annotate DEPENDENCIES/mingw32/Python27/Lib/site-packages/numpy/polynomial/laguerre.py @ 133:4acb5d8d80b6 tip

Don't fail environmental check if README.md exists (but .txt and no-suffix don't)
author Chris Cannam
date Tue, 30 Jul 2019 12:25:44 +0100
parents 2a2c65a20a8b
children
rev   line source
Chris@87 1 """
Chris@87 2 Objects for dealing with Laguerre series.
Chris@87 3
Chris@87 4 This module provides a number of objects (mostly functions) useful for
Chris@87 5 dealing with Laguerre series, including a `Laguerre` class that
Chris@87 6 encapsulates the usual arithmetic operations. (General information
Chris@87 7 on how this module represents and works with such polynomials is in the
Chris@87 8 docstring for its "parent" sub-package, `numpy.polynomial`).
Chris@87 9
Chris@87 10 Constants
Chris@87 11 ---------
Chris@87 12 - `lagdomain` -- Laguerre series default domain, [-1,1].
Chris@87 13 - `lagzero` -- Laguerre series that evaluates identically to 0.
Chris@87 14 - `lagone` -- Laguerre series that evaluates identically to 1.
Chris@87 15 - `lagx` -- Laguerre series for the identity map, ``f(x) = x``.
Chris@87 16
Chris@87 17 Arithmetic
Chris@87 18 ----------
Chris@87 19 - `lagmulx` -- multiply a Laguerre series in ``P_i(x)`` by ``x``.
Chris@87 20 - `lagadd` -- add two Laguerre series.
Chris@87 21 - `lagsub` -- subtract one Laguerre series from another.
Chris@87 22 - `lagmul` -- multiply two Laguerre series.
Chris@87 23 - `lagdiv` -- divide one Laguerre series by another.
Chris@87 24 - `lagval` -- evaluate a Laguerre series at given points.
Chris@87 25 - `lagval2d` -- evaluate a 2D Laguerre series at given points.
Chris@87 26 - `lagval3d` -- evaluate a 3D Laguerre series at given points.
Chris@87 27 - `laggrid2d` -- evaluate a 2D Laguerre series on a Cartesian product.
Chris@87 28 - `laggrid3d` -- evaluate a 3D Laguerre series on a Cartesian product.
Chris@87 29
Chris@87 30 Calculus
Chris@87 31 --------
Chris@87 32 - `lagder` -- differentiate a Laguerre series.
Chris@87 33 - `lagint` -- integrate a Laguerre series.
Chris@87 34
Chris@87 35 Misc Functions
Chris@87 36 --------------
Chris@87 37 - `lagfromroots` -- create a Laguerre series with specified roots.
Chris@87 38 - `lagroots` -- find the roots of a Laguerre series.
Chris@87 39 - `lagvander` -- Vandermonde-like matrix for Laguerre polynomials.
Chris@87 40 - `lagvander2d` -- Vandermonde-like matrix for 2D power series.
Chris@87 41 - `lagvander3d` -- Vandermonde-like matrix for 3D power series.
Chris@87 42 - `laggauss` -- Gauss-Laguerre quadrature, points and weights.
Chris@87 43 - `lagweight` -- Laguerre weight function.
Chris@87 44 - `lagcompanion` -- symmetrized companion matrix in Laguerre form.
Chris@87 45 - `lagfit` -- least-squares fit returning a Laguerre series.
Chris@87 46 - `lagtrim` -- trim leading coefficients from a Laguerre series.
Chris@87 47 - `lagline` -- Laguerre series of given straight line.
Chris@87 48 - `lag2poly` -- convert a Laguerre series to a polynomial.
Chris@87 49 - `poly2lag` -- convert a polynomial to a Laguerre series.
Chris@87 50
Chris@87 51 Classes
Chris@87 52 -------
Chris@87 53 - `Laguerre` -- A Laguerre series class.
Chris@87 54
Chris@87 55 See also
Chris@87 56 --------
Chris@87 57 `numpy.polynomial`
Chris@87 58
Chris@87 59 """
Chris@87 60 from __future__ import division, absolute_import, print_function
Chris@87 61
Chris@87 62 import warnings
Chris@87 63 import numpy as np
Chris@87 64 import numpy.linalg as la
Chris@87 65
Chris@87 66 from . import polyutils as pu
Chris@87 67 from ._polybase import ABCPolyBase
Chris@87 68
Chris@87 69 __all__ = [
Chris@87 70 'lagzero', 'lagone', 'lagx', 'lagdomain', 'lagline', 'lagadd',
Chris@87 71 'lagsub', 'lagmulx', 'lagmul', 'lagdiv', 'lagpow', 'lagval', 'lagder',
Chris@87 72 'lagint', 'lag2poly', 'poly2lag', 'lagfromroots', 'lagvander',
Chris@87 73 'lagfit', 'lagtrim', 'lagroots', 'Laguerre', 'lagval2d', 'lagval3d',
Chris@87 74 'laggrid2d', 'laggrid3d', 'lagvander2d', 'lagvander3d', 'lagcompanion',
Chris@87 75 'laggauss', 'lagweight']
Chris@87 76
Chris@87 77 lagtrim = pu.trimcoef
Chris@87 78
Chris@87 79
Chris@87 80 def poly2lag(pol):
Chris@87 81 """
Chris@87 82 poly2lag(pol)
Chris@87 83
Chris@87 84 Convert a polynomial to a Laguerre series.
Chris@87 85
Chris@87 86 Convert an array representing the coefficients of a polynomial (relative
Chris@87 87 to the "standard" basis) ordered from lowest degree to highest, to an
Chris@87 88 array of the coefficients of the equivalent Laguerre series, ordered
Chris@87 89 from lowest to highest degree.
Chris@87 90
Chris@87 91 Parameters
Chris@87 92 ----------
Chris@87 93 pol : array_like
Chris@87 94 1-D array containing the polynomial coefficients
Chris@87 95
Chris@87 96 Returns
Chris@87 97 -------
Chris@87 98 c : ndarray
Chris@87 99 1-D array containing the coefficients of the equivalent Laguerre
Chris@87 100 series.
Chris@87 101
Chris@87 102 See Also
Chris@87 103 --------
Chris@87 104 lag2poly
Chris@87 105
Chris@87 106 Notes
Chris@87 107 -----
Chris@87 108 The easy way to do conversions between polynomial basis sets
Chris@87 109 is to use the convert method of a class instance.
Chris@87 110
Chris@87 111 Examples
Chris@87 112 --------
Chris@87 113 >>> from numpy.polynomial.laguerre import poly2lag
Chris@87 114 >>> poly2lag(np.arange(4))
Chris@87 115 array([ 23., -63., 58., -18.])
Chris@87 116
Chris@87 117 """
Chris@87 118 [pol] = pu.as_series([pol])
Chris@87 119 deg = len(pol) - 1
Chris@87 120 res = 0
Chris@87 121 for i in range(deg, -1, -1):
Chris@87 122 res = lagadd(lagmulx(res), pol[i])
Chris@87 123 return res
Chris@87 124
Chris@87 125
Chris@87 126 def lag2poly(c):
Chris@87 127 """
Chris@87 128 Convert a Laguerre series to a polynomial.
Chris@87 129
Chris@87 130 Convert an array representing the coefficients of a Laguerre series,
Chris@87 131 ordered from lowest degree to highest, to an array of the coefficients
Chris@87 132 of the equivalent polynomial (relative to the "standard" basis) ordered
Chris@87 133 from lowest to highest degree.
Chris@87 134
Chris@87 135 Parameters
Chris@87 136 ----------
Chris@87 137 c : array_like
Chris@87 138 1-D array containing the Laguerre series coefficients, ordered
Chris@87 139 from lowest order term to highest.
Chris@87 140
Chris@87 141 Returns
Chris@87 142 -------
Chris@87 143 pol : ndarray
Chris@87 144 1-D array containing the coefficients of the equivalent polynomial
Chris@87 145 (relative to the "standard" basis) ordered from lowest order term
Chris@87 146 to highest.
Chris@87 147
Chris@87 148 See Also
Chris@87 149 --------
Chris@87 150 poly2lag
Chris@87 151
Chris@87 152 Notes
Chris@87 153 -----
Chris@87 154 The easy way to do conversions between polynomial basis sets
Chris@87 155 is to use the convert method of a class instance.
Chris@87 156
Chris@87 157 Examples
Chris@87 158 --------
Chris@87 159 >>> from numpy.polynomial.laguerre import lag2poly
Chris@87 160 >>> lag2poly([ 23., -63., 58., -18.])
Chris@87 161 array([ 0., 1., 2., 3.])
Chris@87 162
Chris@87 163 """
Chris@87 164 from .polynomial import polyadd, polysub, polymulx
Chris@87 165
Chris@87 166 [c] = pu.as_series([c])
Chris@87 167 n = len(c)
Chris@87 168 if n == 1:
Chris@87 169 return c
Chris@87 170 else:
Chris@87 171 c0 = c[-2]
Chris@87 172 c1 = c[-1]
Chris@87 173 # i is the current degree of c1
Chris@87 174 for i in range(n - 1, 1, -1):
Chris@87 175 tmp = c0
Chris@87 176 c0 = polysub(c[i - 2], (c1*(i - 1))/i)
Chris@87 177 c1 = polyadd(tmp, polysub((2*i - 1)*c1, polymulx(c1))/i)
Chris@87 178 return polyadd(c0, polysub(c1, polymulx(c1)))
Chris@87 179
Chris@87 180 #
Chris@87 181 # These are constant arrays are of integer type so as to be compatible
Chris@87 182 # with the widest range of other types, such as Decimal.
Chris@87 183 #
Chris@87 184
Chris@87 185 # Laguerre
Chris@87 186 lagdomain = np.array([0, 1])
Chris@87 187
Chris@87 188 # Laguerre coefficients representing zero.
Chris@87 189 lagzero = np.array([0])
Chris@87 190
Chris@87 191 # Laguerre coefficients representing one.
Chris@87 192 lagone = np.array([1])
Chris@87 193
Chris@87 194 # Laguerre coefficients representing the identity x.
Chris@87 195 lagx = np.array([1, -1])
Chris@87 196
Chris@87 197
Chris@87 198 def lagline(off, scl):
Chris@87 199 """
Chris@87 200 Laguerre series whose graph is a straight line.
Chris@87 201
Chris@87 202
Chris@87 203
Chris@87 204 Parameters
Chris@87 205 ----------
Chris@87 206 off, scl : scalars
Chris@87 207 The specified line is given by ``off + scl*x``.
Chris@87 208
Chris@87 209 Returns
Chris@87 210 -------
Chris@87 211 y : ndarray
Chris@87 212 This module's representation of the Laguerre series for
Chris@87 213 ``off + scl*x``.
Chris@87 214
Chris@87 215 See Also
Chris@87 216 --------
Chris@87 217 polyline, chebline
Chris@87 218
Chris@87 219 Examples
Chris@87 220 --------
Chris@87 221 >>> from numpy.polynomial.laguerre import lagline, lagval
Chris@87 222 >>> lagval(0,lagline(3, 2))
Chris@87 223 3.0
Chris@87 224 >>> lagval(1,lagline(3, 2))
Chris@87 225 5.0
Chris@87 226
Chris@87 227 """
Chris@87 228 if scl != 0:
Chris@87 229 return np.array([off + scl, -scl])
Chris@87 230 else:
Chris@87 231 return np.array([off])
Chris@87 232
Chris@87 233
Chris@87 234 def lagfromroots(roots):
Chris@87 235 """
Chris@87 236 Generate a Laguerre series with given roots.
