= Sage Interactions - Geometry =
goto [[interact|interact main page]]

<<TableOfContents>>

== Intersecting tetrahedral reflections FIXME ==
by Marshall Hampton.  Inspired by a question from Hans Schepker of Glass Geometry.
{{{#!sagecell
#Pairs of tetrahedra, one the reflection of the other in the internal face, are joined by union operations:
p1 = Polyhedron(vertices = [[1,1,1],[1,1,0],[0,1,1],[1,0,1]])
p2 = Polyhedron(vertices = [[1/3,1/3,1/3],[1,1,0],[0,1,1],[1,0,1]])
p12 = p1.convex_hull(p2)
p3 = Polyhedron(vertices = [[0,0,1],[0,0,0],[0,1,1],[1,0,1]])
p4 = Polyhedron(vertices = [[2/3,2/3,1/3],[0,0,0],[0,1,1],[1,0,1]])
p34 = p3.convex_hull(p4)
p5 = Polyhedron(vertices = [[1,0,0],[1,0,1],[0,0,0],[1,1,0]])
p6 = Polyhedron(vertices = [[1/3,2/3,2/3],[1,0,1],[0,0,0],[1,1,0]])
p56 = p5.convex_hull(p6)
p7 = Polyhedron(vertices = [[0,1,0],[0,0,0],[1,1,0],[0,1,1]])
p8 = Polyhedron(vertices = [[2/3,1/3,2/3],[0,0,0],[1,1,0],[0,1,1]])
p78 = p7.convex_hull(p8)
pti = p12.intersection(p34).intersection(p56).intersection(p78)
@interact
def tetra_plot(opac = slider(srange(0,1.0,.25), default = .25)):
    p12r = p12.render_wireframe()+p12.render_solid(opacity = opac)
    p34r = p34.render_wireframe()+p34.render_solid(rgbcolor = (0,0,1),opacity = opac)
    p56r = p56.render_wireframe()+p56.render_solid(rgbcolor = (0,1,0),opacity = opac)
    p78r = p78.render_wireframe()+p78.render_solid(rgbcolor = (0,1,1),opacity = opac)
    ptir = pti.render_wireframe()+pti.render_solid(rgbcolor = (1,0,1),opacity = .9)
    show(p12r+p34r+p56r+p78r+ptir, frame = False)
}}}
{{attachment:tetrareflect.png}}

== Evolutes ==
by Pablo Angulo. Computes the evolute of a plane curve given in parametric coordinates. The curve must be parametrized from the interval [0,2pi].
{{{#!sagecell
var('t');
def norma(v):
    return sqrt(sum(x^2 for x in v))    
paso_angulo=5

@interact
def _( gamma1=input_box(default=sin(t)), gamma2=input_box(default=1.3*cos(t)),
    draw_normal_lines=True, 
    rango_angulos=range_slider(0,360,paso_angulo,(0,90),label='Draw lines for these angles'), 
    draw_osculating_circle=True, 
    t0=input_box(default=pi/3,label='parameter value for the osculating circle'), 
    auto_update=False ):
        
    gamma=(gamma1,gamma2)
    gammap=(gamma[0].derivative(),gamma[1].derivative())
    np=norma(gammap)
    gammapp=(gammap[0].derivative(),gammap[1].derivative())
    npp=norma(gammapp)

    normal=(gammap[1]/np, -gammap[0]/np)    
    curvatura=(gammap[1]*gammapp[0]-gammap[0]*gammapp[1])/norma(gammap)^3
    radio=1/curvatura
        
    centros=(gamma[0]+radio*normal[0],gamma[1]+radio*normal[1])
            
    curva=parametric_plot(gamma,(t,0,2*pi))
    evoluta=parametric_plot(centros,(t,0,2*pi), color='red')
    grafica=curva+evoluta

    if draw_normal_lines:
        f=2*pi/360
        lineas=sum(line2d( [ (gamma[0](t=i*f), gamma[1](t=i*f)), 
                             (centros[0](t=i*f), centros[1](t=i*f)) ], 
                           thickness=1,rgbcolor=(1,0.8,0.8)) 
                   for i in range(rango_angulos[0], rango_angulos[1]+paso_angulo, paso_angulo))
        grafica+=lineas
    
    if draw_osculating_circle and 0<t0<2*pi:
        punto=point((gamma[0](t=t0), gamma[1](t=t0)), rgbcolor=hue(0),pointsize=30)
        circulo=circle( (centros[0](t=t0), centros[1](t=t0)), radio(t=t0) )  
        grafica+=punto+circulo

    show(grafica,aspect_ratio=1,xmin=-2,xmax=2,ymin=-2,ymax=2)
}}}
{{attachment:evoluta3.png}}


