Given a data function together with an error corridor for the scattering amplitude along some finite part of the cuts, one can construct effectively the whole set of analytic functions (``admissible amplitudes''), compatible with these conditions and bounded by a certain number M on the remaining part of the cuts. Depending on the actual value of an important constant ε0 computed from the data function and the bound M, this set may be void. If not, in every point of the cut plane the set of values of the admissible amplitudes fills densely a circle; explicit formulas are given for its radius η̂(z) and center f̂(z), the latter being the best possible estimate for the whole set. In contrast to the linear extrapolation obtained by Poisson weighted dispersion relations, here nonlinear functional methods were used. This paper contains an appendix written by Professor C. Foias, on some functional analytical methods used in connection with the computation of the numerical value of the constant ε0.

Topics
Complex functions, Scattering theory
1.
R. E.
Cutkosky
 and
B. B.
Deo
,
Phys. Rev.
174
,
1859
(
1968
).
Google Scholar
Crossref
Search ADS
 
2.
S.
Ciulli
,
Nuovo Cimento A
61
,
787
(
1969
);
Google Scholar
Crossref
Search ADS
 
S.
Ciulli
,
Nuovo Cimento A
62
,
301
(
1969
).
Google Scholar
Crossref
Search ADS
 
3.
R. E.
Cutkosky
,
Ann. Phys. (N.Y.)
54
,
110
(
1969
).
Google Scholar
Crossref
Search ADS
 
4.
J. E.
Bowcock
 and
J.
John
,
Nucl. Phys. B
11
,
659
(
1969
).
Google Scholar
Crossref
Search ADS
 
5.
R. E. Cutkosky, “Partial Wave Phenomenology,” Carnegie Mellon Univ. preprint talk at the ZGS Users Conference, Argonne National Laboratory, May, 1970.
6.
S. Ciulli and G. Nenciu, “Cut to cut extrapolation,” talk at the Lund Conference, May 1969,
and
Commun. Math. Phys.
26
,
237
(
1972
).
Crossref
Search ADS
 
7.
S.
Ciulli
 and
J.
Fischer
,
Nucl. Phys. B
24
,
537
(
1970
).
Google Scholar
Crossref
Search ADS
 
8.
G.
Nenciu
,
Nuovo Cimento Lett.
4
,
96
(
1970
).
Google Scholar
Crossref
Search ADS
 
9.
G.
Calucci
,
L.
Fonda
, and
G. C.
Ghiraldi
,
Phys. Rev.
166
,
1719
(
1968
).
Google Scholar
Crossref
Search ADS
 
10.
A. N.
Tihonov
,
Dokl. Akad. Nauk SSSR
159
,
501
(
1963
).
Google Scholar
 
11.
Usually one maps the whole cut z plane into the ζ‐unit circle; nevertheless, if information on the position of the singularities in the second sheet is available, one could map also parts of the superior Riemann sheets into the unit disk, enhancing in a substantial way [see problem (i), Sec. 3 of Ref. 2] the stability of the extrapolation.
12.
I.
Schur
,
J. Reine Angew. Math.
147
,
205
(
1917
).
Google Scholar
 
13.
Z.
Nehari
,
Ann. Math.
65
,
153
(
1957
).
Google Scholar
Crossref
Search ADS
 
14.
V. M.
Adamjan
,
D. Z.
Arov
,
M. G.
Krein
 (in Russian),
Funk. Anal. ego Prid.
2
, (a) (
4
),
1
(
1968
);
Google Scholar
 
see also
V. M.
Adamjan
,
D. Z.
Arov
,
M. G.
Krein
,
Funk. Anal. ego Prid.
2
, (b) (
1
) (
1968
).
Google Scholar
 
15.
C.
Foias
 and
B. Sz.
Nagy
,
C. R. Acad. Sci. (Paris)
266
,
493
(
1968
);
Google Scholar
 
B. Sz. Nagy and C. Foias, Harmonic Analysis of Hilbert Space Operators (North‐Holland, Amsterdam, 1970).
16.
To find Yχ+ [see Eq. (3.10)] from Eq. (3.7a) one first notices that the θ‐reflection operator Tφ≡φ̌ is self‐adjoint. Indeed, if un are the basis vectors exp (in θ), we have (T+umun)≡(um,Tun)≡(um,Ǔn) = (um,u−n) = δm,−n and hence T+um = u−m, i.e., T+ = T. Further, observing that TU = U+T, starting from (3.7̄a) one gets Yχ+ = (χ̌*U+T)+ = T+Uχ̌ = U+χT = χU+T, where the last equality holds because both χ and U+ are multiplication operators.
17.
This follows directly from the definition of the norm of an operator as being the maximum of the numbers ‖Yφ‖/‖φ‖ for all possible functions φ belonging to the space (or subspace) under consideration.
18.
R.
Douglas
,
P.
Muhly
, and
C.
Pearcy
,
Mich. Math. J.
15
,
385
(
1968
).
Google Scholar
Crossref
Search ADS
 
19.
L. B.
Page
,
Indiana Math. J.
20
,
20
(
1971
).
Google Scholar
 
20.
All these weight functions (⁠C0,C1, and so on) defined by the values of their moduli on the boundary and having no zeroes inside D are commonly known to mathematicians under the name of “exterior functions.” The name “truncated Carleman kernels” or “Carleman functions” was introduced by us at the seminar at the 1969 Lund Conference6 in analogy with Lavrentiev’s “Carleman kernels,” but this appears to have produced much confusion in the literature, since references to Carleman’s book on quasianalytic functions are now floating around although there is no trace in it of “Carleman functions” at all! Similar weight functions were used, for instance, in the weighted L2 norms [see Eq. (5.11)] in the early papers of Szegö and one can certainly say that the weight functions are as old as analysis itself. Therefore, in order to avoid further confusions, it might be better to call them simply “exterior weight functions”.
21.
A. N.
Kolmogoroff
,
Giornale Ist. Ital. Attuari
4
,
83
(
1933
).
Google Scholar
 
22.
We recall that the repartition function F(ξ0) of a random variable ξ, is nothing but the probability of finding a ξ less than ξ0.
23.
A.
Rényi
, “
On the theory of order statistics
,”
Acta Mathematica Acad. Sci. Hungarial.
4
(
1953
).
Google Scholar
 
24.
Sabba‐Stefǎnescu
,
Nucl. Phys. B
56
,
287
(
1973
).
Google Scholar
Crossref
Search ADS
 
25.
S.
Ciulli
 and
G.
Nenciu
,
Nuovo Cim.
8
,
735
(
1972
).
Google Scholar
Crossref
Search ADS
 
26.
I.
Capriani
,
S.
Ciulli
,
A.
Pomponiu
, and
J.
Sabba‐Stefǎnescu
,
Phys. Rev. D
5
,
1658
(
1972
).
Google Scholar
Crossref
Search ADS
 
27.
G. Nenciu, “The Best Analytic Continuation of the Scattering Amplitudes,” invited talk at the IV Colloquium on Theoretical Physics and Elementary Particles, Prague, 30 Sept.–2 Oct. 1970.
This content is only available via PDF.
© 1973 The American Institute of Physics.
1973
The American Institute of Physics
You do not currently have access to this content.