|
|
Line 1: |
Line 1: |
− | ''on a left <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h0470502.png" />-module <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h0470503.png" />''
| + | <!-- |
| + | h0470201.png |
| + | $#A+1 = 17 n = 0 |
| + | $#C+1 = 17 : ~/encyclopedia/old_files/data/H047/H.0407020 Hermite problem |
| + | Automatically converted into TeX, above some diagnostics. |
| + | Please remove this comment and the {{TEX|auto}} line below, |
| + | if TeX found to be correct. |
| + | --> |
| | | |
− | A mapping <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h0470504.png" /> that is linear in the first argument and satisfies the condition
| + | {{TEX|auto}} |
| + | {{TEX|done}} |
| | | |
− | <table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h0470505.png" /></td> </tr></table>
| + | The problem about the homogeneous arithmetical minima of positive $ n $- |
| + | ary quadratic forms with real coefficients. It is equivalent to the problem of the densest lattice packing of $ n $- |
| + | dimensional balls of equal radius (see [[Geometry of numbers|Geometry of numbers]]). |
| | | |
− | Here <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h0470506.png" /> is a ring with a unit element and equipped with an involutory anti-automorphism <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h0470507.png" />. In particular, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h0470508.png" /> is a [[Sesquilinear form|sesquilinear form]] on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h0470509.png" />. The module <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705010.png" /> itself is then called a Hermitian space. By analogy with what is done for bilinear forms, equivalence is defined for Hermitian forms (in another terminology, isometry) and, correspondingly, isomorphism (isometry) of Hermitian spaces (in particular, automorphism). All automorphisms of a Hermitian form <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705011.png" /> form a group <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705012.png" />, which is called the unitary group associated with the Hermitian form <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705013.png" />; its structure has been well studied when <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705014.png" /> is a skew-field (see [[Unitary group|Unitary group]]).
| + | Let $ f = f ( x) $, |
| + | $ x \in \mathbf R ^ {n} $, |
| + | be a positive quadratic form over $ \mathbf R $ |
| + | of determinant $ d = d ( f ) = \mathop{\rm det} f \neq 0 $; |
| + | and let |
| | | |
− | A Hermitian form is a special case of an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705016.png" />-Hermitian form (where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705017.png" /> is an element in the centre of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705018.png" />), that is, a sesquilinear form <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705019.png" /> on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705020.png" /> for which
| + | $$ |
| + | m ( f ) = \inf _ {\begin{array}{c} |
| + | x \in \mathbf Z ^ {n} \\ |
| | | |
− | <table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705021.png" /></td> </tr></table>
| + | x \neq 0 |
| + | \end{array} |
| + | } |
| + | f ( x) = \min _ {\begin{array}{c} |
| + | x \in \mathbf Z ^ {n} \\ |
| | | |
− | When <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705022.png" />, an <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705023.png" />-Hermitian form is Hermitian, and when <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705024.png" /> the form is called skew-Hermitian or anti-Hermitian. If <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705025.png" />, a Hermitian form is a symmetric bilinear form, and a skew-Hermitian form is a skew-symmetric or anti-symmetric bilinear form. If the mapping
| + | {x \neq 0 } |
| + | \end{array} |
| + | } f ( x) |
| + | $$ |
| | | |
− | <table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705026.png" /></td> </tr></table>
| + | be its homogeneous arithmetical minimum. The quantity |
| | | |
− | where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705027.png" /> for any <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705028.png" />, is bijective, then <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705029.png" /> is called a non-degenerate Hermitian form or a Hermitian scalar product on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705030.png" />.
| + | $$ |
| + | \gamma _ {n} = \sup |
| + | \frac{m ( f ) }{\{ d ( f ) \} ^ {1/n} } |
| + | = \max \ |
| | | |
− | If <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705031.png" /> is a free <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705032.png" />-module with a basis <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705033.png" />, then the matrix <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705034.png" />, where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705035.png" />, is called the matrix of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705036.png" /> in the given basis; it is a [[Hermitian matrix|Hermitian matrix]] (that is, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705037.png" />). A Hermitian form <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705038.png" /> is non-degenerate if and only if <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705039.png" /> is invertible. If <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705040.png" /> is a skew-field, if <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705041.png" />, and if <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705042.png" /> is finite-dimensional over <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705043.png" />, then <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705044.png" /> has an orthogonal basis relative to <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705045.png" /> (in which the matrix is diagonal).