== Geodesics on a parametric surface ==
by Antonio Valdés and Pablo Angulo. This example was originally composed of two interacts: 
 - the first allowing the user to introduce a parametric surface, and draw it.
 - the second drawing a geodesic within the surface. 
The separation was so that after the first interact, the geodesic equations were "compiled", thus making the second interact faster.
However, in the following there is only one interact, to make sagecell works.

{{{#!sagecell
from scipy.integrate import odeint

u, v, t, du, dv = var('u v t du dv')

def fading_line3d(points, rgbcolor1, rgbcolor2, *args, **kwds):
    L = len(points)
    vcolor1 = vector(RDF, rgbcolor1)
    vcolor2 = vector(RDF, rgbcolor2)
    return sum(line3d(points[j:j+2], 
                      rgbcolor = tuple( ((L-j)/L)*vcolor1 + (j/L)*vcolor2 ), 
                      *args, **kwds) 
               for j in srange(L-1))

steps = 100

@interact
def _(x = input_box(3*sin(u)*cos(v), 'x'),
      y = input_box(sin(u)*sin(v), 'y'),
      z = input_box(2*cos(u), 'z'),
      int_u = input_grid(1, 2, default = [[0,pi]], label = 'u -interval'), 
      int_v = input_grid(1, 2, default = [[-pi,pi]], label = 'v -interval'),
      init_point = input_grid(1, 2, default = [[-pi/4,pi/8]], label = 'coordinates of \ninitial point'),
      init_vector = input_grid(1, 2, default = [[1,0]], label = 'coordinates of \ninitial vector'),
      int_s = slider(0, 10, 1/10, 
                           default = pi/2, 
                           label = 'geodesic interval'),
      sliding_color = checkbox(True,'change color along the geodesic')):
    int_u = int_u[0]
    int_v = int_v[0]
    u_0, v_0 = init_point[0]
    V_u, V_v = init_vector[0]

    F = vector([x, y, z])

    S_plot = parametric_plot3d( F, 
                                (u, int_u[0], int_u[1]), 
                                (v, int_v[0], int_v[1]))

    dFu = F.diff(u)
    dFv = F.diff(v)

    Fu = fast_float(dFu, u, v)
    Fv = fast_float(dFv, u, v)

    ufunc = function('ufunc')
    vfunc = function('vfunc')

    dFtt = F(u=ufunc(t), v=vfunc(t)).diff(t, t)

    ec1 = dFtt.dot_product(dFu(u=ufunc(t), v=vfunc(t)))
    ec2 = dFtt.dot_product(dFv(u=ufunc(t), v=vfunc(t)))

    dv, ddv, du, ddu = var('dv, ddv, du, ddu')

    diffec1 = ec1.substitute(diff(ufunc(t), t) == du, 
                            diff(ufunc(t), t, t) == ddu,     
                            diff(vfunc(t), t) == dv, 
                            diff(vfunc(t), t, t) == ddv, 
                            ufunc(t) == u, vfunc(t) == v)
    diffec2 = ec2.substitute(diff(ufunc(t), t) == du, 
                            diff(ufunc(t), t, t) == ddu,     
                            diff(vfunc(t), t) == dv, 
                            diff(vfunc(t), t, t) == ddv, 
                            ufunc(t) == u, vfunc(t) == v)
    sols = solve([diffec1 == 0 , diffec2 == 0], ddu, ddv)

    ddu_rhs = (sols[0][0]).rhs().full_simplify()
    ddv_rhs = (sols[0][1]).rhs().full_simplify()

    ddu_ff = fast_float(ddu_rhs, du, dv, u, v)
    ddv_ff = fast_float(ddv_rhs, du, dv, u, v)

    def func(y,t):
        v = list(y)
        return [ddu_ff(*v), ddv_ff(*v), v[0], v[1]]

