lemma
proved
Jcost_unit0
show as:
view math explainer →
open explainer
Generate a durable explainer page for this declaration.
open lean source
IndisputableMonolith.Cost on GitHub at line 12.
browse module
All declarations in this module, on Recognition.
explainer page
depends on
used by
-
pitchCost_at_unison -
hearingLossPenalty_zero -
srCost_zero_at_threshold -
actionJ_const_one -
Jcost_taylor_quadratic -
pleasure_max_at_one -
aestheticCost_zero_at_optimum -
narrativeTension_resolution -
wallpaperJ_p6m_eq_zero -
beautyScore_at_one -
yieldGapCost_at_potential -
J_at_one -
resonant_minimization -
settlementCost_at_fit -
proportionCost_at_ideal -
eccentricity_penalty_zero -
solarWindCost_at_eq -
J_unit_zero -
moon_J_cost_zero -
haberBoschTempCost_at_min -
below_threshold_equilibrium -
physiological_ros -
vacuum_climate_zero_cost -
forecastCost_zero_at_unit -
chainCost_zero_at_unit -
damkohlerCost_at_critical -
abundanceCost_at_predicted -
matter_balance_equilibrium -
vacuum_mode -
flat_minimizes_cost -
homogeneous_minimizes_cost -
potential_min_at_one -
V_vacuum -
jcost_ground_state -
jcost_unbounded_near_zero -
Jcost_eq_zero_iff -
Jmetric_one -
Jlog_eq_zero_iff -
Jcost_agrees_on_exp -
Jcost_unit0
formal source
9 symmetric : ∀ {x}, 0 < x → F x = F x⁻¹
10 unit0 : F 1 = 0
11
12lemma Jcost_unit0 : Jcost 1 = 0 := by
13 simp [Jcost]
14
15/-- J(x) expressed as a squared ratio: J(x) = (x-1)²/(2x). -/
16lemma Jcost_eq_sq {x : ℝ} (hx : x ≠ 0) :
17 Jcost x = (x - 1) ^ 2 / (2 * x) := by
18 unfold Jcost
19 field_simp [hx]
20 ring
21
22lemma Jcost_symm {x : ℝ} (hx : 0 < x) : Jcost x = Jcost x⁻¹ := by
23 have hx0 : x ≠ 0 := ne_of_gt hx
24 rw [Jcost_eq_sq hx0, Jcost_eq_sq (inv_ne_zero hx0)]
25 field_simp [hx0]
26 ring
27
28/-- J(x) ≥ 0 for positive x (AM-GM inequality) -/
29lemma Jcost_nonneg {x : ℝ} (hx : 0 < x) : 0 ≤ Jcost x := by
30 have hx0 : x ≠ 0 := hx.ne'
31 rw [Jcost_eq_sq hx0]
32 positivity
33
34def AgreesOnExp (F : ℝ → ℝ) : Prop := ∀ t : ℝ, F (Real.exp t) = Jcost (Real.exp t)
35
36@[simp] lemma Jcost_exp (t : ℝ) :
37 Jcost (Real.exp t) = ((Real.exp t) + (Real.exp (-t))) / 2 - 1 := by
38 have h : (Real.exp t)⁻¹ = Real.exp (-t) := by
39 symm; simp [Real.exp_neg t]
40 simp [Jcost, h]
41
42class SymmUnit (F : ℝ → ℝ) : Prop where
papers checked against this theorem (showing 23 of 23)
-
Generative models stay exactly on particle phase space manifold
"the 'pure noise' forward process endpoint corresponds to the uniform distribution on phase space"
-
Known operators shrink risk bounds and scale down sample needs
"The estimator decomposes the total risk into a sum over learned layers; every known operator contributes zero to this sum... Replacing learned layer m by a known operator removes the term Am E∥em∥22 from (4)."
-
Density triggers conductor-insulator crossover in ultracold plasmas
"F(r,ε) = A_F exp{−ε/(kBT_vir)} / sqrt(ε − U_eff(r))"
-
Mutual information beats entropy in benchmark selection
"f1(S) = H(X_S) = ½ log det(2πe Σ_SS) ... maximizing log det(Σ_AA), since additive constants do not affect the argmax."