| + | \frac{m ( f ) }{\{ d ( f ) \} ^ {1/n} } |
| + | , |
| + | $$ |
| | | |
− | If <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705046.png" /> is a commutative ring with identity, if <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705047.png" />, and if the matrix of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705048.png" /> is definite, then its determinant lies in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705049.png" />. Under a change of basis in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705050.png" /> this determinant is multiplied by a non-zero element of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705051.png" /> of the form <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705052.png" />, where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705053.png" /> is an invertible element of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705054.png" />. The determinant regarded up to multiplication by such elements is called the determinant of the Hermitian form or of the Hermitian space <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705055.png" />; it is an important invariant and is used in the classification of Hermitian forms.
| + | where the supermum or maximum is over all positive quadratic forms $ f $, |
| + | is called the Hermite constant; $ \gamma _ {n} = \{ \gamma ( F _ {n} ) \} ^ {2} $, |
| + | where $ F _ {n} ( x) = ( x _ {1} ^ {2} + \dots + x _ {n} ^ {2} ) ^ {1/2} $ |
| + | is the radial function corresponding to a ball. |
| | | |
− | Let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705056.png" /> be commutative. Then a Hermitian form <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705057.png" /> on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705058.png" /> gives rise to a quadratic form <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705059.png" /> on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705060.png" /> over <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705061.png" />. The analysis of such forms lies at the basis of the construction of the Witt group of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705062.png" /> with an involution (see [[Witt ring|Witt ring]]; [[Witt decomposition|Witt decomposition]]; [[Witt theorem|Witt theorem]]). When <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705063.png" /> is a maximal ordered field, then the [[Law of inertia|law of inertia]] extends to Hermitian forms (and there arise the corresponding concepts of the signature, the index of inertia, and positive and negative definiteness). If <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705064.png" /> is a field and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705065.png" />, then <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705066.png" /> is a quadratic Galois extension of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705067.png" />, and isometry of two non-degenerate Hermitian forms over <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705068.png" /> is equivalent to isometry of the quadratic forms over <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705069.png" /> generated by them; this reduces the classification of non-degenerate Hermitian forms over <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705070.png" /> to that of non-degenerate quadratic forms over <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705071.png" />.
| + | Originally, one understood by the Hermite problem the task of finding or estimating $ \gamma _ {n} $( |
| + | from above and below). The exact values of $ \gamma _ {n} $ |
| + | are known only for $ n \leq 8 $( |
| + | see [[#References|[1]]]). For estimates of $ \gamma _ {n} $, |
| + | see [[#References|[2]]] or [[#References|[1]]]. |
| | | |
− | If <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705072.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705073.png" /> is the involution of complex conjugation, then a complete system of invariants of Hermitian forms over a finite-dimensional space is given by the rank and the signature of the corresponding quadratic forms. If <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705074.png" /> is a local field or the field of functions of a single variable over a finite field, then a complete system of invariants for non-degenerate Hermitian forms is given by the rank and the determinant. If <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705075.png" /> is a finite field, then there is only one invariant, the rank. For the case when <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705076.png" /> is an algebraic extension of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/h/h047/h047050/h04705077.png" />, see [[#References|[3]]]. Ch. Hermite was the first, in 1853, to consider the forms that bear his name in connection with certain problems of number theory.