Chris@87 237
Chris@87 238 The function returns the coefficients of the polynomial
Chris@87 239
Chris@87 240 .. math:: p(x) = (x - r_0) * (x - r_1) * ... * (x - r_n),
Chris@87 241
Chris@87 242 in Laguerre form, where the `r_n` are the roots specified in `roots`.
Chris@87 243 If a zero has multiplicity n, then it must appear in `roots` n times.
Chris@87 244 For instance, if 2 is a root of multiplicity three and 3 is a root of
Chris@87 245 multiplicity 2, then `roots` looks something like [2, 2, 2, 3, 3]. The
Chris@87 246 roots can appear in any order.
Chris@87 247
Chris@87 248 If the returned coefficients are `c`, then
Chris@87 249
Chris@87 250 .. math:: p(x) = c_0 + c_1 * L_1(x) + ... + c_n * L_n(x)
Chris@87 251
Chris@87 252 The coefficient of the last term is not generally 1 for monic
Chris@87 253 polynomials in Laguerre form.
Chris@87 254
Chris@87 255 Parameters
Chris@87 256 ----------
Chris@87 257 roots : array_like
Chris@87 258 Sequence containing the roots.
Chris@87 259
Chris@87 260 Returns
Chris@87 261 -------
Chris@87 262 out : ndarray
Chris@87 263 1-D array of coefficients. If all roots are real then `out` is a
Chris@87 264 real array, if some of the roots are complex, then `out` is complex
Chris@87 265 even if all the coefficients in the result are real (see Examples
Chris@87 266 below).
Chris@87 267
Chris@87 268 See Also
Chris@87 269 --------
Chris@87 270 polyfromroots, legfromroots, chebfromroots, hermfromroots,
Chris@87 271 hermefromroots.
Chris@87 272
Chris@87 273 Examples
Chris@87 274 --------
Chris@87 275 >>> from numpy.polynomial.laguerre import lagfromroots, lagval
Chris@87 276 >>> coef = lagfromroots((-1, 0, 1))
Chris@87 277 >>> lagval((-1, 0, 1), coef)
Chris@87 278 array([ 0., 0., 0.])
Chris@87 279 >>> coef = lagfromroots((-1j, 1j))
Chris@87 280 >>> lagval((-1j, 1j), coef)
Chris@87 281 array([ 0.+0.j, 0.+0.j])
Chris@87 282
Chris@87 283 """
Chris@87 284 if len(roots) == 0:
Chris@87 285 return np.ones(1)
Chris@87 286 else:
Chris@87 287 [roots] = pu.as_series([roots], trim=False)
Chris@87 288 roots.sort()
Chris@87 289 p = [lagline(-r, 1) for r in roots]
Chris@87 290 n = len(p)
Chris@87 291 while n > 1:
Chris@87 292 m, r = divmod(n, 2)
Chris@87 293 tmp = [lagmul(p[i], p[i+m]) for i in range(m)]
Chris@87 294 if r:
Chris@87 295 tmp[0] = lagmul(tmp[0], p[-1])
Chris@87 296 p = tmp
Chris@87 297 n = m
Chris@87 298 return p[0]
Chris@87 299
Chris@87 300
Chris@87 301 def lagadd(c1, c2):
Chris@87 302 """
Chris@87 303 Add one Laguerre series to another.
Chris@87 304
Chris@87 305 Returns the sum of two Laguerre series `c1` + `c2`. The arguments
Chris@87 306 are sequences of coefficients ordered from lowest order term to
Chris@87 307 highest, i.e., [1,2,3] represents the series ``P_0 + 2*P_1 + 3*P_2``.
Chris@87 308
Chris@87 309 Parameters
Chris@87 310 ----------
Chris@87 311 c1, c2 : array_like
Chris@87 312 1-D arrays of Laguerre series coefficients ordered from low to
Chris@87 313 high.
Chris@87 314
Chris@87 315 Returns
Chris@87 316 -------
Chris@87 317 out : ndarray
Chris@87 318 Array representing the Laguerre series of their sum.
Chris@87 319
Chris@87 320 See Also
Chris@87 321 --------
Chris@87 322 lagsub, lagmul, lagdiv, lagpow
Chris@87 323
Chris@87 324 Notes
Chris@87 325 -----
Chris@87 326 Unlike multiplication, division, etc., the sum of two Laguerre series
Chris@87 327 is a Laguerre series (without having to "reproject" the result onto
Chris@87 328 the basis set) so addition, just like that of "standard" polynomials,
Chris@87 329 is simply "component-wise."
Chris@87 330
Chris@87 331 Examples
Chris@87 332 --------
Chris@87 333 >>> from numpy.polynomial.laguerre import lagadd
Chris@87 334 >>> lagadd([1, 2, 3], [1, 2, 3, 4])
Chris@87 335 array([ 2., 4., 6., 4.])
Chris@87 336
Chris@87 337
Chris@87 338 """
Chris@87 339 # c1, c2 are trimmed copies
Chris@87 340 [c1, c2] = pu.as_series([c1, c2])
Chris@87 341 if len(c1) > len(c2):
Chris@87 342 c1[:c2.size] += c2
Chris@87 343 ret = c1
Chris@87 344 else:
Chris@87 345 c2[:c1.size] += c1
Chris@87 346 ret = c2
Chris@87 347 return pu.trimseq(ret)
Chris@87 348
Chris@87 349
Chris@87 350 def lagsub(c1, c2):
Chris@87 351 """
Chris@87 352 Subtract one Laguerre series from another.
Chris@87 353
Chris@87 354 Returns the difference of two Laguerre series `c1` - `c2`. The
Chris@87 355 sequences of coefficients are from lowest order term to highest, i.e.,
Chris@87 356 [1,2,3] represents the series ``P_0 + 2*P_1 + 3*P_2``.
Chris@87 357
Chris@87 358 Parameters
Chris@87 359 ----------
Chris@87 360 c1, c2 : array_like
Chris@87 361 1-D arrays of Laguerre series coefficients ordered from low to
Chris@87 362 high.
Chris@87 363
Chris@87 364 Returns
Chris@87 365 -------
Chris@87 366 out : ndarray
Chris@87 367 Of Laguerre series coefficients representing their difference.
Chris@87 368
Chris@87 369 See Also
Chris@87 370 --------
Chris@87 371 lagadd, lagmul, lagdiv, lagpow
Chris@87 372
Chris@87 373 Notes
Chris@87 374 -----
Chris@87 375 Unlike multiplication, division, etc., the difference of two Laguerre
Chris@87 376 series is a Laguerre series (without having to "reproject" the result
Chris@87 377 onto the basis set) so subtraction, just like that of "standard"
Chris@87 378 polynomials, is simply "component-wise."
Chris@87 379
Chris@87 380 Examples
Chris@87 381 --------
Chris@87 382 >>> from numpy.polynomial.laguerre import lagsub
Chris@87 383 >>> lagsub([1, 2, 3, 4], [1, 2, 3])
Chris@87 384 array([ 0., 0., 0., 4.])
Chris@87 385
Chris@87 386 """
Chris@87 387 # c1, c2 are trimmed copies
Chris@87 388 [c1, c2] = pu.as_series([c1, c2])
Chris@87 389 if len(c1) > len(c2):
Chris@87 390 c1[:c2.size] -= c2
Chris@87 391 ret = c1
Chris@87 392 else:
Chris@87 393 c2 = -c2
Chris@87 394 c2[:c1.size] += c1
Chris@87 395 ret = c2
Chris@87 396 return pu.trimseq(ret)
Chris@87 397
Chris@87 398
Chris@87 399 def lagmulx(c):
Chris@87 400 """Multiply a Laguerre series by x.
Chris@87 401
Chris@87 402 Multiply the Laguerre series `c` by x, where x is the independent
Chris@87 403 variable.
Chris@87 404
Chris@87 405
Chris@87 406 Parameters
Chris@87 407 ----------
Chris@87 408 c : array_like
Chris@87 409 1-D array of Laguerre series coefficients ordered from low to
Chris@87 410 high.
Chris@87 411
Chris@87 412 Returns
Chris@87 413 -------
Chris@87 414 out : ndarray
Chris@87 415 Array representing the result of the multiplication.
Chris@87 416
Chris@87 417 Notes
Chris@87 418 -----
Chris@87 419 The multiplication uses the recursion relationship for Laguerre
Chris@87 420 polynomials in the form
Chris@87 421
Chris@87 422 .. math::
Chris@87 423
Chris@87 424 xP_i(x) = (-(i + 1)*P_{i + 1}(x) + (2i + 1)P_{i}(x) - iP_{i - 1}(x))
Chris@87 425
Chris@87 426 Examples
Chris@87 427 --------
Chris@87 428 >>> from numpy.polynomial.laguerre import lagmulx
Chris@87 429 >>> lagmulx([1, 2, 3])
Chris@87 430 array([ -1., -1., 11., -9.])
Chris@87 431
Chris@87 432 """
Chris@87 433 # c is a trimmed copy
Chris@87 434 [c] = pu.as_series([c])
Chris@87 435 # The zero series needs special treatment
Chris@87 436 if len(c) == 1 and c[0] == 0:
Chris@87 437 return c
Chris@87 438
Chris@87 439 prd = np.empty(len(c) + 1, dtype=c.dtype)
Chris@87 440 prd[0] = c[0]
Chris@87 441 prd[1] = -c[0]
Chris@87 442 for i in range(1, len(c)):
Chris@87 443 prd[i + 1] = -c[i]*(i + 1)
Chris@87 444 prd[i] += c[i]*(2*i + 1)
Chris@87 445 prd[i - 1] -= c[i]*i
Chris@87 446 return prd
Chris@87 447
Chris@87 448
Chris@87 449 def lagmul(c1, c2):
Chris@87 450 """
Chris@87 451 Multiply one Laguerre series by another.
Chris@87 452
Chris@87 453 Returns the product of two Laguerre series `c1` * `c2`. The arguments
Chris@87 454 are sequences of coefficients, from lowest order "term" to highest,
Chris@87 455 e.g., [1,2,3] represents the series ``P_0 + 2*P_1 + 3*P_2``.
Chris@87 456
Chris@87 457 Parameters
Chris@87 458 ----------
Chris@87 459 c1, c2 : array_like
Chris@87 460 1-D arrays of Laguerre series coefficients ordered from low to
Chris@87 461 high.
Chris@87 462
Chris@87 463 Returns
Chris@87 464 -------
Chris@87 465 out : ndarray
Chris@87 466 Of Laguerre series coefficients representing their product.
Chris@87 467
Chris@87 468 See Also
Chris@87 469 --------
Chris@87 470 lagadd, lagsub, lagdiv, lagpow
Chris@87 471
Chris@87 472 Notes
Chris@87 473 -----
Chris@87 474 In general, the (polynomial) product of two C-series results in terms
Chris@87 475 that are not in the Laguerre polynomial basis set. Thus, to express
Chris@87 476 the product as a Laguerre series, it is necessary to "reproject" the
Chris@87 477 product onto said basis set, which may produce "unintuitive" (but
Chris@87 478 correct) results; see Examples section below.
