to z∗z^* makes z∗az^*a null in both seminorms; its adjoint azaz is consequently null as well. Thus both products belong to IωI_\omega. A one-sided estimate for a general nonsymmetric representative would not give these adjoint identities or both multiplier conditions.

Exercise 8. In Theorem 3.2, why do all the limits XYX_Y agree?

Solution. They give the same vector limit on Ω\Omega, because the first term of every pYp_Y is the same norm. Thus (XY−XZ)Ω=0(X_Y-X_Z)\Omega=0. Faithfulness of the state implies φω((XY−XZ)∗(XY−XZ))=0\varphi^\omega((X_Y-X_Z)^*(X_Y-X_Z))=0 only when XY=XZX_Y=X_Z. This produces one algebra element controlling every dense cyclic vector.

Exercise 9. Prove normality of the constant embedding by increasing positive nets.

Solution. If ai↑aa_i\uparrow a, the image supremum ZZ is bounded by ι(a)\iota(a). Normality of φ\varphi and the vector-state normality of φω\varphi^\omega give zero state on the positive difference, as in (13). Faithfulness makes that difference zero, so the embedding preserves every such supremum.

Exercise 10. When MM has a faithful normal tracial state, show that Nω=ℓ∞(N,M)N_\omega=\ell^\infty(\mathbb N,M).

Solution. Use that trace to define the unchanged ideal. For bounded ana_n, the tracial 22-norm inequalities ∥anzn∥2,∥znan∥2≤sup⁡n∥an∥∥zn∥2\|a_nz_n\|_2,\|z_na_n\|_2\le\sup_n\|a_n\|\|z_n\|_2 show preservation on both sides. Thus every bounded sequence is a multiplier. The ideal restriction is needed in the general nontracial construction.

Exercise 11. Derive the predual error term in (20).

Solution. Replace ψ∘βn\psi\circ\beta_n by the fixed functional ψ∘β\psi\circ\beta inside [xn,⋅][x_n,\cdot]. The uniform estimate ∥[xn,ρ]∥≤2∥xn∥∥ρ∥\|[x_n,\rho]\|\le2\|x_n\|\|\rho\| bounds the difference by the displayed second term. Normalizing by βn−1\beta_n^{-1} does not change its norm. The fixed commutator and the predual error both tend to zero.

Exercise 12. Why is the implementing unitary in (21) allowed to have a nonscalar quotient even though the induced map has a prescribed limit on MM?

Solution. Membership in NωN_\omega means preservation of strong*-null sequences, rather than scalar-triviality. Its quotient unitary can contain additional asymptotic information. Equation (21) specifies its action on the constant copy of MM; it does not determine its action on all varying quotient elements. The tensor-tail counterexample in the next lesson exhibits this distinction.

Exercise 13. Produce bounded representatives from a dense *-subalgebra that does not contain 11.

Solution. Let A0A_0 be the finite-rank operators in B(ℓ2)B(\ell^2), and let ana_n represent XX with bound C>0C>0. Kaplansky approximation in their norm closure, followed by finite-rank norm approximation, gives bn∈A0b_n\in A_0, bounded by 2C2C, with symmetric state error less than 1/n1/n. Their difference lies in IωI_\omega, so they represent the same element and remain multipliers. For the unit itself one can simply use increasing finite-rank projections.

Exercise 14. Identify the role of finite-dimensional compactness in Theorem 7.2.

Solution. The multiplier condition makes the tails in (23) uniformly small on an ω\omega-large set. Within a fixed finite-dimensional range, weak coefficient convergence is norm convergence. Adding the small tails gives norm convergence of each full column and row. Ultraweak convergence alone, without the multiplier tail control, would not give this conclusion.

References

Adrian Ocneanu's freely accessible Actions of discrete amenable groups on factors, thesis, Chapter 5, Sections 5.1–5.2, printed pp.47–52 (PDF pp.61–66), gives the normalizing quotient, two-vector completeness, normal subalgebras and convergent automorphism families. Its completeness proof modifies representatives and takes an operator limit at each coordinate. Here finite simultaneous row tests select one diagonal representative, and the limiting multiplier estimates establish both null seminorms before passing to the quotient. Sections 1–4 and 6–7 require a given faithful normal state; the centralizing subalgebra in Section 5 explicitly retains separable predual. The normality proof handles arbitrary strong-Cauchy nets, and a semi-liftable automorphism keeps its chosen family.

Hiroshi Ando and Uffe Haagerup, Ultraproducts of von Neumann algebras, v3, Lemma 3.13, gives the simultaneous-vector diagonal method comparison. Lemmas 3.8 and 3.10, Propositions 3.14–3.15 and Theorem 3.7 give a separate support-corner realization for sigma-finite algebras. Their Theorem 3.9 quotes the left-ideal/hereditary correspondence from another text; no such correspondence or standard-form corner construction is a premise of the direct completeness argument here. Full faithful-state topology, quotient functional calculus and bounded density proofs are supplied in the programme. Human-source rights remain separate from the course dedication.