Peetre's inequality for the Japanese bracket #
The Japanese bracket ⟨x⟩ = (1 + ‖x‖ ^ 2) ^ (1 / 2) is submultiplicative up to the factor √2:
since ‖x + y‖ ^ 2 ≤ 2 ‖x‖ ^ 2 + 2 ‖y‖ ^ 2,
1 + ‖x + y‖ ^ 2 ≤ 2 (1 + ‖x‖ ^ 2) (1 + ‖y‖ ^ 2).
Raising this to the power s / 2 gives, for s ≥ 0, ⟨x + y⟩ ^ s ≤ 2 ^ (s / 2) ⟨x⟩ ^ s ⟨y⟩ ^ s.
For s < 0 the same bound for -s, applied to the decomposition x = (x + y) + (-y), gives
⟨x + y⟩ ^ s ≤ 2 ^ (-s / 2) ⟨x⟩ ^ s ⟨y⟩ ^ (-s). Both cases are Peetre's inequality
⟨x + y⟩ ^ s ≤ 2 ^ (|s| / 2) ⟨x⟩ ^ s ⟨y⟩ ^ |s|, for every real s.
It says that translating a frequency by y changes the weight ⟨ξ⟩ ^ s of the Bessel-potential
space H^s by at most the factor 2 ^ (|s| / 2) ⟨y⟩ ^ |s|. This is the estimate behind the
boundedness of multiplication by Schwartz functions on H^s for every real s, and behind the
commutator estimates of the pseudodifferential calculus.
The brackets are written out as (1 + ‖x‖ ^ 2) ^ (s / 2), the weight of Mathlib's
TemperedDistribution.besselPotential.
Main declarations #
TauCeti.one_add_norm_add_sq_le:1 + ‖x + y‖ ^ 2 ≤ 2 (1 + ‖x‖ ^ 2) (1 + ‖y‖ ^ 2).TauCeti.one_add_norm_add_sq_rpow_le_mul: Peetre's inequality⟨x + y⟩ ^ s ≤ 2 ^ (|s| / 2) ⟨x⟩ ^ s ⟨y⟩ ^ |s|.TauCeti.one_add_norm_sq_rpow_le_mul_sub: the equivalent form⟨x⟩ ^ s ≤ 2 ^ (|s| / 2) ⟨x - y⟩ ^ |s| ⟨y⟩ ^ s.TauCeti.one_add_norm_sq_rpow_mul_rpow_le_mul_sub_left,TauCeti.one_add_norm_sq_rpow_mul_rpow_le_mul_sub_right: the two-weight forms⟨x⟩ ^ r ⟨y⟩ ^ (-s) ≤ 2 ^ (|s| / 2) ⟨x - y⟩ ^ |s| ⟨x⟩ ^ (r - s)and⟨x⟩ ^ r ⟨y⟩ ^ (-s) ≤ 2 ^ (|r| / 2) ⟨x - y⟩ ^ |r| ⟨y⟩ ^ (r - s), which bound the Fourier kernel of an operator fromH^stoH^rby a function ofx - ytimes a weight of orderr - sin either variable.TauCeti.one_add_norm_sq_rpow_le_of_norm_sub_le_one: for‖x - y‖ ≤ 1,⟨x⟩ ^ s ≤ 2 ^ |s| ⟨y⟩ ^ s.TauCeti.one_add_norm_sq_rpow_le_of_norm_eq_mul: rescaling byc ≥ 1changes the weight⟨x⟩ ^ sby at most the factorc ^ |s|.TauCeti.one_add_norm_smul_sq_rpow_le: the same for the scalar multiplec • x,⟨c • x⟩ ^ s ≤ c ^ |s| ⟨x⟩ ^ s.TauCeti.one_add_norm_add_sq_rpow_le: the subadditive form⟨x + y⟩ ^ s ≤ 2 ^ s (⟨x⟩ ^ s + ⟨y⟩ ^ s)fors ≥ 0, which splits the weight of a convolution between its two factors.TauCeti.one_add_norm_sq_rpow_le_one_add_norm_pow:⟨x⟩ ^ N ≤ (1 + ‖x‖) ^ Nfor naturalN.TauCeti.one_add_norm_sq_rpow_le_one_add_norm_rpow: the bracket is at most the inhomogeneous norm,⟨x⟩ ^ r ≤ (1 + ‖x‖) ^ rforr ≥ 0.TauCeti.one_add_norm_sq_rpow_le_two_pow_mul_one_add_norm_pow: the comparison⟨x⟩ ^ N ≤ 2 ^ N (1 + ‖x‖ ^ N)of a natural power of the bracket with a power of the norm.
References #
- G. B. Folland, Real Analysis: Modern Techniques and Their Applications, 2nd ed., Wiley, 1999,
§9.3 (Peetre's inequality, used for multiplication by Schwartz functions on
H_s).
Peetre's inequality. For every real s, the Japanese bracket ⟨x⟩ = (1 + ‖x‖ ^ 2) ^ (1/2)
satisfies ⟨x + y⟩ ^ s ≤ 2 ^ (|s| / 2) ⟨x⟩ ^ s ⟨y⟩ ^ |s|.
Peetre's inequality, in the form ⟨x⟩ ^ s ≤ 2 ^ (|s| / 2) ⟨x - y⟩ ^ |s| ⟨y⟩ ^ s for every
real s.
Peetre's inequality for a pair of weights, with the decay in x:
⟨x⟩ ^ r ⟨y⟩ ^ (-s) ≤ 2 ^ (|s| / 2) ⟨x - y⟩ ^ |s| ⟨x⟩ ^ (r - s) for all real r and s.
Peetre's inequality for a pair of weights, with the decay in y:
⟨x⟩ ^ r ⟨y⟩ ^ (-s) ≤ 2 ^ (|r| / 2) ⟨x - y⟩ ^ |r| ⟨y⟩ ^ (r - s) for all real r and s.
Rescaling by a factor c ≥ 1 changes the Japanese bracket weight by at most c ^ |s|: if
‖x‖ = c ‖y‖, then (1 + ‖x‖ ^ 2) ^ (s / 2) ≤ c ^ |s| (1 + ‖y‖ ^ 2) ^ (s / 2) for every real s.
For s ≥ 0 this uses ‖x‖ ≤ c ‖y‖, and for s < 0 it uses ‖y‖ ≤ ‖x‖.
Scaling by c ≥ 1 changes the Japanese bracket weight by at most c ^ |s|:
(1 + ‖c • x‖ ^ 2) ^ (s / 2) ≤ c ^ |s| (1 + ‖x‖ ^ 2) ^ (s / 2) for every real s.
Peetre's inequality, subadditive form. The Japanese bracket weight
(1 + ‖x‖ ^ 2) ^ (s / 2) is subadditive up to a constant: for s ≥ 0,
(1 + ‖x + y‖ ^ 2) ^ (s / 2) ≤ 2 ^ s ((1 + ‖x‖ ^ 2) ^ (s / 2) + (1 + ‖y‖ ^ 2) ^ (s / 2)).
A natural power of the Japanese bracket is comparable to the same power of the norm:
(1 + ‖x‖ ^ 2) ^ (N / 2) ≤ 2 ^ N (1 + ‖x‖ ^ N).