While this submission is a draft, it cannot be used by other submissions.

Deterministic quasipolynomial-time approximation schemes

Lax263307.QPTAS · concepts/Lax263307/QPTAS.lean · lax-263307

definition

Loading review…

Sign in with ORCID

Community review

Flags

Each flag is tied to a public ORCID identity and explains why this concept may be incorrect.

No flags have been submitted.

    Community review

    Flag this concept

    State precisely what appears incorrect. This explanation will be public under your ORCID name.

    No source line selected.

    Natural Language Statement

    Definition

    A deterministic QPTAS for maximum-weight independent set on a class C\mathcal C uses one finite program, depending only on C\mathcal C, which returns a (1−ε)(1-\varepsilon)-approximation on every member with rational weights and rational 0<ε<10<\varepsilon<1. Its running time has the form

    2(ε−1log⁡(n+2))OC(1)+(ℓ+1)OC(1),2^{(\varepsilon^{-1}\log(n+2))^{O_{\mathcal C}(1)}} +(\ell+1)^{O_{\mathcal C}(1)},

    where ℓ\ell is the binary length of the weights and accuracy. Neither class recognition nor a separator is supplied as input.

    Concept map
    5 concepts; 1 descendant hidden
    100%
    Proven claimDefinitionThis conceptRelated conceptA → B: B builds on A

    Lean source view on GitHub

    1import Lax263307.GraphClasses
    2import Lax263307.WeightedIndependentSet
    3import Lax808846.Ram
    4import Mathlib.Data.Nat.Log
    5
    6/-!
    7---
    8title: Deterministic quasipolynomial-time approximation schemes
    9type: definition
    10---
    11A deterministic QPTAS for maximum-weight independent set on a class
    12C\mathcal C uses one finite program, depending only on C\mathcal C,
    13which returns a (1−ε)(1-\varepsilon)-approximation on every member with
    14rational weights and rational 0<ε<10<\varepsilon<1.
    15Its running time has the form
    162(ε−1log⁡(n+2))OC(1)+(ℓ+1)OC(1),2^{(\varepsilon^{-1}\log(n+2))^{O_{\mathcal C}(1)}} +(\ell+1)^{O_{\mathcal C}(1)},
    17
    18where ℓ\ell is the binary length of the weights and accuracy.
    19Neither class recognition nor a separator is supplied as input.
    20
    21# Formalization notes
    22
    23Execution is `Lax808846.Ram.RunsTo` on the registered word RAM.
    24The supplied input is exactly `encodeInput`, with no implicit length prefix.
    25The constants and program precede the input and word width in the quantifiers.
    26The natural-number time bound uses ⌈1/ε⌉\lceil1/\varepsilon\rceil and
    271+⌊log⁡2(n+2)⌋1+\lfloor\log_2(n+2)\rfloor to cover small inputs.
    28
    29The same bound is a sufficient word width, and the program must work at
    30every larger width. In particular it works at that explicitly prescribed
    31width; there is no advice hidden in a choice of width. Allowing this many
    32bits accommodates exact rational subroutines. Simulating these bounded-word
    33operations with bit operations costs a polynomial in the time bound,
    34which remains of the displayed form after increasing its constants and
    35exponent: (A+B)k≤2k−1(Ak+Bk)(A+B)^k\le2^{k-1}(A^k+B^k). Thus the additive polynomial
    36dependence on ℓ\ell is preserved. Correctness and time are promised only
    37on well-formed inputs whose graphs belong to C\mathcal C.
    38-/
    39
    40namespace Lax263307.QPTAS
    41
    42open GraphClasses WeightedIndependentSet Lax808846.Ram
    43
    44/-- An explicit bound representing the paper's two-parameter running time. -/
    45def timeBound (c d n ℓ : ℕ) (ε : ℚ) : ℕ :=
    46 2 ^ (c * (⌈ε⁻¹⌉₊ * (Nat.log2 (n + 2) + 1)) ^ d) + c * (ℓ + 1) ^ d
    47
    48/-- One uniform deterministic word-RAM program realizes the approximation scheme. -/
    49def HasQPTAS (C : GraphClass) : Prop :=
    50 ∃ (program : Program) (c d : ℕ), 0 < c ∧ 0 < d ∧
    51 ∀ (n : ℕ) (G : SimpleGraph (Fin n)) [DecidableRel G.Adj],
    52 G ∈ C n → ∀ (weights : Fin n → ℚ) (ε : ℚ), 0 < ε → ε < 1 →
    53 let T := timeBound c d n (rationalBitSize weights ε) ε
    54 ∀ wordLength : ℕ, T ≤ wordLength →
    55 ∃ I : Finset (Fin n), IsApproximation G weights ε I ∧
    56 ∃ t ≤ T, RunsTo wordLength program
    57 (encodeInput G weights ε) (encodeOutput I) t
    58
    59end Lax263307.QPTAS
    60
    Formalization notes

    Execution is Lax808846.Ram.RunsToLax808846.Ram.RunsTo on the registered word RAM. The supplied input is exactly encodeInputencodeInput, with no implicit length prefix. The constants and program precede the input and word width in the quantifiers. The natural-number time bound uses ⌈1/ε⌉\lceil1/\varepsilon\rceil and 1+⌊log⁡2(n+2)⌋1+\lfloor\log_2(n+2)\rfloor to cover small inputs.

    The same bound is a sufficient word width, and the program must work at every larger width. In particular it works at that explicitly prescribed width; there is no advice hidden in a choice of width. Allowing this many bits accommodates exact rational subroutines. Simulating these bounded-word operations with bit operations costs a polynomial in the time bound, which remains of the displayed form after increasing its constants and exponent: (A+B)k≤2k−1(Ak+Bk)(A+B)^k\le2^{k-1}(A^k+B^k). Thus the additive polynomial dependence on ℓ\ell is preserved. Correctness and time are promised only on well-formed inputs whose graphs belong to C\mathcal C.

    Discussion

    Ask a question or add context. Endorsements and structured flags are kept in the review panel above.

    Loading discussion…