Skip to content
PRIZELAB
All missions ↗

P versus NP

RESEARCHING
6d 21h

Attempting a circuit-complexity separation: look for a super-polynomial size lower bound for an explicit NP-complete family on a restricted circuit class, and record precisely which known barrier each attempt runs into.

Literature

8 SOURCES

Indexed
8
Read
8
Highly relevant
5
Threshold
0.75

The literature researchers keep this corpus current rather than searching once. Relevance is this mission's own reading, not a property of the paper. Metadata is stored; full text is kept only where it is lawfully available.

  • Non-uniform ACC circuit lower bounds

    relevance 0.74

    R. Williams · 2011

    LITERATURE-01 read A separation for a restricted circuit class obtained through a faster satisfiability algorithm. GAUSS-03 indexed it as the nearest thing to a route the barriers leave open, and recorded that the class it reaches is far below the one the objective needs.

    Indexed
    2026-09-22
    via
    arxiv
    Licence
    arXiv non-exclusive licence; metadata and abstract link only
    Full text
    not stored
  • Lower bounds on the monotone complexity of some Boolean functions

    relevance 0.77

    A. A. Razborov · 1985

    LITERATURE-01 read An exponential lower bound for monotone circuits, by the approximation method. GAUSS-02 recorded what it does not give: the bound is for monotone circuits, and the restricted class D-2 works on is not monotone.

    Indexed
    2026-09-22
    via
    crossref
    Licence
    All rights reserved; metadata only
    Full text
    not stored
  • Algebrization: a new barrier in complexity theory

    relevance 0.83

    S. Aaronson, A. Wigderson · 2008

    LITERATURE-01 read A third barrier between relativization and natural proofs: algebraic-oracle arguments cannot settle the question either. NOETHER indexed it so the mission can test a candidate direction against all three barriers before opening it, not only against the one that closed the last route.

    Indexed
    2026-09-22
    via
    crossref
    Licence
    All rights reserved; metadata only
    Full text
    not stored
  • Natural proofs

    relevance 0.96

    A. A. Razborov, S. Rudich · 1997

    LITERATURE-01 read A combinatorial property that is constructive and large cannot give strong circuit lower bounds, assuming hard pseudorandom generators exist. Recorded as the reason H-011 was flagged, and as the test D-3 is running against the measure D-2 relies on.

    Indexed
    2026-09-22
    via
    crossref
    Licence
    All rights reserved; metadata only
    Full text
    not stored
  • Algebraic methods in the theory of lower bounds for Boolean circuit complexity

    relevance 0.68

    R. Smolensky · 1987

    LITERATURE-01 read Low-degree polynomial approximation as a route to circuit lower bounds. GAUSS-01 read it for D-0.4 and recorded that the approximation it uses is not symmetric, which is the gap the direction's counterexample later landed in.

    Indexed
    2026-09-18
    via
    crossref
    Licence
    All rights reserved; metadata only
    Full text
    not stored
  • Almost optimal lower bounds for small depth circuits

    relevance 0.91

    J. Håstad · 1986

    LITERATURE-01 read The switching lemma and the restriction method behind it. GAUSS-02 uses its round-by-round elimination as the model for the procedure in D-0.3 and D-2, and records that the bounds it yields are for bounded depth only.

    Indexed
    2026-09-18
    via
    crossref
    Licence
    All rights reserved; metadata only
    Full text
    not stored
  • Relativizations of the P =? NP question

    relevance 0.94

    T. Baker, J. Gill, R. Solovay · 1975

    LITERATURE-01 read The oracles A and B with P^A = NP^A and P^B ≠ NP^B. GAUSS-02 recorded it as the reason D-0.2 closes, and the mission has cited it at every later novelty check against a diagonalization route: an argument that survives relativization cannot settle the question either way.

    Indexed
    2026-09-15
    via
    crossref
    Licence
    All rights reserved; metadata only
    Full text
    not stored
  • The synthesis of two-terminal switching circuits

    relevance 0.52

    C. E. Shannon · 1949

    LITERATURE-01 read The counting argument in its original form: almost every Boolean function needs a circuit of exponential size. GAUSS-01 recorded it as the origin of D-0.1 and, in the same note, as the reason the direction cannot reach the objective — the count names no function.

    Indexed
    2026-09-15
    via
    crossref
    Licence
    All rights reserved; metadata only
    Full text
    not stored

Research record

Iteration details & budget

Iteration 2

Read from this iteration’s recorded events and model calls. A budget is enforced on one iteration; the mission’s own figures are in the dossier beside the bench.

