Scribd
Upload
12 views3 pages

Gaur 2

The paper establishes the equivalence of various norms on the unitization of nonunital Banach algebras, providing uniform bounds for these norms. A tighter bound is specifically derived for C*-algebras involving Hermitian elements. The results contribute to the understanding of norm extensions and their implications in the context of Banach algebras.

Uploaded by

jackdabhi07
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
12 views3 pages

Gaur 2

The paper establishes the equivalence of various norms on the unitization of nonunital Banach algebras, providing uniform bounds for these norms. A tighter bound is specifically derived for C*-algebras involving Hermitian elements. The results contribute to the understanding of norm extensions and their implications in the context of Banach algebras.

Uploaded by

jackdabhi07
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

proceedings of the

american mathematical society


Volume 117, Number 1, January 1993

NORMS ON UNITIZATIONS OF BANACHALGEBRAS

A. K. GAUR AND Z. V. KOVARfK

(Communicated by Palle E. T. Jorgensen)

Abstract. Equivalence of various norms on the unitization of a nonunital


Banach algebra is established, with bounds (1 and 6exp(l)) uniform over the
class of such algebras. A tighter bound, 3, is obtained in C*-algebras for ele-
ments with Hermitian nonunital parts.

The algebra norm || • || on a nonunital Banach algebra A can be extended to


an algebra norm on the unitization A+ in many ways. Proposition 4.3 in [3]
states that among these extensions, the /i-norm
||Ae+ a||i = |A|-(-||a||
is maximal and the operator norm
\\Xe+ a\\op = sup{||Ax + ax\\ : \\x\\ < 1}
is minimal, provided that it does extend || • ||, i.e., that || • || is a regular
(= operator) norm.
In the latter case, A+ is complete under both || • ||, and || • ||op, so by the
"two-norm lemma" [2, II.2.5] these two norms are equivalent; the pure existence
nature of the lemma does not yield an explicit bound M in || • ||i < M\\ • ||op
and such a bound seems to depend on the algebra A .
The present theorem establishes uniform equivalence of the two unitization
norms over the class of nonunital Banach algebras with regular norms.
Theorem. For every nonunital Banach algebra A with unitization A+ and with
regular norm, and for every X £ C and a £ A, we have
\\Xe+ a\\op < ||/te + a||i < (6exp 1)||A^+ a||op.
If A is a C*-algebra, a £ A is hermitian, and X is complex then
\\Xe+ a\\x <3||Ae + a||op
and the constant 3 is best (minimal) possible.
Proof. In a general algebra A with a regular norm, we have an extension of the
classical inequality for the numerical radius v(a) [1, Theorem 4.1]:
v(a) < \\a\\ < (expl)v(a).
Received by the editors March 12, 1991 and, in revised form, May 13, 1991; presented at the
AMS Baltimore Meeting (871) on January 10, 1992.
1991 Mathematics Subject Classification. Primary 46J99, 46J15.
Key words and phrases. Banach algebra, unitization, equivalent norms, Hermitian element.

© 1993 American Mathematical Society


0002-9939/93 $1.00 + $.25 per page

111
112 A. K. GAUR AND Z. V. KOVARIK

Without loss of generality, assume that a ^ 0. We know that the closure K of


the numerical range of a in a nonunital algebra contains 0; our first task is to
estimate v(X + a) from below: From the geometry of the complex plane we see
that the diameter d of the compact K is realized as the distance d = \a —B\
with a, B £ K, and comparison with the special case Xo= -(a + /?)/2 leads
to
v(Xe+ a) = maxiJA+ £,\:£,eK}>\d.
Also, since 0 G K, we have d >v(a). Altogether,
v(Xe + a) > \v(a).
Now we split estimates into cases \X\< 2||a || and \X\> 2\\a\\. The former case
gives
||Ag + fl||op> v(a)/2 > v(a)/2 = 1 .
\X\+ \\a\\ ~ 2\\a\\ + (exp l)v(a) ~ (3expl)v(a) 6expl'
the latter case \X\> 2\\a\\ gives, using the triangle inequality and the fact that
the fraction in the middle increases with \X\,
\\Xe+ a\\op \X\-\\a\\ 1
|A|+ |N -|a| + H-3-
We conclude that for all complex X,
||A+ fl||i <(6expl)||A + a||op.
Now the C*-algebra case: The closure of the numerical range of a Hermitian
a is the smallest real interval [a, /?] containing the spectrum of a, and for all
complex X we have
p.e + a||. = |A| + max(|a|, \B\),
\\Xe+ a\\op = max(|A + a\,\X + B\).
The expression to minimize is
max(|A + a|,|A + ^|)
q{ > |A| + max(|a|,|/?|) •
Without loss of generality, we assume that a < 0 < B and y = (a + B)/2 > 0
(recall that 0 is in the spectrum of a); otherwise we replace a with -a .
From now on, this is a problem about complex numbers. We split it into
four cases:
(CI) A real,
(C2) A not real, 3U > -y,
(C3) A is not real, SRA < -y,
(C4) A not real, 3tt = -y.
In (CI) q is continuous, piecewise monotone with breakpoints —B, 0, -y,
-a, and respective values,
B + \a\ 1 §__. B + \a\ 1 B + \a\
2$ -2' B ' 3j8-|a|-3' B + \a\
and q approaches 1 as \X\ —»oo. The best we can say about q, therefore, is
q>\, attained when a = 0.
NORMSON UNITIZATIONSOF BANACHALGEBRAS 113

Case (C4). Write, for symmetry, a — y - p < 0, B = y + p>0,so that


p = (B - a)/2. Also, we substitute p = -y + >Jy2 + v2 (note p > 0), so that
v2 = p2 + 2py. To prove that Q(v) = q(-y + iv) > \ , write

Q(v) = _!i±4_ = Vp2 + Vl


^ ' \X\+ max(M ,\B\) ^T^A + y + p'
2 _\_ 9(p2+p2 + 2py)-(p + p + 2y)2
^[) 9 9(p+p + 2y)2

= 2(p-p-y)2 + 6(p + y)(p-y + p) >Q


9(p + p + 2y)2
since both p + y > 0 and p-y+p = \a\+p>0.
Cases (C2) and (C3). Except on the set {A|3tA= -y or A = 0}, q has a
gradient
VqW=(i±jm±m^z^jm form>-y,
(X+ a)(\X\+ P)/\X + a\-\X + a\X/\X\
(\M+ P)2
Remark. The bound 6exp 1 is not the best; by splitting at (1 + l/(2exp l))||a||
instead of at 2||a|| in the proof, we could reduce the bound 6expl to 1 +
4 exp 1, but we suspect that even this can be improved.

References
1. F. F. Bonsall and J. Duncan, Numerical ranges of operators on normed spaces and of elements
of normed algebras, London Math. Soc. Lecture Notes Ser., vol. 2, Cambridge Univ. Press,
Cambridge and New York, 1971.
2. N. Dunford and J. T. Schwartz, Linear operators, Part I, Interscience, New York, 1958.
3. A. K. Gaur and Z. V. Kovah'k, Norms, states and numerical ranges on direct sums, Analysis,
Internat. Math. J. Anal. Appl. 11 (1991), 155-164.

Department of Mathematics, Duquesne University, Pittsburgh, Pennsylvania 158282


Department of Mathematics and Statistics, McMaster University, Hamilton, On-
tario, Canada L8S 4K1

You might also like