-
Segmentation masking improves sign language AI accuracy
"L = (1/Ω) Σ |I(p) − Î(p)|² ... mean squared error loss based only on masked patches"
-
Two quantum Wasserstein distances coincide on qubits with one cost operator
"DDPT(ϱ,ϱ)² = Σ Iϱ(Hn), where Iϱ(H) = Tr(H²ϱ) − Tr(H√ϱ H√ϱ) is the Wigner–Yanase skew information."
-
Morphing wings cut control cost 65% in obstacle avoidance
"we adopt a physics-based cost model that directly computes actuation effort from aerodynamic loading using the virtual work principle"
-
MARL controller lifts 5G cell throughput with small fairness cost
"a PRB agent learns user-level resource shares ... and a power agent distributes the base-station power budget ... a fairness-aware reward based on smoothed throughput and Jain's fairness index"
-
VLM judge reproduces human video rankings with perfect match
"PHAS(m) = (1/|P|) Σ_p Σ_d w_d s_{m,p,d} / Σ_d w_d × λ(m,p), where w_d are calibrated by non-negative ridge logistic regression on human preference annotations."
-
Flow-matching super-resolves ground astro images without fakes
"W(k) = H*(k) / (|H(k)|² + λ_SNR⁻¹) ... Wiener-regularized approximate adjoint"
-
Run-and-tumble motion keeps chemotaxis robust to noise
"Weber's law / log-sensing regime where v_drift ∼ g independent of c"
-
Supervision preferences and confrontation control rumor spread
"c_i(t) = ξ/(1+e^{-d_{i→r}(t)}) ... ĉ_i(t) = min{c_i(t),1}"
-
DINOv3 segments remote sensing images without RS fine-tuning
"CAT-Seg [8] introduced cost aggregation for OVSS by framing similarity scores between CLIP text and image embeddings as cost maps. ... The cost aggregation refines the raw cost maps into per-class probability maps."
-
Fidelity stays exactly 1 when coupling tuned to -1 in four-qubit chain
"F = |cos(φ/2)| ... C_l1 = sin²(φ/2) ... freezing at α = −1"
-
Reward centering now works for episodic RL
"Φ(s) := b/(1−γ) ... F(s,a,s') = γ·b/(1−γ) − b/(1−γ) = −b"
-
Hybrids make image restoration up to 3.4x faster on edge CPUs
"We minimize the loss L_distill = |O_S − O_T|_2^2 to make MambaIR blocks serve as lightweight surrogates."
-
Encoder models beat larger decoders on noisy medical report QA
"ϕ(u, v) = 1 − d_lev(u, v)/max(|u|,|v|)"
-
BerLU smooths activations with Bernstein polynomials
"BerLU(x) = αx for x<-ε; (1-α)/(4ε) x² + (1+α)/2 x + (1-α)ε/4 for -ε≤x≤ε; x for x>ε"
-
One measurement stabilizes bistable qubit frequency
"the optimal probing time τ_opt ≈ 1/(2 Δ_TLS) ... maximizes sensitivity to the TLS-induced frequency shift"
-
Unanswerable math prompts deviate in LLM geometry before output
"we extract last-layer hidden states, apply mean pooling over all input tokens, and subtract the global mean vector... All distances are cosine distances (1−cos θ)."
-
Mirror descent gives O(1/n) rate for convex optimal control
"Take h(u)=½|u|². Then D_h(v|u)=½|v−u|². Hence the mirror step becomes... Euclidean projected gradient descent is the special case of mirror descent associated with the quadratic mirror map h(u)=½|u|²."
-
Recurrent RL hidden states match optimal control co-states
"Assuming a standard continuous action space and a quadratic control penalty c(u) = u⊤Ru ... yields a closed-form optimal control law u⋆ = -½ R⁻¹ G_θ(y) h⋆."
-
Exact entropy and temperature for any number of equispaced levels
"Ω_p(E,N) = [x^q] ((1−x^p)/(1−x))^N ... the contour-integral form and asymptotic evaluation follow standard methods of coefficient extraction and saddle-point analysis"