Chris@87 479
Chris@87 480 Examples
Chris@87 481 --------
Chris@87 482 >>> from numpy.polynomial.laguerre import lagmul
Chris@87 483 >>> lagmul([1, 2, 3], [0, 1, 2])
Chris@87 484 array([ 8., -13., 38., -51., 36.])
Chris@87 485
Chris@87 486 """
Chris@87 487 # s1, s2 are trimmed copies
Chris@87 488 [c1, c2] = pu.as_series([c1, c2])
Chris@87 489
Chris@87 490 if len(c1) > len(c2):
Chris@87 491 c = c2
Chris@87 492 xs = c1
Chris@87 493 else:
Chris@87 494 c = c1
Chris@87 495 xs = c2
Chris@87 496
Chris@87 497 if len(c) == 1:
Chris@87 498 c0 = c[0]*xs
Chris@87 499 c1 = 0
Chris@87 500 elif len(c) == 2:
Chris@87 501 c0 = c[0]*xs
Chris@87 502 c1 = c[1]*xs
Chris@87 503 else:
Chris@87 504 nd = len(c)
Chris@87 505 c0 = c[-2]*xs
Chris@87 506 c1 = c[-1]*xs
Chris@87 507 for i in range(3, len(c) + 1):
Chris@87 508 tmp = c0
Chris@87 509 nd = nd - 1
Chris@87 510 c0 = lagsub(c[-i]*xs, (c1*(nd - 1))/nd)
Chris@87 511 c1 = lagadd(tmp, lagsub((2*nd - 1)*c1, lagmulx(c1))/nd)
Chris@87 512 return lagadd(c0, lagsub(c1, lagmulx(c1)))
Chris@87 513
Chris@87 514
Chris@87 515 def lagdiv(c1, c2):
Chris@87 516 """
Chris@87 517 Divide one Laguerre series by another.
Chris@87 518
Chris@87 519 Returns the quotient-with-remainder of two Laguerre series
Chris@87 520 `c1` / `c2`. The arguments are sequences of coefficients from lowest
Chris@87 521 order "term" to highest, e.g., [1,2,3] represents the series
Chris@87 522 ``P_0 + 2*P_1 + 3*P_2``.
Chris@87 523
Chris@87 524 Parameters
Chris@87 525 ----------
Chris@87 526 c1, c2 : array_like
Chris@87 527 1-D arrays of Laguerre series coefficients ordered from low to
Chris@87 528 high.
Chris@87 529
Chris@87 530 Returns
Chris@87 531 -------
Chris@87 532 [quo, rem] : ndarrays
Chris@87 533 Of Laguerre series coefficients representing the quotient and
Chris@87 534 remainder.
Chris@87 535
Chris@87 536 See Also
Chris@87 537 --------
Chris@87 538 lagadd, lagsub, lagmul, lagpow
Chris@87 539
Chris@87 540 Notes
Chris@87 541 -----
Chris@87 542 In general, the (polynomial) division of one Laguerre series by another
Chris@87 543 results in quotient and remainder terms that are not in the Laguerre
Chris@87 544 polynomial basis set. Thus, to express these results as a Laguerre
Chris@87 545 series, it is necessary to "reproject" the results onto the Laguerre
Chris@87 546 basis set, which may produce "unintuitive" (but correct) results; see
Chris@87 547 Examples section below.
Chris@87 548
Chris@87 549 Examples
Chris@87 550 --------
Chris@87 551 >>> from numpy.polynomial.laguerre import lagdiv
Chris@87 552 >>> lagdiv([ 8., -13., 38., -51., 36.], [0, 1, 2])
Chris@87 553 (array([ 1., 2., 3.]), array([ 0.]))
Chris@87 554 >>> lagdiv([ 9., -12., 38., -51., 36.], [0, 1, 2])
Chris@87 555 (array([ 1., 2., 3.]), array([ 1., 1.]))
Chris@87 556
Chris@87 557 """
Chris@87 558 # c1, c2 are trimmed copies
Chris@87 559 [c1, c2] = pu.as_series([c1, c2])
Chris@87 560 if c2[-1] == 0:
Chris@87 561 raise ZeroDivisionError()
Chris@87 562
Chris@87 563 lc1 = len(c1)
Chris@87 564 lc2 = len(c2)
Chris@87 565 if lc1 < lc2:
Chris@87 566 return c1[:1]*0, c1
Chris@87 567 elif lc2 == 1:
Chris@87 568 return c1/c2[-1], c1[:1]*0
Chris@87 569 else:
Chris@87 570 quo = np.empty(lc1 - lc2 + 1, dtype=c1.dtype)
Chris@87 571 rem = c1
Chris@87 572 for i in range(lc1 - lc2, - 1, -1):
Chris@87 573 p = lagmul([0]*i + [1], c2)
Chris@87 574 q = rem[-1]/p[-1]
Chris@87 575 rem = rem[:-1] - q*p[:-1]
Chris@87 576 quo[i] = q
Chris@87 577 return quo, pu.trimseq(rem)
Chris@87 578
Chris@87 579
Chris@87 580 def lagpow(c, pow, maxpower=16):
Chris@87 581 """Raise a Laguerre series to a power.
Chris@87 582
Chris@87 583 Returns the Laguerre series `c` raised to the power `pow`. The
Chris@87 584 argument `c` is a sequence of coefficients ordered from low to high.
Chris@87 585 i.e., [1,2,3] is the series ``P_0 + 2*P_1 + 3*P_2.``
Chris@87 586
Chris@87 587 Parameters
Chris@87 588 ----------
Chris@87 589 c : array_like
Chris@87 590 1-D array of Laguerre series coefficients ordered from low to
Chris@87 591 high.
Chris@87 592 pow : integer
Chris@87 593 Power to which the series will be raised
Chris@87 594 maxpower : integer, optional
Chris@87 595 Maximum power allowed. This is mainly to limit growth of the series
Chris@87 596 to unmanageable size. Default is 16
Chris@87 597
Chris@87 598 Returns
Chris@87 599 -------
Chris@87 600 coef : ndarray
Chris@87 601 Laguerre series of power.
Chris@87 602
Chris@87 603 See Also
Chris@87 604 --------
Chris@87 605 lagadd, lagsub, lagmul, lagdiv
Chris@87 606
Chris@87 607 Examples
Chris@87 608 --------
Chris@87 609 >>> from numpy.polynomial.laguerre import lagpow
Chris@87 610 >>> lagpow([1, 2, 3], 2)
Chris@87 611 array([ 14., -16., 56., -72., 54.])
Chris@87 612
Chris@87 613 """
Chris@87 614 # c is a trimmed copy
Chris@87 615 [c] = pu.as_series([c])
Chris@87 616 power = int(pow)
Chris@87 617 if power != pow or power < 0:
Chris@87 618 raise ValueError("Power must be a non-negative integer.")
Chris@87 619 elif maxpower is not None and power > maxpower:
Chris@87 620 raise ValueError("Power is too large")
Chris@87 621 elif power == 0:
Chris@87 622 return np.array([1], dtype=c.dtype)
Chris@87 623 elif power == 1:
Chris@87 624 return c
Chris@87 625 else:
Chris@87 626 # This can be made more efficient by using powers of two
Chris@87 627 # in the usual way.
Chris@87 628 prd = c
Chris@87 629 for i in range(2, power + 1):
Chris@87 630 prd = lagmul(prd, c)
Chris@87 631 return prd
Chris@87 632
Chris@87 633
Chris@87 634 def lagder(c, m=1, scl=1, axis=0):
Chris@87 635 """
Chris@87 636 Differentiate a Laguerre series.
Chris@87 637
Chris@87 638 Returns the Laguerre series coefficients `c` differentiated `m` times
Chris@87 639 along `axis`. At each iteration the result is multiplied by `scl` (the
Chris@87 640 scaling factor is for use in a linear change of variable). The argument
Chris@87 641 `c` is an array of coefficients from low to high degree along each
Chris@87 642 axis, e.g., [1,2,3] represents the series ``1*L_0 + 2*L_1 + 3*L_2``
Chris@87 643 while [[1,2],[1,2]] represents ``1*L_0(x)*L_0(y) + 1*L_1(x)*L_0(y) +
Chris@87 644 2*L_0(x)*L_1(y) + 2*L_1(x)*L_1(y)`` if axis=0 is ``x`` and axis=1 is
Chris@87 645 ``y``.
Chris@87 646
Chris@87 647 Parameters
Chris@87 648 ----------
Chris@87 649 c : array_like
Chris@87 650 Array of Laguerre series coefficients. If `c` is multidimensional
Chris@87 651 the different axis correspond to different variables with the
Chris@87 652 degree in each axis given by the corresponding index.
Chris@87 653 m : int, optional
Chris@87 654 Number of derivatives taken, must be non-negative. (Default: 1)
Chris@87 655 scl : scalar, optional
Chris@87 656 Each differentiation is multiplied by `scl`. The end result is
Chris@87 657 multiplication by ``scl**m``. This is for use in a linear change of
Chris@87 658 variable. (Default: 1)
Chris@87 659 axis : int, optional
Chris@87 660 Axis over which the derivative is taken. (Default: 0).
Chris@87 661
Chris@87 662 .. versionadded:: 1.7.0
Chris@87 663
Chris@87 664 Returns
Chris@87 665 -------
Chris@87 666 der : ndarray
Chris@87 667 Laguerre series of the derivative.
Chris@87 668
Chris@87 669 See Also
Chris@87 670 --------
Chris@87 671 lagint
Chris@87 672
Chris@87 673 Notes
Chris@87 674 -----
Chris@87 675 In general, the result of differentiating a Laguerre series does not
Chris@87 676 resemble the same operation on a power series. Thus the result of this
Chris@87 677 function may be "unintuitive," albeit correct; see Examples section
Chris@87 678 below.
Chris@87 679
Chris@87 680 Examples
Chris@87 681 --------
Chris@87 682 >>> from numpy.polynomial.laguerre import lagder
Chris@87 683 >>> lagder([ 1., 1., 1., -3.])
Chris@87 684 array([ 1., 2., 3.])
Chris@87 685 >>> lagder([ 1., 0., 0., -4., 3.], m=2)
Chris@87 686 array([ 1., 2., 3.])