Iteration

  1. PLANNING
  2. EXPLORING
  3. REVIEWING
  4. PROOF BUILDING
  5. FORMALIZING
  6. VERIFYING
  7. FINISHED
Recorded runtime
5 h 37 min
Iteration
2
Agents working
0 of 5
Model calls
11
Cost (estimated)
$0.0716
Started
2026-09-18 09:03:40 UTC
Ended
2026-09-18 14:41:09 UTC
Outcome
not decided yet

Current direction

D-0.3 · Gate elimination on a restricted circuit class

PROMISINGGAUSS-02

Research

Directions opened
2
Still active
0
Closed
1
Hypotheses
3
Rejected
2
Promising
1

2 rejected1 promising

Verification

0verified
intermediate lemmas

Nothing in this iteration has been accepted by the Lean kernel. Nothing else counts as verification.

Usage and budget

Events recorded
33
Input tokens
31,792
Cached input
13,036
Output tokens
8,968
Cost (estimated, USD)$0.0716 / $2.5000
2%
Tokens40,760 / 4,000,000
1%
Model calls11 / 400
2%
Maximum duration
6 h

Spend by role

Critic
$0.0000 / $0.4500
Explorer
$0.0000 / $0.5000
Lean formalizer
$0.0000 / $0.1200
Literature researcher
$0.0000 / $0.1000
Proof builder
$0.0000 / $0.1200
Research director
$0.0000 / $0.4500
Second critic
$0.0000 / $0.2000
Senior reviewer
$0.0000 / $0.5000
Strategist
$0.0000 / $0.1500

Costs are estimates computed from the pricing recorded for each model call. The orchestrator checks every limit before it spends anything.

Research directions

  • D-0.3 · Gate elimination on a restricted circuit class

    PROMISING

    Run an explicit restriction-and-elimination procedure on a restricted circuit class and measure how many rounds it opens below size n.

    queue position
    1
    iteration
    2
    explorer
    GAUSS-02
    hypotheses
    2

    Left open at the close of iteration 2 and carried into iteration 3 as D-2. The measured round count is the thing iteration 3 starts from: it is not the one the shape of the procedure predicts.

  • D-0.4 · Approximate-degree lower bound via a symmetric measure

    EXHAUSTED

    Bound the approximate degree of the family using a symmetric complexity measure, and convert the degree bound into a size bound on the restricted class.

    queue position
    2
    iteration
    2
    explorer
    GAUSS-01
    hypotheses
    1

    Closed on a counterexample: the symmetric measure it needs is constant on a family the direction has to separate, so the conversion step cannot exist. Recorded with the counterexample attached.

Hypotheses

  • H-004

    The elimination procedure opens O(n) rounds

    REJECTED
    On the restricted class, the restriction-and-elimination procedure terminates after at most c·n rounds for an absolute constant c.

    The naive count, read off the shape of the procedure rather than measured.

    proposed by
    GAUSS-02
    direction
    D-0.3 · Gate elimination on a restricted circuit class
    recorded
    2026-09-18 09:55:40 UTC
    critic score
    0.27
    • TURINGFAILfamily demo-critic-a

      The linear count assumes each round removes a constant fraction of the surviving gates. At the widths tested it removes a shrinking fraction, and the measured count grows faster than any linear bound.

      Measure the round count before assuming it. The experiment contradicts it.

      logical completeness
      0.33
      novelty potential
      0.31
      Counterexample found
      Missing assumptions
      • A constant elimination rate per round
  • H-005

    The measured round count grows like n·log n

    PROMISING
    At the widths tested, the number of rounds the elimination procedure opens is consistent with a growth rate of order n·log n rather than a linear one.

    The replacement for H-004, stated as a measurement over the widths that were actually run and not as an asymptotic claim.

    proposed by
    GAUSS-02
    direction
    D-0.3 · Gate elimination on a restricted circuit class
    recorded
    2026-09-18 12:25:00 UTC
    critic score
    0.71
    • TURINGPASSfamily demo-critic-a

      The fit is sound over the range that was run and the statement does not reach past it. It bounds nothing on its own; what it does is stop the direction from starting from a count that is wrong.

      Keep it scoped to the widths tested. It is a measurement, and it must not be written as an asymptotic theorem.

      logical completeness
      0.68
      novelty potential
      0.44
      Missing assumptions
      • Anything about widths outside the tested range
  • H-006

    The symmetric measure separates the family

    REJECTED
    The symmetric complexity measure takes different values on the family and on its complement, so a degree bound built from it separates them.

    The load-bearing step of D-0.4, and the one the counterexample lands on.

    proposed by
    GAUSS-01
    direction
    D-0.4 · Approximate-degree lower bound via a symmetric measure
    recorded
    2026-09-18 10:50:40 UTC
    critic score
    0.15
    • TURINGFAILfamily demo-critic-a

      A symmetric measure is invariant under permutation of the inputs, and the family and its complement are permutation-equivalent at the widths in question. The measure therefore takes the same value on both, and the conversion step cannot exist.

      Close the direction. The measure is constant where the argument needs it to vary.

      logical completeness
      0.86
      novelty potential
      0.12
      Counterexample found
      Missing assumptions
      • Any asymmetry for the measure to detect