    Point = [u_0, v_0]
    velocity = [V_u, V_v]
    Point = list(map(float, Point))
    velocity = list(map(float, velocity))

    geo2D_aux = odeint(func,
                       y0 = [velocity[0], velocity[1], Point[0], Point[1]],
                       t = srange(0, int_s, 0.01))

    geo3D = [F(u=l,v=r) for [j, k, l, r] in geo2D_aux]

    if sliding_color:
        g_plot = fading_line3d(geo3D, rgbcolor1 = (1, 0, 0), rgbcolor2 = (0, 1, 0), thickness=4)
    else:
        g_plot = line3d(geo3D, rgbcolor=(0, 1, 0), thickness=4)

    P = F(u=Point[0], v=Point[1])
    P_plot = point3d((P[0], P[1], P[2]), rgbcolor = (0, 0, 0), pointsize = 30)
    V = velocity[0] * Fu(u = Point[0], v = Point[1]) + \
        velocity[1] * Fv(u= Point[0], v = Point[1])
    V_plot = arrow3d(P, P + V, color = 'black')

    show(g_plot + S_plot + V_plot + P_plot,aspect_ratio = [1, 1, 1])
}}}
{{attachment:geodesics1.png}}
{{attachment:geodesics2.png}}

== Dimensional Explorer ==
By Eviatar Bach

Renders 2D images (perspective or spring-layout) and 3D models of 0-10 dimensional hypercubes. It also displays number of edges and vertices.
{{{#!sagecell
@interact
def render(Display=selector(['2D Perspective', '2D Spring-layout', '3D']), Dimension=slider(0,10,default=4, step_size=1), Size=slider(0,10,default=5,step_size=1), Vertices=False, Calculations=False):
    
    if Display=='2D Perspective':
        
        if Dimension==0:
            g=graphs.GridGraph([1])
            print('Vertices:', len(g.vertices()), ('(2^%s)'%Dimension if Calculations else ''), '\nEdges:', len(g.edges()), ('(%s*(%s/2))' %(len(g.vertices()), Dimension) if Calculations else ''))
            g.show(figsize=[Size,Size], vertex_size=30, vertex_labels=False, transparent=True, vertex_colors='black')
            
        else:
            g=graphs.CubeGraph(Dimension)
            print('Vertices:', len(g.vertices()), ('(2^%s)'%Dimension if Calculations else ''), '\nEdges:', len(g.edges()), ('(%s*(%s/2))' %(len(g.vertices()), Dimension) if Calculations else ''))
            g.show(figsize=[Size,Size], vertex_size=(20 if Vertices else 0), vertex_labels=False, transparent=True, vertex_colors='black')
            
    if Display=='2D Spring-layout':
        
        if Dimension==0:
            s=graphs.GridGraph([1])
            print('Vertices:', len(s.vertices()), ('(2^%s)'%Dimension if Calculations else ''), '\nEdges:', len(s.edges()), ('(%s*(%s/2))' %(len(s.vertices()), Dimension) if Calculations else ''))
            s.show(figsize=[Size,Size], vertex_size=30, vertex_labels=False, transparent=True, vertex_colors='black')
            
        else:
            s=graphs.GridGraph([2]*Dimension)
            print('Vertices:', len(s.vertices()), ('(2^%s)'%Dimension if Calculations else ''), '\nEdges:', len(s.edges()), ('(%s*(%s/2))' %(len(s.vertices()), Dimension) if Calculations else ''))
            s.show(figsize=[Size,Size], vertex_size=(20 if Vertices else 0), vertex_labels=False, transparent=True, vertex_colors='black')
            
    if Display=='3D':
        if Dimension==0:
            d=graphs.GridGraph([1])
            print('Vertices:', len(d.vertices()), ('(2^%s)'%Dimension if Calculations else ''), '\nEdges:', len(d.edges()), ('(%s*(%s/2))' %(len(d.vertices()), Dimension) if Calculations else ''))
            d.show3d(figsize=[Size/2,Size/2], vertex_size=0.001)

        else:
            d=graphs.CubeGraph(Dimension)
            print('Vertices:', len(d.vertices()), ('(2^%s)'%Dimension if Calculations else ''), '\nEdges:', len(d.edges()), ('(%s*(%s/2))' %(len(d.vertices()), Dimension) if Calculations else ''))
            d.show3d(figsize=[Size,Size], vertex_size=(0.03 if Vertices else 0.001))
}}}
{{attachment:dimensions.png}}

== Crofton's formula ==
by Pablo Angulo. Illustrates [[http://en.wikipedia.org/wiki/Crofton%27s_formula| Crofton's formula]] by throwing some random lines and computing the intersection number with a given curve. May use either solve for exact computation of the intersections, or may also approximate the curve by straight segments (this is the default).