Chris@87 687
Chris@87 688 """
Chris@87 689 c = np.array(c, ndmin=1, copy=1)
Chris@87 690 if c.dtype.char in '?bBhHiIlLqQpP':
Chris@87 691 c = c.astype(np.double)
Chris@87 692 cnt, iaxis = [int(t) for t in [m, axis]]
Chris@87 693
Chris@87 694 if cnt != m:
Chris@87 695 raise ValueError("The order of derivation must be integer")
Chris@87 696 if cnt < 0:
Chris@87 697 raise ValueError("The order of derivation must be non-negative")
Chris@87 698 if iaxis != axis:
Chris@87 699 raise ValueError("The axis must be integer")
Chris@87 700 if not -c.ndim <= iaxis < c.ndim:
Chris@87 701 raise ValueError("The axis is out of range")
Chris@87 702 if iaxis < 0:
Chris@87 703 iaxis += c.ndim
Chris@87 704
Chris@87 705 if cnt == 0:
Chris@87 706 return c
Chris@87 707
Chris@87 708 c = np.rollaxis(c, iaxis)
Chris@87 709 n = len(c)
Chris@87 710 if cnt >= n:
Chris@87 711 c = c[:1]*0
Chris@87 712 else:
Chris@87 713 for i in range(cnt):
Chris@87 714 n = n - 1
Chris@87 715 c *= scl
Chris@87 716 der = np.empty((n,) + c.shape[1:], dtype=c.dtype)
Chris@87 717 for j in range(n, 1, -1):
Chris@87 718 der[j - 1] = -c[j]
Chris@87 719 c[j - 1] += c[j]
Chris@87 720 der[0] = -c[1]
Chris@87 721 c = der
Chris@87 722 c = np.rollaxis(c, 0, iaxis + 1)
Chris@87 723 return c
Chris@87 724
Chris@87 725
Chris@87 726 def lagint(c, m=1, k=[], lbnd=0, scl=1, axis=0):
Chris@87 727 """
Chris@87 728 Integrate a Laguerre series.
Chris@87 729
Chris@87 730 Returns the Laguerre series coefficients `c` integrated `m` times from
Chris@87 731 `lbnd` along `axis`. At each iteration the resulting series is
Chris@87 732 **multiplied** by `scl` and an integration constant, `k`, is added.
Chris@87 733 The scaling factor is for use in a linear change of variable. ("Buyer
Chris@87 734 beware": note that, depending on what one is doing, one may want `scl`
Chris@87 735 to be the reciprocal of what one might expect; for more information,
Chris@87 736 see the Notes section below.) The argument `c` is an array of
Chris@87 737 coefficients from low to high degree along each axis, e.g., [1,2,3]
Chris@87 738 represents the series ``L_0 + 2*L_1 + 3*L_2`` while [[1,2],[1,2]]
Chris@87 739 represents ``1*L_0(x)*L_0(y) + 1*L_1(x)*L_0(y) + 2*L_0(x)*L_1(y) +
Chris@87 740 2*L_1(x)*L_1(y)`` if axis=0 is ``x`` and axis=1 is ``y``.
Chris@87 741
Chris@87 742
Chris@87 743 Parameters
Chris@87 744 ----------
Chris@87 745 c : array_like
Chris@87 746 Array of Laguerre series coefficients. If `c` is multidimensional
Chris@87 747 the different axis correspond to different variables with the
Chris@87 748 degree in each axis given by the corresponding index.
Chris@87 749 m : int, optional
Chris@87 750 Order of integration, must be positive. (Default: 1)
Chris@87 751 k : {[], list, scalar}, optional
Chris@87 752 Integration constant(s). The value of the first integral at
Chris@87 753 ``lbnd`` is the first value in the list, the value of the second
Chris@87 754 integral at ``lbnd`` is the second value, etc. If ``k == []`` (the
Chris@87 755 default), all constants are set to zero. If ``m == 1``, a single
Chris@87 756 scalar can be given instead of a list.
Chris@87 757 lbnd : scalar, optional
Chris@87 758 The lower bound of the integral. (Default: 0)
Chris@87 759 scl : scalar, optional
Chris@87 760 Following each integration the result is *multiplied* by `scl`
Chris@87 761 before the integration constant is added. (Default: 1)
Chris@87 762 axis : int, optional
Chris@87 763 Axis over which the integral is taken. (Default: 0).
Chris@87 764
Chris@87 765 .. versionadded:: 1.7.0
Chris@87 766
Chris@87 767 Returns
Chris@87 768 -------
Chris@87 769 S : ndarray
Chris@87 770 Laguerre series coefficients of the integral.
Chris@87 771
Chris@87 772 Raises
Chris@87 773 ------
Chris@87 774 ValueError
Chris@87 775 If ``m < 0``, ``len(k) > m``, ``np.isscalar(lbnd) == False``, or
Chris@87 776 ``np.isscalar(scl) == False``.
Chris@87 777
Chris@87 778 See Also
Chris@87 779 --------
Chris@87 780 lagder
Chris@87 781
Chris@87 782 Notes
Chris@87 783 -----
Chris@87 784 Note that the result of each integration is *multiplied* by `scl`.
Chris@87 785 Why is this important to note? Say one is making a linear change of
Chris@87 786 variable :math:`u = ax + b` in an integral relative to `x`. Then
Chris@87 787 .. math::`dx = du/a`, so one will need to set `scl` equal to
Chris@87 788 :math:`1/a` - perhaps not what one would have first thought.
Chris@87 789
Chris@87 790 Also note that, in general, the result of integrating a C-series needs
Chris@87 791 to be "reprojected" onto the C-series basis set. Thus, typically,
Chris@87 792 the result of this function is "unintuitive," albeit correct; see
Chris@87 793 Examples section below.
Chris@87 794
Chris@87 795 Examples
Chris@87 796 --------
Chris@87 797 >>> from numpy.polynomial.laguerre import lagint
Chris@87 798 >>> lagint([1,2,3])
Chris@87 799 array([ 1., 1., 1., -3.])
Chris@87 800 >>> lagint([1,2,3], m=2)
Chris@87 801 array([ 1., 0., 0., -4., 3.])
Chris@87 802 >>> lagint([1,2,3], k=1)
Chris@87 803 array([ 2., 1., 1., -3.])
Chris@87 804 >>> lagint([1,2,3], lbnd=-1)
Chris@87 805 array([ 11.5, 1. , 1. , -3. ])
Chris@87 806 >>> lagint([1,2], m=2, k=[1,2], lbnd=-1)
Chris@87 807 array([ 11.16666667, -5. , -3. , 2. ])
Chris@87 808
Chris@87 809 """
Chris@87 810 c = np.array(c, ndmin=1, copy=1)
Chris@87 811 if c.dtype.char in '?bBhHiIlLqQpP':
Chris@87 812 c = c.astype(np.double)
Chris@87 813 if not np.iterable(k):
Chris@87 814 k = [k]
Chris@87 815 cnt, iaxis = [int(t) for t in [m, axis]]
Chris@87 816
Chris@87 817 if cnt != m:
Chris@87 818 raise ValueError("The order of integration must be integer")
Chris@87 819 if cnt < 0:
Chris@87 820 raise ValueError("The order of integration must be non-negative")
Chris@87 821 if len(k) > cnt:
Chris@87 822 raise ValueError("Too many integration constants")
Chris@87 823 if iaxis != axis:
Chris@87 824 raise ValueError("The axis must be integer")
Chris@87 825 if not -c.ndim <= iaxis < c.ndim:
Chris@87 826 raise ValueError("The axis is out of range")
Chris@87 827 if iaxis < 0:
Chris@87 828 iaxis += c.ndim
Chris@87 829
Chris@87 830 if cnt == 0:
Chris@87 831 return c
Chris@87 832
Chris@87 833 c = np.rollaxis(c, iaxis)
Chris@87 834 k = list(k) + [0]*(cnt - len(k))
Chris@87 835 for i in range(cnt):
Chris@87 836 n = len(c)
Chris@87 837 c *= scl
Chris@87 838 if n == 1 and np.all(c[0] == 0):
Chris@87 839 c[0] += k[i]
Chris@87 840 else:
Chris@87 841 tmp = np.empty((n + 1,) + c.shape[1:], dtype=c.dtype)
Chris@87 842 tmp[0] = c[0]
Chris@87 843 tmp[1] = -c[0]
Chris@87 844 for j in range(1, n):
Chris@87 845 tmp[j] += c[j]
Chris@87 846 tmp[j + 1] = -c[j]
Chris@87 847 tmp[0] += k[i] - lagval(lbnd, tmp)
Chris@87 848 c = tmp
Chris@87 849 c = np.rollaxis(c, 0, iaxis + 1)
Chris@87 850 return c
Chris@87 851
Chris@87 852
Chris@87 853 def lagval(x, c, tensor=True):
Chris@87 854 """
Chris@87 855 Evaluate a Laguerre series at points x.
Chris@87 856
Chris@87 857 If `c` is of length `n + 1`, this function returns the value:
Chris@87 858
Chris@87 859 .. math:: p(x) = c_0 * L_0(x) + c_1 * L_1(x) + ... + c_n * L_n(x)
Chris@87 860
Chris@87 861 The parameter `x` is converted to an array only if it is a tuple or a
Chris@87 862 list, otherwise it is treated as a scalar. In either case, either `x`
Chris@87 863 or its elements must support multiplication and addition both with
Chris@87 864 themselves and with the elements of `c`.
Chris@87 865
Chris@87 866 If `c` is a 1-D array, then `p(x)` will have the same shape as `x`. If
Chris@87 867 `c` is multidimensional, then the shape of the result depends on the
Chris@87 868 value of `tensor`. If `tensor` is true the shape will be c.shape[1:] +
Chris@87 869 x.shape. If `tensor` is false the shape will be c.shape[1:]. Note that
Chris@87 870 scalars have shape (,).
Chris@87 871
Chris@87 872 Trailing zeros in the coefficients will be used in the evaluation, so
Chris@87 873 they should be avoided if efficiency is a concern.
Chris@87 874
Chris@87 875 Parameters
Chris@87 876 ----------
Chris@87 877 x : array_like, compatible object
Chris@87 878 If `x` is a list or tuple, it is converted to an ndarray, otherwise
Chris@87 879 it is left unchanged and treated as a scalar. In either case, `x`
Chris@87 880 or its elements must support addition and multiplication with
Chris@87 881 with themselves and with the elements of `c`.
Chris@87 882 c : array_like
Chris@87 883 Array of coefficients ordered so that the coefficients for terms of
Chris@87 884 degree n are contained in c[n]. If `c` is multidimensional the
Chris@87 885 remaining indices enumerate multiple polynomials. In the two
Chris@87 886 dimensional case the coefficients may be thought of as stored in
Chris@87 887 the columns of `c`.
Chris@87 888 tensor : boolean, optional
Chris@87 889 If True, the shape of the coefficient array is extended with ones
Chris@87 890 on the right, one for each dimension of `x`. Scalars have dimension 0
Chris@87 891 for this action. The result is that every column of coefficients in
Chris@87 892 `c` is evaluated for every element of `x`. If False, `x` is broadcast
Chris@87 893 over the columns of `c` for the evaluation. This keyword is useful
Chris@87 894 when `c` is multidimensional. The default value is True.
Chris@87 895
Chris@87 896 .. versionadded:: 1.7.0
Chris@87 897
Chris@87 898 Returns
Chris@87 899 -------
Chris@87 900 values : ndarray, algebra_like
Chris@87 901 The shape of the return value is described above.
Chris@87 902
Chris@87 903 See Also
Chris@87 904 --------
Chris@87 905 lagval2d, laggrid2d, lagval3d, laggrid3d
Chris@87 906
Chris@87 907 Notes
Chris@87 908 -----
Chris@87 909 The evaluation uses Clenshaw recursion, aka synthetic division.
