"""Exact phase architecture for A(a,b)=[[0,a],[-b,0]], positive rectangles. All numerical root solutions and quadratures are non-certified floats. """ from __future__ import annotations from dataclasses import dataclass import math import numpy as np from scipy.integrate import quad from scipy.optimize import brentq @dataclass(frozen=True) class Rectangle: amin:float amax:float bmin:float bmax:float def __post_init__(self): if not all(map(math.isfinite,(self.amin,self.amax,self.bmin,self.bmax))): raise ValueError('Finite bounds required.') if not (0np.ndarray: if min(a,b)<=0 or t<0: raise ValueError('a,b positive and t nonnegative required.') w=math.sqrt(a*b) c,s=math.cos(w*t),math.sin(w*t) return np.array([[c,a*s/w],[-b*s/w,c]]) def rectangle_cell(box:Rectangle,eta:float)->RectangleCell: am,ap,bm,bp=box.amin,box.amax,box.bmin,box.bmax if eta>0: cuts=[0,math.atan2(bp,eta),math.pi/2,math.pi-math.atan2(eta,ap),math.pi] states=[(am,bp),(ap,bp),(ap,bm),(ap,bp)] elif eta<0: e=-eta cuts=[0,math.atan2(e,am),math.pi/2,math.pi-math.atan2(bm,e),math.pi] states=[(am,bm),(am,bp),(am,bm),(ap,bm)] else: cuts=[0,math.pi/2,math.pi] states=[(am,bp),(ap,bm)] arcs=[]; T=0.; G=0. for x,y,(a,b) in zip(cuts[:-1],cuts[1:],states): h=quad(lambda t:1/phase_speed(t,a,b),x,y,epsabs=1e-12,epsrel=1e-12)[0] G+=0.5*math.log(phase_speed(x,a,b)/phase_speed(y,a,b)) T+=h arcs.append((x,y,a,b,h)) return RectangleCell(eta,T,G,tuple(arcs)) def rectangle_fixed(box:Rectangle,S:float): if not math.isfinite(S) or S<=0: raise ValueError('S must be positive and finite.') slow=math.sqrt(box.amin*box.bmin);fast=math.sqrt(box.amax*box.bmax) candidates=[] for n in range(1,math.floor(fast*S/math.pi)+1): tau=S/n if not math.pi/fast < tau < math.pi/slow:continue K=1. while rectangle_cell(box,-K).period < tau or rectangle_cell(box,K).period > tau: K*=2 if K>1e12:raise ArithmeticError('Endpoint requires higher precision.') eta=brentq(lambda e:rectangle_cell(box,e).period-tau,-K,K,xtol=1e-12) cell=rectangle_cell(box,eta) candidates.append((n,cell)) return max((n*c.log_gain for n,c in candidates),default=0.),candidates def rectangle_free(box:Rectangle)->RectangleCell: hi=1. fun=lambda eta:rectangle_cell(box,eta).log_gain-eta*rectangle_cell(box,eta).period while fun(hi)>0:hi*=2 root=brentq(fun,0,hi,xtol=1e-13) return rectangle_cell(box,root) def cell_monodromy(cell:RectangleCell): P=np.eye(2) for _,_,a,b,t in cell.arcs:P=generator_transfer(a,b,t)@P return P