Strong and plain hardness
Lax434175.StrongHardness · concepts/Lax434175/StrongHardness.lean · lax-434175
No public endorsements yet.
Loading review…
Sign in with ORCIDNatural Language Statement
Theorem
A problem is strongly NP-hard, W[t]-hard or para-NP-hard when it is hard already on the words whose numbers are at most a fixed polynomial in the length of the word, as when the numbers are written in unary. Plain hardness asks the same of all words, the numbers written in binary. The words with small numbers form a subproblem, and the identity reduces it to the whole problem.
Theorem. Strong NP-hardness implies NP-hardness, strong W[t]-hardness implies W[t]-hardness, and strong para-NP-hardness implies para-NP-hardness, for every parameter.
The algorithmic counterpart of strong hardness: a problem is pseudo-polynomially fixed-parameter tractable in when, for every polynomial bound on the numbers, it is fixed-parameter tractable in on the words whose numbers obey the bound. A problem that is W[t]-hard and pseudo-polynomially fixed-parameter tractable is weakly W[t]-hard: its hardness needs large numbers.
Theorem. If a problem is strongly W[t]-hard and pseudo-polynomially fixed-parameter tractable for the same parameter, then W[t] ⊆ FPT.
Concept map
Evidence
Lean source view on GitHub
| 1 | import Lax434175.Results |
| 2 | |
| 3 | /-! |
| 4 | --- |
| 5 | title: Strong and plain hardness |
| 6 | type: theorem |
| 7 | --- |
| 8 | A problem is *strongly* NP-hard, W[t]-hard or para-NP-hard when it is hard already on |
| 9 | the words whose numbers are at most a fixed polynomial in the length of the word, as |
| 10 | when the numbers are written in unary. Plain hardness asks the same of all words, |
| 11 | the numbers written in binary. The words with small numbers form a subproblem, and |
| 12 | the identity reduces it to the whole problem. |
| 13 | |
| 14 | **Theorem.** Strong NP-hardness implies NP-hardness, strong W[t]-hardness implies |
| 15 | W[t]-hardness, and strong para-NP-hardness implies para-NP-hardness, for every |
| 16 | parameter. |
| 17 | |
| 18 | The algorithmic counterpart of strong hardness: a problem is *pseudo-polynomially |
| 19 | fixed-parameter tractable* in when, for every polynomial bound on the |
| 20 | numbers, it is fixed-parameter tractable in on the words whose numbers obey |
| 21 | the bound. A problem that is W[t]-hard and pseudo-polynomially fixed-parameter |
| 22 | tractable is *weakly* W[t]-hard: its hardness needs large numbers. |
| 23 | |
| 24 | **Theorem.** If a problem is strongly W[t]-hard and pseudo-polynomially |
| 25 | fixed-parameter tractable for the same parameter, then W[t] ⊆ FPT. |
| 26 | -/ |
| 27 | |
| 28 | namespace Lax434175.StrongHardness |
| 29 | open Lax434175.Problems Lax434175.Results |
| 30 | open Lax496464.WH_A2_FptReductions Lax496464.WH_B4_Hierarchies |
| 31 | |
| 32 | variable {P : Problem} (R : PReading P) |
| 33 | |
| 34 | /-- Fixed-parameter tractable in `κ` on the words with numbers at most any fixed |
| 35 | polynomial in their length. -/ |
| 36 | def PseudoFPT (κ : TParameter P) : Prop := |
| 37 | ∀ c d, restrict (parameterized R κ) (Small c d) ∈ FPT |
| 38 | |
| 39 | /-- W[t]-hard, and pseudo-polynomially fixed-parameter tractable for the same |
| 40 | parameter. -/ |
| 41 | def WeaklyWHard (t : ℕ) (κ : TParameter P) : Prop := WHard R t κ ∧ PseudoFPT R κ |
| 42 | |
| 43 | variable {R} |
| 44 | |
| 45 | axiom npHard_of_strongly : StronglyNPHard R → NPHard R |
| 46 | |
| 47 | axiom wHard_of_strongly {t : ℕ} {κ : TParameter P} : StronglyWHard R t κ → WHard R t κ |
| 48 | |
| 49 | axiom paraNPHard_of_strongly {κ : TParameter P} : StronglyParaNPHard R κ → ParaNPHard R κ |
| 50 | |
| 51 | axiom exclusive_weak {t : ℕ} {κ : TParameter P} : |
| 52 | StronglyWHard R t κ → PseudoFPT R κ → W t ⊆ FPT |
| 53 | |
| 54 | end Lax434175.StrongHardness |
| 55 |
Discussion
Ask a question or add context. Endorsements and structured flags are kept in the review panel above.
0 comments