Chris@87 910
Chris@87 911 Examples
Chris@87 912 --------
Chris@87 913 >>> from numpy.polynomial.laguerre import lagval
Chris@87 914 >>> coef = [1,2,3]
Chris@87 915 >>> lagval(1, coef)
Chris@87 916 -0.5
Chris@87 917 >>> lagval([[1,2],[3,4]], coef)
Chris@87 918 array([[-0.5, -4. ],
Chris@87 919 [-4.5, -2. ]])
Chris@87 920
Chris@87 921 """
Chris@87 922 c = np.array(c, ndmin=1, copy=0)
Chris@87 923 if c.dtype.char in '?bBhHiIlLqQpP':
Chris@87 924 c = c.astype(np.double)
Chris@87 925 if isinstance(x, (tuple, list)):
Chris@87 926 x = np.asarray(x)
Chris@87 927 if isinstance(x, np.ndarray) and tensor:
Chris@87 928 c = c.reshape(c.shape + (1,)*x.ndim)
Chris@87 929
Chris@87 930 if len(c) == 1:
Chris@87 931 c0 = c[0]
Chris@87 932 c1 = 0
Chris@87 933 elif len(c) == 2:
Chris@87 934 c0 = c[0]
Chris@87 935 c1 = c[1]
Chris@87 936 else:
Chris@87 937 nd = len(c)
Chris@87 938 c0 = c[-2]
Chris@87 939 c1 = c[-1]
Chris@87 940 for i in range(3, len(c) + 1):
Chris@87 941 tmp = c0
Chris@87 942 nd = nd - 1
Chris@87 943 c0 = c[-i] - (c1*(nd - 1))/nd
Chris@87 944 c1 = tmp + (c1*((2*nd - 1) - x))/nd
Chris@87 945 return c0 + c1*(1 - x)
Chris@87 946
Chris@87 947
Chris@87 948 def lagval2d(x, y, c):
Chris@87 949 """
Chris@87 950 Evaluate a 2-D Laguerre series at points (x, y).
Chris@87 951
Chris@87 952 This function returns the values:
Chris@87 953
Chris@87 954 .. math:: p(x,y) = \\sum_{i,j} c_{i,j} * L_i(x) * L_j(y)
Chris@87 955
Chris@87 956 The parameters `x` and `y` are converted to arrays only if they are
Chris@87 957 tuples or a lists, otherwise they are treated as a scalars and they
Chris@87 958 must have the same shape after conversion. In either case, either `x`
Chris@87 959 and `y` or their elements must support multiplication and addition both
Chris@87 960 with themselves and with the elements of `c`.
Chris@87 961
Chris@87 962 If `c` is a 1-D array a one is implicitly appended to its shape to make
Chris@87 963 it 2-D. The shape of the result will be c.shape[2:] + x.shape.
Chris@87 964
Chris@87 965 Parameters
Chris@87 966 ----------
Chris@87 967 x, y : array_like, compatible objects
Chris@87 968 The two dimensional series is evaluated at the points `(x, y)`,
Chris@87 969 where `x` and `y` must have the same shape. If `x` or `y` is a list
Chris@87 970 or tuple, it is first converted to an ndarray, otherwise it is left
Chris@87 971 unchanged and if it isn't an ndarray it is treated as a scalar.
Chris@87 972 c : array_like
Chris@87 973 Array of coefficients ordered so that the coefficient of the term
Chris@87 974 of multi-degree i,j is contained in ``c[i,j]``. If `c` has
Chris@87 975 dimension greater than two the remaining indices enumerate multiple
Chris@87 976 sets of coefficients.
Chris@87 977
Chris@87 978 Returns
Chris@87 979 -------
Chris@87 980 values : ndarray, compatible object
Chris@87 981 The values of the two dimensional polynomial at points formed with
Chris@87 982 pairs of corresponding values from `x` and `y`.
Chris@87 983
Chris@87 984 See Also
Chris@87 985 --------
Chris@87 986 lagval, laggrid2d, lagval3d, laggrid3d
Chris@87 987
Chris@87 988 Notes
Chris@87 989 -----
Chris@87 990
Chris@87 991 .. versionadded::1.7.0
Chris@87 992
Chris@87 993 """
Chris@87 994 try:
Chris@87 995 x, y = np.array((x, y), copy=0)
Chris@87 996 except:
Chris@87 997 raise ValueError('x, y are incompatible')
Chris@87 998
Chris@87 999 c = lagval(x, c)
Chris@87 1000 c = lagval(y, c, tensor=False)
Chris@87 1001 return c
Chris@87 1002
Chris@87 1003
Chris@87 1004 def laggrid2d(x, y, c):
Chris@87 1005 """
Chris@87 1006 Evaluate a 2-D Laguerre series on the Cartesian product of x and y.
Chris@87 1007
Chris@87 1008 This function returns the values:
Chris@87 1009
Chris@87 1010 .. math:: p(a,b) = \sum_{i,j} c_{i,j} * L_i(a) * L_j(b)
Chris@87 1011
Chris@87 1012 where the points `(a, b)` consist of all pairs formed by taking
Chris@87 1013 `a` from `x` and `b` from `y`. The resulting points form a grid with
Chris@87 1014 `x` in the first dimension and `y` in the second.
Chris@87 1015
Chris@87 1016 The parameters `x` and `y` are converted to arrays only if they are
Chris@87 1017 tuples or a lists, otherwise they are treated as a scalars. In either
Chris@87 1018 case, either `x` and `y` or their elements must support multiplication
Chris@87 1019 and addition both with themselves and with the elements of `c`.
Chris@87 1020
Chris@87 1021 If `c` has fewer than two dimensions, ones are implicitly appended to
Chris@87 1022 its shape to make it 2-D. The shape of the result will be c.shape[2:] +
Chris@87 1023 x.shape + y.shape.
Chris@87 1024
Chris@87 1025 Parameters
Chris@87 1026 ----------
Chris@87 1027 x, y : array_like, compatible objects
Chris@87 1028 The two dimensional series is evaluated at the points in the
Chris@87 1029 Cartesian product of `x` and `y`. If `x` or `y` is a list or
Chris@87 1030 tuple, it is first converted to an ndarray, otherwise it is left
Chris@87 1031 unchanged and, if it isn't an ndarray, it is treated as a scalar.
Chris@87 1032 c : array_like
Chris@87 1033 Array of coefficients ordered so that the coefficient of the term of
Chris@87 1034 multi-degree i,j is contained in `c[i,j]`. If `c` has dimension
Chris@87 1035 greater than two the remaining indices enumerate multiple sets of
Chris@87 1036 coefficients.
Chris@87 1037
Chris@87 1038 Returns
Chris@87 1039 -------
Chris@87 1040 values : ndarray, compatible object
Chris@87 1041 The values of the two dimensional Chebyshev series at points in the
Chris@87 1042 Cartesian product of `x` and `y`.
Chris@87 1043
Chris@87 1044 See Also
Chris@87 1045 --------
Chris@87 1046 lagval, lagval2d, lagval3d, laggrid3d
Chris@87 1047
Chris@87 1048 Notes
Chris@87 1049 -----
Chris@87 1050
Chris@87 1051 .. versionadded::1.7.0
Chris@87 1052
Chris@87 1053 """
Chris@87 1054 c = lagval(x, c)
Chris@87 1055 c = lagval(y, c)
Chris@87 1056 return c
Chris@87 1057
Chris@87 1058
Chris@87 1059 def lagval3d(x, y, z, c):
Chris@87 1060 """
Chris@87 1061 Evaluate a 3-D Laguerre series at points (x, y, z).
Chris@87 1062
Chris@87 1063 This function returns the values:
Chris@87 1064
Chris@87 1065 .. math:: p(x,y,z) = \\sum_{i,j,k} c_{i,j,k} * L_i(x) * L_j(y) * L_k(z)
Chris@87 1066
Chris@87 1067 The parameters `x`, `y`, and `z` are converted to arrays only if
Chris@87 1068 they are tuples or a lists, otherwise they are treated as a scalars and
Chris@87 1069 they must have the same shape after conversion. In either case, either
Chris@87 1070 `x`, `y`, and `z` or their elements must support multiplication and
Chris@87 1071 addition both with themselves and with the elements of `c`.
Chris@87 1072
Chris@87 1073 If `c` has fewer than 3 dimensions, ones are implicitly appended to its
Chris@87 1074 shape to make it 3-D. The shape of the result will be c.shape[3:] +
Chris@87 1075 x.shape.
Chris@87 1076
Chris@87 1077 Parameters
Chris@87 1078 ----------
Chris@87 1079 x, y, z : array_like, compatible object
Chris@87 1080 The three dimensional series is evaluated at the points
Chris@87 1081 `(x, y, z)`, where `x`, `y`, and `z` must have the same shape. If
Chris@87 1082 any of `x`, `y`, or `z` is a list or tuple, it is first converted
Chris@87 1083 to an ndarray, otherwise it is left unchanged and if it isn't an
Chris@87 1084 ndarray it is treated as a scalar.
Chris@87 1085 c : array_like
Chris@87 1086 Array of coefficients ordered so that the coefficient of the term of
Chris@87 1087 multi-degree i,j,k is contained in ``c[i,j,k]``. If `c` has dimension
Chris@87 1088 greater than 3 the remaining indices enumerate multiple sets of
Chris@87 1089 coefficients.
Chris@87 1090
Chris@87 1091 Returns
Chris@87 1092 -------
Chris@87 1093 values : ndarray, compatible object
Chris@87 1094 The values of the multidimension polynomial on points formed with
Chris@87 1095 triples of corresponding values from `x`, `y`, and `z`.
Chris@87 1096
Chris@87 1097 See Also
Chris@87 1098 --------
Chris@87 1099 lagval, lagval2d, laggrid2d, laggrid3d
Chris@87 1100
Chris@87 1101 Notes
Chris@87 1102 -----
Chris@87 1103
Chris@87 1104 .. versionadded::1.7.0
Chris@87 1105
Chris@87 1106 """
Chris@87 1107 try:
Chris@87 1108 x, y, z = np.array((x, y, z), copy=0)
Chris@87 1109 except:
Chris@87 1110 raise ValueError('x, y, z are incompatible')
Chris@87 1111
Chris@87 1112 c = lagval(x, c)
Chris@87 1113 c = lagval(y, c, tensor=False)
Chris@87 1114 c = lagval(z, c, tensor=False)
Chris@87 1115 return c
Chris@87 1116
Chris@87 1117
Chris@87 1118 def laggrid3d(x, y, z, c):
Chris@87 1119 """
Chris@87 1120 Evaluate a 3-D Laguerre series on the Cartesian product of x, y, and z.
Chris@87 1121
Chris@87 1122 This function returns the values:
Chris@87 1123
Chris@87 1124 .. math:: p(a,b,c) = \\sum_{i,j,k} c_{i,j,k} * L_i(a) * L_j(b) * L_k(c)
Chris@87 1125
Chris@87 1126 where the points `(a, b, c)` consist of all triples formed by taking
Chris@87 1127 `a` from `x`, `b` from `y`, and `c` from `z`. The resulting points form
Chris@87 1128 a grid with `x` in the first dimension, `y` in the second, and `z` in
Chris@87 1129 the third.
Chris@87 1130
Chris@87 1131 The parameters `x`, `y`, and `z` are converted to arrays only if they
Chris@87 1132 are tuples or a lists, otherwise they are treated as a scalars. In
Chris@87 1133 either case, either `x`, `y`, and `z` or their elements must support
Chris@87 1134 multiplication and addition both with themselves and with the elements
Chris@87 1135 of `c`.
