pukpr

LTE Parameter Optimization Framework

A multi-stage optimization pipeline for the Laplace's Tidal Equation (LTE) climate index model, using coordinate descent, Pareto front analysis, and cross-index forcing manifold convergence.

Overview

The framework optimizes LTE model parameters across multiple climate indices (AMO, NAO, PDO, NINO4, TNA, Baltic) using a staged approach:

1. IR + ltep tuning — coarse optimization of feedback and modulation frequencies
2. lpap micro-tuning — fine adjustment of tidal amplitudes/phases (sub-0.01 shifts)

pukpr

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from io import StringIO
from pathlib import Path

# Gray et al. 2004 tree-ring reconstruction (1567–1990)
raw = """age	ssta	amo
1567	-1.43849	NA
1568	-0.73836	NA

pukpr

LTE/AMO Fitting Parameters — Objective DOF Characterization

Source: lt.exe.p / lt.exe.resp, AMO forward pass, as unwrapped in lte_forward.py. Each parameter is classified by what it mechanistically claims versus what role it plays in the fit, so that "degrees of freedom" isn't reported as a single undifferentiated count.

1. Externally anchored — not free relative to the AMO target

| Parameter | Role | Basis |

pukpr

#!/usr/bin/env python3
"""lte_forward.py — sequential, non-optimizing equivalent of the combined
Ada (lt.exe / enso_opt) + Python (lte_gui.py) software program.

Rationale
---------
The full system runs an Ada random-descent *optimizer* (N threads, up to
MAXLOOPS restarts) wrapped around a deterministic *forward computation*:
given a parameter set (lt.exe.p) and a climate index (.dat), the pipeline
produces the forced climate response and dumps it to lte_results.csv.

pukpr

#!/usr/bin/env python3
"""Plot hidden latent forcing for the standard LTE result directories."""

from __future__ import annotations

import argparse
from bisect import bisect_right
import csv
import math
from pathlib import Path

pukpr

Substantiating the LTE Draconic-Tidal Model of the QBO: A Plausibility-and-Parsimony Assessment

(Revised: corrected aliasing arithmetic; reframed stratosphere–troposphere/ocean correlation as common-mode forcing rather than QBO-as-causal-intermediary)

TL;DR

• The figures show a physically motivated, cross-validated fit in which a single dominant lunar draconic (nodical, 27.212-day) tidal term, modulated by semi-annual solar impulses, reproduces the QBO with training CC≈0.93 and out-of-sample validation CC≈0.75 — a genuinely strong result that is plausible (the draconic month seasonally aliases to ~2.366 years, matching the QBO's ~2.37-year asymptotic period to within ~1.5 days) and arguably parsimonious (the forcing frequencies are fixed by ephemerides, not fit), but it directly contradicts the mainstream consensus that the QBO period is internally generated by wave–mean-flow interaction and, per the standard review literature, "unrelated to any known periodic forcing."
• The strongest substantia

pukpr

# python3 tidal_chaos.py --k 5 --C 1
# see end of https://geoenergymath.com/2026/06/16/non-homogeneous-non-autonomous/

#!/usr/bin/env python3
"""
Plot M(t) = L(t) + C * sin(k * ∫₀ᵗ L(τ) dτ) for t in [0, 20] years,
where L(t) = 1 + cos(t). The modulation frequency is proportional to L(t),
so higher L gives faster modulation, and L≈0 gives almost no modulation.
Usage: python script.py --k <value> --C <value>
"""

pukpr

import itertools

def frequency(i, j, k):
    return (365.242 * i / 27.3216 + j / 8.848 + k / 18.6) % 1.0

def main():
    indices = range(-4, 5)
    rows = []
    for i, j, k in itertools.product(indices, repeat=3):
        f = frequency(i, j, k)

pukpr

"""
Extract the symmetric "comb" component from LTE data.

The script reads time from column 1 and an amplitude column (default: column 3),
finds mirrored spectral pairs about annualised symmetry centres (0.5, 1.5, 2.5,
...), reconstructs only those paired frequencies, and writes the filtered signal
to lte_filtered.csv.
"""

from __future__ import annotations

pukpr

#!/usr/bin/env python3
import sys
import json
import numpy as np
import matplotlib.pyplot as plt
from pathlib import Path

def read_lpap(subdir):
    """
    Read lpap list from lt.exe.p in the given subdirectory.