{{{#!sagecell
from collections import defaultdict

var('t x y')
pin = pi.n()

def longitud(curva, t0, t1):
    dxdt = derivative(curva[0], t)
    dydt = derivative(curva[1], t)
    integrando(t) = sqrt(dxdt^2 + dydt^2)
    i,_ = numerical_integral(integrando, t0, t1)
    return  i

def random_line(cota):
    theta = random()*pin
    k = 2*cota*random() - cota
    return sin(theta)*x + cos(theta)*y + k

def crofton_exact(curva, t0, t1, L, M):
    forget()
    assume(t>t0)
    assume(t<t1)
    pp = parametric_plot(curva, (t, t0, t1), color='red')
    cortesd = defaultdict(int)
    for k in range(L):
        rl = random_line(M)
        ss = solve(rl(x=curva[0], y=curva[1]), t)
        cortes = 0
        for s in ss:
            tt = s.rhs()
            x0,y0 = curva[0](t=tt), curva[1](t=tt)
            if x0 in RR and y0 in RR:
                pp += point2d((x0,y0), pointsize = 30)
                cortes += 1
        if cortes:
            pp += implicit_plot(rl, (x,-M,M), (y,-M,M), color='green')
        else:
            pp += implicit_plot(rl, (x,-M,M), (y,-M,M), color='blue')
        cortesd[cortes] += 1
    return cortesd, pp
    
def random_line_n(cota):
    theta = random()*pin
    k = 2*cota*random() - cota
    return sin(theta), cos(theta), k

def interseccion_sr(punto1, punto2, recta):
    'Devuelve el punto de interseccion de una recta y un segmento, o None si no se cortan'
    x1, y1 = punto1
    x2, y2 = punto2
    a, b, c   = recta
    num = (-c - a*x1 - b*y1)
    den = (a*(x2 - x1) + b*(y2 - y1))
    if (0 < num < den) or (den < num < 0):
        t_i = num/den
        return ((1-t_i)*x1 + t_i*x2, (1-t_i)*y1 + t_i*y2)
    else:
        return None

def interseccion_cr(curva, t0, t1, recta, partes=50):
    '''Devuelve el numero de puntos de interseccion de una curva y una recta'''
    x,y = curva
    paso = (t1 - t0)/partes
    puntos = [(x(t=tr), y(t=tr)) for tr in srange(t0, t1 + paso, paso)]
    intersecciones = (interseccion_sr(puntos[j], puntos[j+1], recta)
                      for j in range(partes - 1))
    return [p for p in intersecciones if p ]
    
def crofton_aprox(curva, t0, t1, L, M):
    cortesd = defaultdict(int)
    pp = parametric_plot(curva, (t, t0, t1), color='red')
    for k in range(L):
        a,b,c = random_line_n(M)
        rl = a*x + b*y + c
        cortes = interseccion_cr(curva, t0, t1, (a,b,c))
        if cortes:
            pp += sum(point2d(p, pointsize = 30) for p in cortes)
            pp += implicit_plot(rl, (x,-M,M), (y,-M,M), color='green')
        else:
            pp += implicit_plot(rl, (x,-M,M), (y,-M,M), color='blue')
        cortesd[len(cortes)] += 1
    return cortesd, pp

def print_stats(d):
    print('Number of lines with k intersection points:')
    print(', '.join('%d:%d' % kv for kv in d.items()))
    