Chris@87 1136
Chris@87 1137 If `c` has fewer than three dimensions, ones are implicitly appended to
Chris@87 1138 its shape to make it 3-D. The shape of the result will be c.shape[3:] +
Chris@87 1139 x.shape + y.shape + z.shape.
Chris@87 1140
Chris@87 1141 Parameters
Chris@87 1142 ----------
Chris@87 1143 x, y, z : array_like, compatible objects
Chris@87 1144 The three dimensional series is evaluated at the points in the
Chris@87 1145 Cartesian product of `x`, `y`, and `z`. If `x`,`y`, or `z` is a
Chris@87 1146 list or tuple, it is first converted to an ndarray, otherwise it is
Chris@87 1147 left unchanged and, if it isn't an ndarray, it is treated as a
Chris@87 1148 scalar.
Chris@87 1149 c : array_like
Chris@87 1150 Array of coefficients ordered so that the coefficients for terms of
Chris@87 1151 degree i,j are contained in ``c[i,j]``. If `c` has dimension
Chris@87 1152 greater than two the remaining indices enumerate multiple sets of
Chris@87 1153 coefficients.
Chris@87 1154
Chris@87 1155 Returns
Chris@87 1156 -------
Chris@87 1157 values : ndarray, compatible object
Chris@87 1158 The values of the two dimensional polynomial at points in the Cartesian
Chris@87 1159 product of `x` and `y`.
Chris@87 1160
Chris@87 1161 See Also
Chris@87 1162 --------
Chris@87 1163 lagval, lagval2d, laggrid2d, lagval3d
Chris@87 1164
Chris@87 1165 Notes
Chris@87 1166 -----
Chris@87 1167
Chris@87 1168 .. versionadded::1.7.0
Chris@87 1169
Chris@87 1170 """
Chris@87 1171 c = lagval(x, c)
Chris@87 1172 c = lagval(y, c)
Chris@87 1173 c = lagval(z, c)
Chris@87 1174 return c
Chris@87 1175
Chris@87 1176
Chris@87 1177 def lagvander(x, deg):
Chris@87 1178 """Pseudo-Vandermonde matrix of given degree.
Chris@87 1179
Chris@87 1180 Returns the pseudo-Vandermonde matrix of degree `deg` and sample points
Chris@87 1181 `x`. The pseudo-Vandermonde matrix is defined by
Chris@87 1182
Chris@87 1183 .. math:: V[..., i] = L_i(x)
Chris@87 1184
Chris@87 1185 where `0 <= i <= deg`. The leading indices of `V` index the elements of
Chris@87 1186 `x` and the last index is the degree of the Laguerre polynomial.
Chris@87 1187
Chris@87 1188 If `c` is a 1-D array of coefficients of length `n + 1` and `V` is the
Chris@87 1189 array ``V = lagvander(x, n)``, then ``np.dot(V, c)`` and
Chris@87 1190 ``lagval(x, c)`` are the same up to roundoff. This equivalence is
Chris@87 1191 useful both for least squares fitting and for the evaluation of a large
Chris@87 1192 number of Laguerre series of the same degree and sample points.
Chris@87 1193
Chris@87 1194 Parameters
Chris@87 1195 ----------
Chris@87 1196 x : array_like
Chris@87 1197 Array of points. The dtype is converted to float64 or complex128
Chris@87 1198 depending on whether any of the elements are complex. If `x` is
Chris@87 1199 scalar it is converted to a 1-D array.
Chris@87 1200 deg : int
Chris@87 1201 Degree of the resulting matrix.
Chris@87 1202
Chris@87 1203 Returns
Chris@87 1204 -------
Chris@87 1205 vander : ndarray
Chris@87 1206 The pseudo-Vandermonde matrix. The shape of the returned matrix is
Chris@87 1207 ``x.shape + (deg + 1,)``, where The last index is the degree of the
Chris@87 1208 corresponding Laguerre polynomial. The dtype will be the same as
Chris@87 1209 the converted `x`.
Chris@87 1210
Chris@87 1211 Examples
Chris@87 1212 --------
Chris@87 1213 >>> from numpy.polynomial.laguerre import lagvander
Chris@87 1214 >>> x = np.array([0, 1, 2])
Chris@87 1215 >>> lagvander(x, 3)
Chris@87 1216 array([[ 1. , 1. , 1. , 1. ],
Chris@87 1217 [ 1. , 0. , -0.5 , -0.66666667],
Chris@87 1218 [ 1. , -1. , -1. , -0.33333333]])
Chris@87 1219
Chris@87 1220 """
Chris@87 1221 ideg = int(deg)
Chris@87 1222 if ideg != deg:
Chris@87 1223 raise ValueError("deg must be integer")
Chris@87 1224 if ideg < 0:
Chris@87 1225 raise ValueError("deg must be non-negative")
Chris@87 1226
Chris@87 1227 x = np.array(x, copy=0, ndmin=1) + 0.0
Chris@87 1228 dims = (ideg + 1,) + x.shape
Chris@87 1229 dtyp = x.dtype
Chris@87 1230 v = np.empty(dims, dtype=dtyp)
Chris@87 1231 v[0] = x*0 + 1
Chris@87 1232 if ideg > 0:
Chris@87 1233 v[1] = 1 - x
Chris@87 1234 for i in range(2, ideg + 1):
Chris@87 1235 v[i] = (v[i-1]*(2*i - 1 - x) - v[i-2]*(i - 1))/i
Chris@87 1236 return np.rollaxis(v, 0, v.ndim)
Chris@87 1237
Chris@87 1238
Chris@87 1239 def lagvander2d(x, y, deg):
Chris@87 1240 """Pseudo-Vandermonde matrix of given degrees.
Chris@87 1241
Chris@87 1242 Returns the pseudo-Vandermonde matrix of degrees `deg` and sample
Chris@87 1243 points `(x, y)`. The pseudo-Vandermonde matrix is defined by
Chris@87 1244
Chris@87 1245 .. math:: V[..., deg[1]*i + j] = L_i(x) * L_j(y),
Chris@87 1246
Chris@87 1247 where `0 <= i <= deg[0]` and `0 <= j <= deg[1]`. The leading indices of
Chris@87 1248 `V` index the points `(x, y)` and the last index encodes the degrees of
Chris@87 1249 the Laguerre polynomials.
Chris@87 1250
Chris@87 1251 If ``V = lagvander2d(x, y, [xdeg, ydeg])``, then the columns of `V`
Chris@87 1252 correspond to the elements of a 2-D coefficient array `c` of shape
Chris@87 1253 (xdeg + 1, ydeg + 1) in the order
Chris@87 1254
Chris@87 1255 .. math:: c_{00}, c_{01}, c_{02} ... , c_{10}, c_{11}, c_{12} ...
Chris@87 1256
Chris@87 1257 and ``np.dot(V, c.flat)`` and ``lagval2d(x, y, c)`` will be the same
Chris@87 1258 up to roundoff. This equivalence is useful both for least squares
Chris@87 1259 fitting and for the evaluation of a large number of 2-D Laguerre
Chris@87 1260 series of the same degrees and sample points.
Chris@87 1261
Chris@87 1262 Parameters
Chris@87 1263 ----------
Chris@87 1264 x, y : array_like
Chris@87 1265 Arrays of point coordinates, all of the same shape. The dtypes
Chris@87 1266 will be converted to either float64 or complex128 depending on
Chris@87 1267 whether any of the elements are complex. Scalars are converted to
Chris@87 1268 1-D arrays.
Chris@87 1269 deg : list of ints
Chris@87 1270 List of maximum degrees of the form [x_deg, y_deg].
Chris@87 1271
Chris@87 1272 Returns
Chris@87 1273 -------
Chris@87 1274 vander2d : ndarray
Chris@87 1275 The shape of the returned matrix is ``x.shape + (order,)``, where
Chris@87 1276 :math:`order = (deg[0]+1)*(deg([1]+1)`. The dtype will be the same
Chris@87 1277 as the converted `x` and `y`.
Chris@87 1278
Chris@87 1279 See Also
Chris@87 1280 --------
Chris@87 1281 lagvander, lagvander3d. lagval2d, lagval3d
Chris@87 1282
Chris@87 1283 Notes
Chris@87 1284 -----
Chris@87 1285
Chris@87 1286 .. versionadded::1.7.0
Chris@87 1287
Chris@87 1288 """
Chris@87 1289 ideg = [int(d) for d in deg]
Chris@87 1290 is_valid = [id == d and id >= 0 for id, d in zip(ideg, deg)]
Chris@87 1291 if is_valid != [1, 1]:
Chris@87 1292 raise ValueError("degrees must be non-negative integers")
Chris@87 1293 degx, degy = ideg
Chris@87 1294 x, y = np.array((x, y), copy=0) + 0.0
Chris@87 1295
Chris@87 1296 vx = lagvander(x, degx)
Chris@87 1297 vy = lagvander(y, degy)
Chris@87 1298 v = vx[..., None]*vy[..., None,:]
Chris@87 1299 return v.reshape(v.shape[:-2] + (-1,))
Chris@87 1300
Chris@87 1301
Chris@87 1302 def lagvander3d(x, y, z, deg):
Chris@87 1303 """Pseudo-Vandermonde matrix of given degrees.
Chris@87 1304
Chris@87 1305 Returns the pseudo-Vandermonde matrix of degrees `deg` and sample
Chris@87 1306 points `(x, y, z)`. If `l, m, n` are the given degrees in `x, y, z`,
Chris@87 1307 then The pseudo-Vandermonde matrix is defined by
Chris@87 1308
Chris@87 1309 .. math:: V[..., (m+1)(n+1)i + (n+1)j + k] = L_i(x)*L_j(y)*L_k(z),
Chris@87 1310
Chris@87 1311 where `0 <= i <= l`, `0 <= j <= m`, and `0 <= j <= n`. The leading
Chris@87 1312 indices of `V` index the points `(x, y, z)` and the last index encodes
Chris@87 1313 the degrees of the Laguerre polynomials.
Chris@87 1314
Chris@87 1315 If ``V = lagvander3d(x, y, z, [xdeg, ydeg, zdeg])``, then the columns
Chris@87 1316 of `V` correspond to the elements of a 3-D coefficient array `c` of
Chris@87 1317 shape (xdeg + 1, ydeg + 1, zdeg + 1) in the order
Chris@87 1318
Chris@87 1319 .. math:: c_{000}, c_{001}, c_{002},... , c_{010}, c_{011}, c_{012},...
Chris@87 1320
Chris@87 1321 and ``np.dot(V, c.flat)`` and ``lagval3d(x, y, z, c)`` will be the
Chris@87 1322 same up to roundoff. This equivalence is useful both for least squares
Chris@87 1323 fitting and for the evaluation of a large number of 3-D Laguerre
Chris@87 1324 series of the same degrees and sample points.
Chris@87 1325
Chris@87 1326 Parameters
Chris@87 1327 ----------
Chris@87 1328 x, y, z : array_like
Chris@87 1329 Arrays of point coordinates, all of the same shape. The dtypes will
Chris@87 1330 be converted to either float64 or complex128 depending on whether
Chris@87 1331 any of the elements are complex. Scalars are converted to 1-D
Chris@87 1332 arrays.
