The complete book in PDF is now available under a Creative Commons BY-NC-ND License:
The book is also available in printed form from Amazon:
The book builds a numerical library from the ground up, called src/nlib.py. It is a pure python library for numerical computations. It doesn't require numpy.
>>> from nlib import *
>>> A = Matrix([[1,2],[4,9]])
>>> print 1/A
>>> print (A+2)*A
>>> B = Matrix(2,2,lambda i,j: i+j**2)
>>> points = [(x0,y0,dy0), (x1,y1,dy1), (x2,y2,dy2), ...]
>>> coefficients, chi2, fitting_function = fit_least_squares(points,POLYNOMIAL(2))
>>> for x,y,dy in points:
>>> print x, y, '~', fitting_function(x)
>>> from math import sin
>>> def f(x): return sin(x)-1+x
>>> x0 = solve_newton(f, 0.0, ap=0.01, rp=0.01, ns=100)
>>> print 'f(%s)=%s ~ 0' % (x0, f(x0))
(ap is target absolute precision, rp is target relative precision, ns is max number of steps)
>>> def f(x): return (sin(x)-1+x)**2
>>> x0 = optimize_newton(f, 0.0, ap=0.01, rp=0.01, ns=100)
>>> print 'f(%s)=%s ~ min f' % (x0, f(x0))
>>> print 'f'(%s)=%s ~ 0' % (x0, D(f)(x0))
>>> x = [random.random() for k in range(100)]
>>> print 'mu =', mean(x)
>>> print 'sigma =', sd(x)
>>> print 'E[x] =', E(lambda x:x, x)
>>> print 'E[x^2] =', E(lambda x:x**2, x)
>>> print 'E[x^3] =', E(lambda x:x**3, x)
>>> y = [random.random() for k in range(100)]
>>> print 'corr(x,y) = ', correlation(x,y)
>>> print 'cov(x,y) = ', covariance(x,y)
>>> google = YStock('GOOG')
>>> current = google.current()
>>> print current['price']
>>> print current['market_cap']
>>> for day in google.historical():
>>> print day['date'], day['adjusted_close'], day['log_return']
>>> d = PersistentDictionary(path='test.sqlite')
>>> d['key'] = 'value'
>>> print d['key']
>>> del d['key']
d works like a drop-in preplacement for any normal Python dictionary except that the data is stored in a sqlite database in a file called "test.sqlite" so it is still there if you re-start the program. Kind of like the shelve module but shelve files cannot safely be accessed by multiple threads/processes unless locked and locking the entire file is not efficient.
>>> pat = [[[0,0], [0]], [[0,1], [1]], [[1,0], [1]], [[1,1], [0]]]
>>> n = NeuralNetwork(2, 2, 1)
>>> n.train(pat)
>>> n.test(pat)
[0, 0] -> [0.00...]
[0, 1] -> [0.98...]
[1, 0] -> [0.98...]
[1, 1] -> [-0.00...]
>>> data = [(x0,y0), ...]
>>> Canvas(title='my plot').plot(data, color='red').save('myplot.png')
nlib plotting requires matplotlib/numpy for the Canvas object only plots are chainable. methods: .plot, .hist, .errorbar, .ellipses
CONSTANT
CUBIC
Canvas
Cholesky
Cluster
D
DD
Dijkstra
DisjointSets
E
Ellipse
HAVE_MATPLOTLIB
Jacobi_eigenvalues
Kruskal
LINEAR
MCEngine
MCG
Markowitz
MarsenneTwister
Matrix
NeuralNetwork
POLYNOMIAL
PersistentDictionary
Prim
PrimVertex
QUADRATIC
QUARTIC
QuadratureIntegrator
RandomSource
StringIO
Trader
YStock
bootstrap
breadth_first_search
compute_correlation
condition_number
confidence_intervals
continuum_knapsack
correlation
covariance
decode_huffman
depth_first_search
encode_huffman
fib
fit
fit_least_squares
gradient
hessian
integrate
integrate_naive
integrate_quadrature_naive
invert_bicgstab
invert_minimum_residual
is_almost_symmetric
is_almost_zero
is_positive_definite
jacobian
lcs
leapfrog
make_maze
mean
memoize
memoize_persistent
needleman_wunsch
norm
optimize_bisection
optimize_golden_search
optimize_newton
optimize_newton_multi (multi-dimentional optimizer)
optimize_newton_multi_imporved
optimize_secant
partial
random
resample
sd
solve_bisection
solve_fixed_point
solve_newton
solve_newton_multi (multi-dimensional solver)
solve_secant
variance
Created by Massimo Di Pierro ([email protected]) @2016 BSDv3 License