@interact
def crofton_interact(u1 = text_control('x and y coordinates of curve'),
                     curvax = input_box(t^2, label='x(t)' ),
                     curvay = input_box(2*t-1, label='y(t)' ),
                     u2 = text_control('Interval of definition'),
                     t0 = 0, t1 = 1,
                     u3 = text_control('Draw L lines randomly cos(t)x + sin(t)y + K, |K|&lt;M, 0 <= t < 2pi'),
                     M  = 2,
                     L  = 5,
                     u4    = text_control('Use function "solve" from maxima for exact computations?'),
                     exact = checkbox(False),
                     u5    = text_control('Otherwise, a curve is approximated by how many segments?'),
                     steps = slider(4, 40, 4, 8)):

    if exact:
        cortesd, p = crofton_exact((curvax, curvay), t0, t1, L, M)
    else:
        cortesd, p = crofton_aprox((curvax, curvay), t0, t1, L, M)
    p.show(aspect_ratio=1, xmin=-2, xmax=2, ymin=-2,ymax=2)
    print('A curve of length %f'%longitud((curvax, curvay), t0, t1))
    print_stats(cortesd)
    cortes_tot = sum(k*v for k,v in cortesd.items())
    print('Approx length using Crofton\'s formula: %f'%((cortes_tot/L)*(pi*M)))
}}}
{{attachment:crofton4.png}}


== Banchoff-Pohl "area" ==
by Pablo Angulo. Approximates the Banchoff-Pohl "area enclosed by a spatial curve", by throwing some random lines and computing the linking number with the given curve. Lines not linked to the given curve are displayed in red, linked lines are displayed in green.

{{{#!sagecell
from collections import defaultdict
var('t')
a = 0; b= 2*pi

def random_line3d(bound):
    '''Random line in R^3: first choose the guiding vector of the line,
    then choose a point in the plane p perpendicular to that vector.
    '''
    p = vector([(2*random() - 1) for _ in range(3)])
    v = p/norm(p)
    v2, v3 = matrix(v).right_kernel().basis()
    if det(matrix([v, v2, v3]))<0:
        v2, v3 = v3, v2
    v2 = v2/norm(v2)
    v3 = v3 - (v3*v2)*v2
    v3 = v3/norm(v3)
    return v, (2*random()*bound - bound, v2), (2*random()*bound - bound, v3)

def winding_number(x, y, x0, y0):
    x -= x0
    y -= y0
    r2 = x^2 + y^2
    xp = x.derivative(t)
    yp = y.derivative(t)
    integrando(t) = (x*yp -y*xp)/r2
    i,e = numerical_integral(integrando, a, b)
    return round(i/(2*pi))

def linking_number(curve, line):
    v, (a2, v2), (a3, v3) = line
    M = matrix([v, v2, v3])
#    curve2d = (curve*M.inverse())[1:] #This is VERY slow, for some reason!
    curve2d = sum(c*v for c,v in zip(curve,M.inverse()))[1:]
    x, y = curve2d
    return winding_number(x, y, a2,a3)

def color_ln(number):
    if number:
        return (0,1-1/(1+number),0)
    else:
        return (1,0,0)

def banchoff_pohl(curve, L, M):
    ln_d = defaultdict(int)
    pp = parametric_plot3d(curve, (t,0,2*pi), thickness=2)
    for k in range(L):
        a_line = random_line3d(M)
        ln = abs(linking_number(curve, a_line))
        v, (l1, v1), (l2, v2) = a_line
        # fix 30-08-23: coercion of (l1*v1+l2*v2) to SymbolicRing did not work
        pp += parametric_plot3d((1+0*t)*(l1*v1+l2*v2)+t*v,(t,-2,2), 
                                color=color_ln(ln))
        ln_d[ln] += 1
    return ln_d, pp

def print_stats(d):
    print('Number of lines with linking number k:')
    print(', '.join('%d:%d' % kv for kv in d.items()))

@interact
def bp_interact( u1 = text_control('x, y, z coordinates of a closed space curve in [0,2*pi]'),
                 curvax = input_box(cos(t), label='x(t)' ),
                 curvay = input_box(sin(t), label='y(t)' ),
                 curvaz = input_box(0, label='y(t)' ),
                 u2 = text_control('The curve should be contained in a ball of radius M'),
                 M  = 1,
                 u3 = text_control('Use L lines chosen randomly'),
                 L  = 10):
    ln_d, p = banchoff_pohl(vector((curvax, curvay, curvaz)), L, M)
    p.show(aspect_ratio=1, xmin=-2, xmax=2, ymin=-2,ymax=2)
    bp_area_aprox = (sum(k^2*v for k,v in ln_d.iteritems())/L)*2*pi*M^2
    print('Bahnchoff-Pohl area of the curve(aprox): %f' % bp_area_aprox)
    print_stats(ln_d)
}}}

{{attachment:banchoff-pohl.png}}