Chris@87 1333 deg : list of ints
Chris@87 1334 List of maximum degrees of the form [x_deg, y_deg, z_deg].
Chris@87 1335
Chris@87 1336 Returns
Chris@87 1337 -------
Chris@87 1338 vander3d : ndarray
Chris@87 1339 The shape of the returned matrix is ``x.shape + (order,)``, where
Chris@87 1340 :math:`order = (deg[0]+1)*(deg([1]+1)*(deg[2]+1)`. The dtype will
Chris@87 1341 be the same as the converted `x`, `y`, and `z`.
Chris@87 1342
Chris@87 1343 See Also
Chris@87 1344 --------
Chris@87 1345 lagvander, lagvander3d. lagval2d, lagval3d
Chris@87 1346
Chris@87 1347 Notes
Chris@87 1348 -----
Chris@87 1349
Chris@87 1350 .. versionadded::1.7.0
Chris@87 1351
Chris@87 1352 """
Chris@87 1353 ideg = [int(d) for d in deg]
Chris@87 1354 is_valid = [id == d and id >= 0 for id, d in zip(ideg, deg)]
Chris@87 1355 if is_valid != [1, 1, 1]:
Chris@87 1356 raise ValueError("degrees must be non-negative integers")
Chris@87 1357 degx, degy, degz = ideg
Chris@87 1358 x, y, z = np.array((x, y, z), copy=0) + 0.0
Chris@87 1359
Chris@87 1360 vx = lagvander(x, degx)
Chris@87 1361 vy = lagvander(y, degy)
Chris@87 1362 vz = lagvander(z, degz)
Chris@87 1363 v = vx[..., None, None]*vy[..., None,:, None]*vz[..., None, None,:]
Chris@87 1364 return v.reshape(v.shape[:-3] + (-1,))
Chris@87 1365
Chris@87 1366
Chris@87 1367 def lagfit(x, y, deg, rcond=None, full=False, w=None):
Chris@87 1368 """
Chris@87 1369 Least squares fit of Laguerre series to data.
Chris@87 1370
Chris@87 1371 Return the coefficients of a Laguerre series of degree `deg` that is the
Chris@87 1372 least squares fit to the data values `y` given at points `x`. If `y` is
Chris@87 1373 1-D the returned coefficients will also be 1-D. If `y` is 2-D multiple
Chris@87 1374 fits are done, one for each column of `y`, and the resulting
Chris@87 1375 coefficients are stored in the corresponding columns of a 2-D return.
Chris@87 1376 The fitted polynomial(s) are in the form
Chris@87 1377
Chris@87 1378 .. math:: p(x) = c_0 + c_1 * L_1(x) + ... + c_n * L_n(x),
Chris@87 1379
Chris@87 1380 where `n` is `deg`.
Chris@87 1381
Chris@87 1382 Parameters
Chris@87 1383 ----------
Chris@87 1384 x : array_like, shape (M,)
Chris@87 1385 x-coordinates of the M sample points ``(x[i], y[i])``.
Chris@87 1386 y : array_like, shape (M,) or (M, K)
Chris@87 1387 y-coordinates of the sample points. Several data sets of sample
Chris@87 1388 points sharing the same x-coordinates can be fitted at once by
Chris@87 1389 passing in a 2D-array that contains one dataset per column.
Chris@87 1390 deg : int
Chris@87 1391 Degree of the fitting polynomial
Chris@87 1392 rcond : float, optional
Chris@87 1393 Relative condition number of the fit. Singular values smaller than
Chris@87 1394 this relative to the largest singular value will be ignored. The
Chris@87 1395 default value is len(x)*eps, where eps is the relative precision of
Chris@87 1396 the float type, about 2e-16 in most cases.
Chris@87 1397 full : bool, optional
Chris@87 1398 Switch determining nature of return value. When it is False (the
Chris@87 1399 default) just the coefficients are returned, when True diagnostic
Chris@87 1400 information from the singular value decomposition is also returned.
Chris@87 1401 w : array_like, shape (`M`,), optional
Chris@87 1402 Weights. If not None, the contribution of each point
Chris@87 1403 ``(x[i],y[i])`` to the fit is weighted by `w[i]`. Ideally the
Chris@87 1404 weights are chosen so that the errors of the products ``w[i]*y[i]``
Chris@87 1405 all have the same variance. The default value is None.
Chris@87 1406
Chris@87 1407 Returns
Chris@87 1408 -------
Chris@87 1409 coef : ndarray, shape (M,) or (M, K)
Chris@87 1410 Laguerre coefficients ordered from low to high. If `y` was 2-D,
Chris@87 1411 the coefficients for the data in column k of `y` are in column
Chris@87 1412 `k`.
Chris@87 1413
Chris@87 1414 [residuals, rank, singular_values, rcond] : list
Chris@87 1415 These values are only returned if `full` = True
Chris@87 1416
Chris@87 1417 resid -- sum of squared residuals of the least squares fit
Chris@87 1418 rank -- the numerical rank of the scaled Vandermonde matrix
Chris@87 1419 sv -- singular values of the scaled Vandermonde matrix
Chris@87 1420 rcond -- value of `rcond`.
Chris@87 1421
Chris@87 1422 For more details, see `linalg.lstsq`.
Chris@87 1423
Chris@87 1424 Warns
Chris@87 1425 -----
Chris@87 1426 RankWarning
Chris@87 1427 The rank of the coefficient matrix in the least-squares fit is
Chris@87 1428 deficient. The warning is only raised if `full` = False. The
Chris@87 1429 warnings can be turned off by
Chris@87 1430
Chris@87 1431 >>> import warnings
Chris@87 1432 >>> warnings.simplefilter('ignore', RankWarning)
Chris@87 1433
Chris@87 1434 See Also
Chris@87 1435 --------
Chris@87 1436 chebfit, legfit, polyfit, hermfit, hermefit
Chris@87 1437 lagval : Evaluates a Laguerre series.
Chris@87 1438 lagvander : pseudo Vandermonde matrix of Laguerre series.
Chris@87 1439 lagweight : Laguerre weight function.
Chris@87 1440 linalg.lstsq : Computes a least-squares fit from the matrix.
Chris@87 1441 scipy.interpolate.UnivariateSpline : Computes spline fits.
Chris@87 1442
Chris@87 1443 Notes
Chris@87 1444 -----
Chris@87 1445 The solution is the coefficients of the Laguerre series `p` that
Chris@87 1446 minimizes the sum of the weighted squared errors
Chris@87 1447
Chris@87 1448 .. math:: E = \\sum_j w_j^2 * |y_j - p(x_j)|^2,
Chris@87 1449
Chris@87 1450 where the :math:`w_j` are the weights. This problem is solved by
Chris@87 1451 setting up as the (typically) overdetermined matrix equation
Chris@87 1452
Chris@87 1453 .. math:: V(x) * c = w * y,
Chris@87 1454
Chris@87 1455 where `V` is the weighted pseudo Vandermonde matrix of `x`, `c` are the
Chris@87 1456 coefficients to be solved for, `w` are the weights, and `y` are the
Chris@87 1457 observed values. This equation is then solved using the singular value
Chris@87 1458 decomposition of `V`.
Chris@87 1459
Chris@87 1460 If some of the singular values of `V` are so small that they are
Chris@87 1461 neglected, then a `RankWarning` will be issued. This means that the
Chris@87 1462 coefficient values may be poorly determined. Using a lower order fit
Chris@87 1463 will usually get rid of the warning. The `rcond` parameter can also be
Chris@87 1464 set to a value smaller than its default, but the resulting fit may be
Chris@87 1465 spurious and have large contributions from roundoff error.
Chris@87 1466
Chris@87 1467 Fits using Laguerre series are probably most useful when the data can
Chris@87 1468 be approximated by ``sqrt(w(x)) * p(x)``, where `w(x)` is the Laguerre
Chris@87 1469 weight. In that case the weight ``sqrt(w(x[i])`` should be used
Chris@87 1470 together with data values ``y[i]/sqrt(w(x[i])``. The weight function is
Chris@87 1471 available as `lagweight`.
Chris@87 1472
Chris@87 1473 References
Chris@87 1474 ----------
Chris@87 1475 .. [1] Wikipedia, "Curve fitting",
Chris@87 1476 http://en.wikipedia.org/wiki/Curve_fitting
Chris@87 1477
Chris@87 1478 Examples
Chris@87 1479 --------
Chris@87 1480 >>> from numpy.polynomial.laguerre import lagfit, lagval
Chris@87 1481 >>> x = np.linspace(0, 10)
Chris@87 1482 >>> err = np.random.randn(len(x))/10
Chris@87 1483 >>> y = lagval(x, [1, 2, 3]) + err
Chris@87 1484 >>> lagfit(x, y, 2)
Chris@87 1485 array([ 0.96971004, 2.00193749, 3.00288744])
Chris@87 1486
Chris@87 1487 """
Chris@87 1488 order = int(deg) + 1
Chris@87 1489 x = np.asarray(x) + 0.0
Chris@87 1490 y = np.asarray(y) + 0.0
Chris@87 1491
Chris@87 1492 # check arguments.
Chris@87 1493 if deg < 0:
Chris@87 1494 raise ValueError("expected deg >= 0")
Chris@87 1495 if x.ndim != 1:
Chris@87 1496 raise TypeError("expected 1D vector for x")
Chris@87 1497 if x.size == 0:
Chris@87 1498 raise TypeError("expected non-empty vector for x")
Chris@87 1499 if y.ndim < 1 or y.ndim > 2:
Chris@87 1500 raise TypeError("expected 1D or 2D array for y")
Chris@87 1501 if len(x) != len(y):
Chris@87 1502 raise TypeError("expected x and y to have same length")
Chris@87 1503
Chris@87 1504 # set up the least squares matrices in transposed form
Chris@87 1505 lhs = lagvander(x, deg).T
Chris@87 1506 rhs = y.T
Chris@87 1507 if w is not None:
Chris@87 1508 w = np.asarray(w) + 0.0
Chris@87 1509 if w.ndim != 1:
Chris@87 1510 raise TypeError("expected 1D vector for w")
Chris@87 1511 if len(x) != len(w):
Chris@87 1512 raise TypeError("expected x and w to have same length")
Chris@87 1513 # apply weights. Don't use inplace operations as they
Chris@87 1514 # can cause problems with NA.
Chris@87 1515 lhs = lhs * w
Chris@87 1516 rhs = rhs * w
Chris@87 1517
Chris@87 1518 # set rcond
Chris@87 1519 if rcond is None:
Chris@87 1520 rcond = len(x)*np.finfo(x.dtype).eps
Chris@87 1521
Chris@87 1522 # Determine the norms of the design matrix columns.
Chris@87 1523 if issubclass(lhs.dtype.type, np.complexfloating):
Chris@87 1524 scl = np.sqrt((np.square(lhs.real) + np.square(lhs.imag)).sum(1))
Chris@87 1525 else:
Chris@87 1526 scl = np.sqrt(np.square(lhs).sum(1))
Chris@87 1527 scl[scl == 0] = 1
Chris@87 1528
Chris@87 1529 # Solve the least squares problem.