| + | Subsequently, the term Hermite problem was used for the search for local maxima (boundary or extremal) of $ m ( f ) / \{ d ( f ) \} ^ {1/n} $ |
| + | in the space of coefficients and their corresponding forms $ f $. |
| + | Algorithms are known for finding all classes of boundary forms. In particular, the Voronoi algorithm for perfect forms (see [[#References|[1]]], [[#References|[3]]], [[#References|[4]]]). |
| + | |
| + | The problem was posed by Ch. Hermite in 1850. |
| | | |
| ====References==== | | ====References==== |
− | <table><TR><TD valign="top">[1]</TD> <TD valign="top"> N. Bourbaki, "Elements of mathematics. Algebra: Algebraic structures. Linear algebra" , '''1''' , Addison-Wesley (1974) pp. Chapt.1;2 (Translated from French) {{MR|0354207}} {{ZBL|}} </TD></TR><TR><TD valign="top">[2]</TD> <TD valign="top"> J.A. Dieudonné, "La géométrie des groups classiques" , Springer (1955) {{MR|}} {{ZBL|0221.20056}} </TD></TR><TR><TD valign="top">[3]</TD> <TD valign="top"> J. Milnor, D. Husemoller, "Symmetric bilinear forms" , Springer (1973) {{MR|0506372}} {{ZBL|0292.10016}} </TD></TR><TR><TD valign="top">[4]</TD> <TD valign="top"> O.T. O'Meara, "Introduction to quadratic forms" , Springer (1973) {{MR|}} {{ZBL|0259.10018}} </TD></TR></table> | + | <table><TR><TD valign="top">[1]</TD> <TD valign="top"> P.M. Gruber, C.G. Lekkerkerker, "Geometry of numbers" , North-Holland (1987) (Updated reprint) {{MR|0893813}} {{ZBL|0611.10017}} </TD></TR><TR><TD valign="top">[2]</TD> <TD valign="top"> C.A. Rogers, "Packing and covering" , Cambridge Univ. Press (1964) {{MR|0172183}} {{ZBL|0176.51401}} </TD></TR><TR><TD valign="top">[3]</TD> <TD valign="top"> B.N. Delone, "The Peterburg school of number theory" , Moscow-Leningrad (1947) (In Russian)</TD></TR><TR><TD valign="top">[4]</TD> <TD valign="top"> E.P. Baranovskii, "Packings, coverings, partitionings and certain other distributions in spaces of constant curvature" ''Progress in Math.'' , '''9''' (1971) pp. 209–253 ''Itogi Nauk. Mat. Algebra Topol. Geom. 1967'' (1969) pp. 189–225</TD></TR></table> |
The problem about the homogeneous arithmetical minima of positive $ n $-
ary quadratic forms with real coefficients. It is equivalent to the problem of the densest lattice packing of $ n $-
dimensional balls of equal radius (see Geometry of numbers).
Let $ f = f ( x) $,
$ x \in \mathbf R ^ {n} $,
be a positive quadratic form over $ \mathbf R $
of determinant $ d = d ( f ) = \mathop{\rm det} f \neq 0 $;
and let
$$
m ( f ) = \inf _ {\begin{array}{c}
x \in \mathbf Z ^ {n} \\
x \neq 0
\end{array}
}
f ( x) = \min _ {\begin{array}{c}
x \in \mathbf Z ^ {n} \\
{x \neq 0 }
\end{array}
} f ( x)
$$
be its homogeneous arithmetical minimum. The quantity
$$
\gamma _ {n} = \sup
\frac{m ( f ) }{\{ d ( f ) \} ^ {1/n} }
= \max \
\frac{m ( f ) }{\{ d ( f ) \} ^ {1/n} }
,
$$
where the supermum or maximum is over all positive quadratic forms $ f $,
is called the Hermite constant; $ \gamma _ {n} = \{ \gamma ( F _ {n} ) \} ^ {2} $,
where $ F _ {n} ( x) = ( x _ {1} ^ {2} + \dots + x _ {n} ^ {2} ) ^ {1/2} $
is the radial function corresponding to a ball.
Originally, one understood by the Hermite problem the task of finding or estimating $ \gamma _ {n} $(
from above and below). The exact values of $ \gamma _ {n} $
are known only for $ n \leq 8 $(
see [1]). For estimates of $ \gamma _ {n} $,
see [2] or [1].
Subsequently, the term Hermite problem was used for the search for local maxima (boundary or extremal) of $ m ( f ) / \{ d ( f ) \} ^ {1/n} $
in the space of coefficients and their corresponding forms $ f $.
Algorithms are known for finding all classes of boundary forms. In particular, the Voronoi algorithm for perfect forms (see [1], [3], [4]).
The problem was posed by Ch. Hermite in 1850.
References
[1] | P.M. Gruber, C.G. Lekkerkerker, "Geometry of numbers" , North-Holland (1987) (Updated reprint) MR0893813 Zbl 0611.10017 |
[2] | C.A. Rogers, "Packing and covering" , Cambridge Univ. Press (1964) MR0172183 Zbl 0176.51401 |
[3] | B.N. Delone, "The Peterburg school of number theory" , Moscow-Leningrad (1947) (In Russian) |
[4] | E.P. Baranovskii, "Packings, coverings, partitionings and certain other distributions in spaces of constant curvature" Progress in Math. , 9 (1971) pp. 209–253 Itogi Nauk. Mat. Algebra Topol. Geom. 1967 (1969) pp. 189–225 |