Generated history
A candidate sequence before survival filtering.
Reference
A companion reference for RSG v1.4 and the visualisation. The formulas are written as copyable text rather than images, with claim status kept visible.
(X_n, \phi_n, \mu_n, S_n).\sigma_n = (X_n, \phi_n, \mu_n, S_n)
\sigma_{n+1} = R(\sigma_n)
\gamma_i = (\sigma_{i,0}, \sigma_{i,1}, \sigma_{i,2}, ...)
\phi_n = (\Theta_n, \Pi_n)
J_n = \Theta_n^2 + \ell^2 \Pi_n^2
W_n = \Theta_n^2 / J_n
\Lambda_surv(\sigma,\phi) = \Gamma(\sigma) W(\phi)
A_i(t) = integral_0^t \Gamma_i(\tau) W_i(\tau) d\tau
S_i(t) = exp(-A_i(t))
G_i(t) := S_i(t) = exp(-A_i(t))
p_i(t) = q_i G_i(t) / sum_j q_j G_j(t)
equal-preparation shorthand:
p_i(t) = S_i(t) / sum_j S_j(t)
H_surv(t) = -sum_i p_i(t) ln p_i(t)
N_eff(t) = exp(H_surv(t))
ln(p_i / p_j) = ln(q_i / q_j) - A_i + A_j
equal-preparation shorthand:
ln(p_i / p_j) = -A_i + A_j
\hat p_i = (I_i - I_{i,bg}) / sum_j (I_j - I_{j,bg})
\hat L_ij = ln(\hat p_i / \hat p_j)
A candidate sequence before survival filtering.
The normalised measure over histories that remain live after filtering.
The multiplier G_i = S_i = exp(-A_i) applied to a prepared represented amount before normalisation.
The portion of a projected state that is vulnerable to the loss channel.
Recursive or topological persistence that lets a history become record-like.
The rule that only the controlled analogue-media prediction is the developed empirical claim in RSG v1.4.
The requirement that \Theta, \Pi, \ell, phase zero, wrapping, and singular-state rules are fixed before output data.
The comparison model A_i = integral Gamma dt; RSG-specific support requires improvement over it in a positive-control regime.
Effectively lossless, projected-norm-preserving transport with no recurrent internal clock and therefore d tau_R = 0.
A conditional future application requiring its own variables, calibration, and test.