Stationarity and detailed balance for finite continuous-time Markov chains
Probability Theory
DefinitionDetailed balance#✓fully proved
The row vector π is in detailed balance with the matrix Q:
π i * Q i j = π j * Q j i for all i j (Norris 1997, §3.7). This is only the algebraic
equation: it holds for π = 0 and every Q. For a probability vector π and a stochastic
matrix it is the reversibility of the chain started from π.
def IsDetailedBalance (π : ι → ℝ) (Q : Matrix ι ι ℝ) : Prop :=
∀ i j, π i * Q i j = π j * Q j i
Used by 10
π is the equilibrium of the SiPhy matrix · Reversible continuous-time models give reversible semigroups · Time-reversibility of the GTR chain · Detailed balance of the F81 model · GTR is reversible · Detailed balance of the HKY85 model · Detailed balance propagates to the transition function · Detailed balance implies invariance for the transition semigroup · Detailed balance implies invariance for the generatorTheoremInvariance under the semigroup#✓fully proved
π Q = 0 if and only if π P t = π for every real
t (Norris 1997, §3.5, finite state space).
theorem vecMul_eq_zero_iff_forall_vecMul_ctmcTransition [DecidableEq ι] (π : ι → ℝ)
(Q : Matrix ι ι ℝ) :
π ᵥ* Q = 0 ↔ ∀ t, π ᵥ* ctmcTransition Q t = π
noncomputable def ctmcTransition (Q : Matrix ι ι ℝ) (t : ℝ) : Matrix ι ι ℝ :=
exp (t • Q)
MathlibExt/Probability/Process/ContinuousTimeMarkovChain/Semigroup.lean:64
Is Detailed Balance
TheoremDetailed balance implies invariance for the generator#✓fully proved
Detailed balance with a matrix whose rows sum to 0 implies invariance π Q = 0
(Norris 1997, §3.7).
theorem vecMul_eq_zero (h : IsDetailedBalance π Q) (hQ : ∀ i, ∑ j, Q i j = 0) :
π ᵥ* Q = 0
def IsDetailedBalance (π : ι → ℝ) (Q : Matrix ι ι ℝ) : Prop :=
∀ i j, π i * Q i j = π j * Q j i
MathlibExt/Probability/Process/ContinuousTimeMarkovChain/Stationary.lean:64
Uses 1 · Used by 2
TheoremDetailed balance propagates to the transition function#✓fully proved
Detailed balance for Q
implies π i P t i j = π j P t j i for every t (Norris 1997, §3.7). For a probability
vector π and a rate matrix Q this is the reversibility of the chain started from π.
theorem ctmcTransition [DecidableEq ι] (h : IsDetailedBalance π Q) (t : ℝ) :
IsDetailedBalance π (ctmcTransition Q t)
def IsDetailedBalance (π : ι → ℝ) (Q : Matrix ι ι ℝ) : Prop :=
∀ i j, π i * Q i j = π j * Q j i
MathlibExt/Probability/Process/ContinuousTimeMarkovChain/Stationary.lean:64
noncomputable def ctmcTransition (Q : Matrix ι ι ℝ) (t : ℝ) : Matrix ι ι ℝ :=
exp (t • Q)
MathlibExt/Probability/Process/ContinuousTimeMarkovChain/Semigroup.lean:64
Uses 2 · Used by 2
TheoremDetailed balance implies invariance for the transition semigroup#✓fully proved
Detailed balance with a matrix whose rows sum to 0 (for instance a rate matrix) makes π
invariant for every P t.
theorem vecMul_ctmcTransition [DecidableEq ι] (h : IsDetailedBalance π Q)
(hQ : ∀ i, ∑ j, Q i j = 0) (t : ℝ) :
π ᵥ* ProbabilityTheory.ctmcTransition Q t = π
def IsDetailedBalance (π : ι → ℝ) (Q : Matrix ι ι ℝ) : Prop :=
∀ i j, π i * Q i j = π j * Q j i
MathlibExt/Probability/Process/ContinuousTimeMarkovChain/Stationary.lean:64
noncomputable def ctmcTransition (Q : Matrix ι ι ℝ) (t : ℝ) : Matrix ι ι ℝ :=
exp (t • Q)
MathlibExt/Probability/Process/ContinuousTimeMarkovChain/Semigroup.lean:64
Uses 4 · Used by 1
Loading dependency explorer…
The dependency explorer needs JavaScript and its graph data.