Negative type inequalities arise in
the study of embedding properties of metric spaces, but
they often reduce to intractable combinatorial problems. In this paper we study more quantitative versions of
these inequalities involving
the so-called
p -negative type gap. In particular, we focus our attention on
the class of finite ultrametric spaces which are important in areas such as phylogenetics and data mining. Let
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022247X16305856&_mathId=si1.gif&_user=111111111&_pii=S0022247X16305856&_rdoc=1&_issn=0022247X&md5=2c619669fd7d5cb26a068936bb1f487c" title="Click to view the MathML source">(X,d)class="mathContainer hidden">class="mathCode"> be a given finite ultrametric space with minimum non-zero distance
α. Then
the p -negative type gap
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022247X16305856&_mathId=si2.gif&_user=111111111&_pii=S0022247X16305856&_rdoc=1&_issn=0022247X&md5=a89a72507b75dac00e2e758d997e59ea" title="Click to view the MathML source">ΓX(p)class="mathContainer hidden">class="mathCode"> of
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022247X16305856&_mathId=si1.gif&_user=111111111&_pii=S0022247X16305856&_rdoc=1&_issn=0022247X&md5=2c619669fd7d5cb26a068936bb1f487c" title="Click to view the MathML source">(X,d)class="mathContainer hidden">class="mathCode"> is positive for all
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022247X16305856&_mathId=si3.gif&_user=111111111&_pii=S0022247X16305856&_rdoc=1&_issn=0022247X&md5=f930879dd72285404df7e00e42badcc0" title="Click to view the MathML source">p≥0class="mathContainer hidden">class="mathCode">. In this paper we compute
the value of
the limit
class="formula" id="fm0010">
It turns out that this value is positive and it may be given explicitly by an elegant combinatorial formula. This formula allows us to characterize when
the ratio
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022247X16305856&_mathId=si5.gif&_user=111111111&_pii=S0022247X16305856&_rdoc=1&_issn=0022247X&md5=e4808912dc368c5cfb5c361ec1205a75" title="Click to view the MathML source">ΓX(p)/αpclass="mathContainer hidden">class="mathCode"> is a constant independent of
p . The determination of
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022247X16305856&_mathId=si321.gif&_user=111111111&_pii=S0022247X16305856&_rdoc=1&_issn=0022247X&md5=0e9ef2006b2509bac73a4ca6738f064e" title="Click to view the MathML source">ΓX(∞)class="mathContainer hidden">class="mathCode"> also leads to new, asymptotically sharp, families of enhanced
p -negative type inequalities for
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022247X16305856&_mathId=si1.gif&_user=111111111&_pii=S0022247X16305856&_rdoc=1&_issn=0022247X&md5=2c619669fd7d5cb26a068936bb1f487c" title="Click to view the MathML source">(X,d)class="mathContainer hidden">class="mathCode">. Indeed, suppose that
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022247X16305856&_mathId=si66.gif&_user=111111111&_pii=S0022247X16305856&_rdoc=1&_issn=0022247X&md5=35b8bc05e92e30d66e0be2cbe3d617fc" title="Click to view the MathML source">G∈(0,ΓX(∞))class="mathContainer hidden">class="mathCode">. Then, for all sufficiently large
p,
the inequality
class="formula" id="fm0020">
holds for each finite subset
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022247X16305856&_mathId=si68.gif&_user=111111111&_pii=S0022247X16305856&_rdoc=1&_issn=0022247X&md5=324e2c53a3652686fa30fcc6056a8f21" title="Click to view the MathML source">{z1,…,zn}⊆Xclass="mathContainer hidden">class="mathCode">, and each scalar
n -tuple
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022247X16305856&_mathId=si10.gif&_user=111111111&_pii=S0022247X16305856&_rdoc=1&_issn=0022247X&md5=7f1eb2b2fdb15313d6129848871ba4b5" title="Click to view the MathML source">ζ=(ζ1,…,ζn)∈Rnclass="mathContainer hidden">class="mathCode"> that satisfies
class="mathmlsrc">class="formulatext stixSupport mathImg" data-mathURL="/science?_ob=MathURL&_method=retrieve&_eid=1-s2.0-S0022247X16305856&_mathId=si11.gif&_user=111111111&_pii=S0022247X16305856&_rdoc=1&_issn=0022247X&md5=1c195f8e4e472e777093e796833384d2" title="Click to view the MathML source">ζ1+⋯+ζn=0class="mathContainer hidden">class="mathCode">. Notably,
these results do not extend to general finite metric spaces.