Chris@87 1530 c, resids, rank, s = la.lstsq(lhs.T/scl, rhs.T, rcond)
Chris@87 1531 c = (c.T/scl).T
Chris@87 1532
Chris@87 1533 # warn on rank reduction
Chris@87 1534 if rank != order and not full:
Chris@87 1535 msg = "The fit may be poorly conditioned"
Chris@87 1536 warnings.warn(msg, pu.RankWarning)
Chris@87 1537
Chris@87 1538 if full:
Chris@87 1539 return c, [resids, rank, s, rcond]
Chris@87 1540 else:
Chris@87 1541 return c
Chris@87 1542
Chris@87 1543
Chris@87 1544 def lagcompanion(c):
Chris@87 1545 """
Chris@87 1546 Return the companion matrix of c.
Chris@87 1547
Chris@87 1548 The usual companion matrix of the Laguerre polynomials is already
Chris@87 1549 symmetric when `c` is a basis Laguerre polynomial, so no scaling is
Chris@87 1550 applied.
Chris@87 1551
Chris@87 1552 Parameters
Chris@87 1553 ----------
Chris@87 1554 c : array_like
Chris@87 1555 1-D array of Laguerre series coefficients ordered from low to high
Chris@87 1556 degree.
Chris@87 1557
Chris@87 1558 Returns
Chris@87 1559 -------
Chris@87 1560 mat : ndarray
Chris@87 1561 Companion matrix of dimensions (deg, deg).
Chris@87 1562
Chris@87 1563 Notes
Chris@87 1564 -----
Chris@87 1565
Chris@87 1566 .. versionadded::1.7.0
Chris@87 1567
Chris@87 1568 """
Chris@87 1569 # c is a trimmed copy
Chris@87 1570 [c] = pu.as_series([c])
Chris@87 1571 if len(c) < 2:
Chris@87 1572 raise ValueError('Series must have maximum degree of at least 1.')
Chris@87 1573 if len(c) == 2:
Chris@87 1574 return np.array([[1 + c[0]/c[1]]])
Chris@87 1575
Chris@87 1576 n = len(c) - 1
Chris@87 1577 mat = np.zeros((n, n), dtype=c.dtype)
Chris@87 1578 top = mat.reshape(-1)[1::n+1]
Chris@87 1579 mid = mat.reshape(-1)[0::n+1]
Chris@87 1580 bot = mat.reshape(-1)[n::n+1]
Chris@87 1581 top[...] = -np.arange(1, n)
Chris@87 1582 mid[...] = 2.*np.arange(n) + 1.
Chris@87 1583 bot[...] = top
Chris@87 1584 mat[:, -1] += (c[:-1]/c[-1])*n
Chris@87 1585 return mat
Chris@87 1586
Chris@87 1587
Chris@87 1588 def lagroots(c):
Chris@87 1589 """
Chris@87 1590 Compute the roots of a Laguerre series.
Chris@87 1591
Chris@87 1592 Return the roots (a.k.a. "zeros") of the polynomial
Chris@87 1593
Chris@87 1594 .. math:: p(x) = \\sum_i c[i] * L_i(x).
Chris@87 1595
Chris@87 1596 Parameters
Chris@87 1597 ----------
Chris@87 1598 c : 1-D array_like
Chris@87 1599 1-D array of coefficients.
Chris@87 1600
Chris@87 1601 Returns
Chris@87 1602 -------
Chris@87 1603 out : ndarray
Chris@87 1604 Array of the roots of the series. If all the roots are real,
Chris@87 1605 then `out` is also real, otherwise it is complex.
Chris@87 1606
Chris@87 1607 See Also
Chris@87 1608 --------
Chris@87 1609 polyroots, legroots, chebroots, hermroots, hermeroots
Chris@87 1610
Chris@87 1611 Notes
Chris@87 1612 -----
Chris@87 1613 The root estimates are obtained as the eigenvalues of the companion
Chris@87 1614 matrix, Roots far from the origin of the complex plane may have large
Chris@87 1615 errors due to the numerical instability of the series for such
Chris@87 1616 values. Roots with multiplicity greater than 1 will also show larger
Chris@87 1617 errors as the value of the series near such points is relatively
Chris@87 1618 insensitive to errors in the roots. Isolated roots near the origin can
Chris@87 1619 be improved by a few iterations of Newton's method.
Chris@87 1620
Chris@87 1621 The Laguerre series basis polynomials aren't powers of `x` so the
Chris@87 1622 results of this function may seem unintuitive.
Chris@87 1623
Chris@87 1624 Examples
Chris@87 1625 --------
Chris@87 1626 >>> from numpy.polynomial.laguerre import lagroots, lagfromroots
Chris@87 1627 >>> coef = lagfromroots([0, 1, 2])
Chris@87 1628 >>> coef
Chris@87 1629 array([ 2., -8., 12., -6.])
Chris@87 1630 >>> lagroots(coef)
Chris@87 1631 array([ -4.44089210e-16, 1.00000000e+00, 2.00000000e+00])
Chris@87 1632
Chris@87 1633 """
Chris@87 1634 # c is a trimmed copy
Chris@87 1635 [c] = pu.as_series([c])
Chris@87 1636 if len(c) <= 1:
Chris@87 1637 return np.array([], dtype=c.dtype)
Chris@87 1638 if len(c) == 2:
Chris@87 1639 return np.array([1 + c[0]/c[1]])
Chris@87 1640
Chris@87 1641 m = lagcompanion(c)
Chris@87 1642 r = la.eigvals(m)
Chris@87 1643 r.sort()
Chris@87 1644 return r
Chris@87 1645
Chris@87 1646
Chris@87 1647 def laggauss(deg):
Chris@87 1648 """
Chris@87 1649 Gauss-Laguerre quadrature.
Chris@87 1650
Chris@87 1651 Computes the sample points and weights for Gauss-Laguerre quadrature.
Chris@87 1652 These sample points and weights will correctly integrate polynomials of
Chris@87 1653 degree :math:`2*deg - 1` or less over the interval :math:`[0, \inf]`
Chris@87 1654 with the weight function :math:`f(x) = \exp(-x)`.
Chris@87 1655
Chris@87 1656 Parameters
Chris@87 1657 ----------
Chris@87 1658 deg : int
Chris@87 1659 Number of sample points and weights. It must be >= 1.
Chris@87 1660
Chris@87 1661 Returns
Chris@87 1662 -------
Chris@87 1663 x : ndarray
Chris@87 1664 1-D ndarray containing the sample points.
Chris@87 1665 y : ndarray
Chris@87 1666 1-D ndarray containing the weights.
Chris@87 1667
Chris@87 1668 Notes
Chris@87 1669 -----
Chris@87 1670
Chris@87 1671 .. versionadded::1.7.0
Chris@87 1672
Chris@87 1673 The results have only been tested up to degree 100 higher degrees may
Chris@87 1674 be problematic. The weights are determined by using the fact that
Chris@87 1675
Chris@87 1676 .. math:: w_k = c / (L'_n(x_k) * L_{n-1}(x_k))
Chris@87 1677
Chris@87 1678 where :math:`c` is a constant independent of :math:`k` and :math:`x_k`
Chris@87 1679 is the k'th root of :math:`L_n`, and then scaling the results to get
Chris@87 1680 the right value when integrating 1.
Chris@87 1681
Chris@87 1682 """
Chris@87 1683 ideg = int(deg)
Chris@87 1684 if ideg != deg or ideg < 1:
Chris@87 1685 raise ValueError("deg must be a non-negative integer")
Chris@87 1686
Chris@87 1687 # first approximation of roots. We use the fact that the companion
Chris@87 1688 # matrix is symmetric in this case in order to obtain better zeros.
Chris@87 1689 c = np.array([0]*deg + [1])
Chris@87 1690 m = lagcompanion(c)
Chris@87 1691 x = la.eigvals(m)
Chris@87 1692 x.sort()
Chris@87 1693
Chris@87 1694 # improve roots by one application of Newton
Chris@87 1695 dy = lagval(x, c)
Chris@87 1696 df = lagval(x, lagder(c))
Chris@87 1697 x -= dy/df
Chris@87 1698
Chris@87 1699 # compute the weights. We scale the factor to avoid possible numerical
Chris@87 1700 # overflow.
Chris@87 1701 fm = lagval(x, c[1:])
Chris@87 1702 fm /= np.abs(fm).max()
Chris@87 1703 df /= np.abs(df).max()
Chris@87 1704 w = 1/(fm * df)
Chris@87 1705
Chris@87 1706 # scale w to get the right value, 1 in this case
Chris@87 1707 w /= w.sum()
Chris@87 1708
Chris@87 1709 return x, w
Chris@87 1710
Chris@87 1711
Chris@87 1712 def lagweight(x):
Chris@87 1713 """Weight function of the Laguerre polynomials.
Chris@87 1714
Chris@87 1715 The weight function is :math:`exp(-x)` and the interval of integration
Chris@87 1716 is :math:`[0, \inf]`. The Laguerre polynomials are orthogonal, but not
Chris@87 1717 normalized, with respect to this weight function.
Chris@87 1718
Chris@87 1719 Parameters
Chris@87 1720 ----------
Chris@87 1721 x : array_like
Chris@87 1722 Values at which the weight function will be computed.
Chris@87 1723
Chris@87 1724 Returns
Chris@87 1725 -------
Chris@87 1726 w : ndarray
Chris@87 1727 The weight function at `x`.
Chris@87 1728
Chris@87 1729 Notes
Chris@87 1730 -----
Chris@87 1731
Chris@87 1732 .. versionadded::1.7.0
Chris@87 1733
Chris@87 1734 """
Chris@87 1735 w = np.exp(-x)
Chris@87 1736 return w
Chris@87 1737
Chris@87 1738 #
Chris@87 1739 # Laguerre series class
Chris@87 1740 #
Chris@87 1741
Chris@87 1742 class Laguerre(ABCPolyBase):
Chris@87 1743 """A Laguerre series class.
Chris@87 1744
Chris@87 1745 The Laguerre class provides the standard Python numerical methods
Chris@87 1746 '+', '-', '*', '//', '%', 'divmod', '**', and '()' as well as the
Chris@87 1747 attributes and methods listed in the `ABCPolyBase` documentation.
Chris@87 1748
Chris@87 1749 Parameters
Chris@87 1750 ----------
Chris@87 1751 coef : array_like
Chris@87 1752 Laguerre coefficients in order of increasing degree, i.e,
Chris@87 1753 ``(1, 2, 3)`` gives ``1*L_0(x) + 2*L_1(X) + 3*L_2(x)``.
Chris@87 1754 domain : (2,) array_like, optional
Chris@87 1755 Domain to use. The interval ``[domain[0], domain[1]]`` is mapped
Chris@87 1756 to the interval ``[window[0], window[1]]`` by shifting and scaling.
Chris@87 1757 The default value is [0, 1].
Chris@87 1758 window : (2,) array_like, optional
Chris@87 1759 Window, see `domain` for its use. The default value is [0, 1].
Chris@87 1760
Chris@87 1761 .. versionadded:: 1.6.0
Chris@87 1762
Chris@87 1763 """
Chris@87 1764 # Virtual Functions
Chris@87 1765 _add = staticmethod(lagadd)
Chris@87 1766 _sub = staticmethod(lagsub)
Chris@87 1767 _mul = staticmethod(lagmul)
Chris@87 1768 _div = staticmethod(lagdiv)
Chris@87 1769 _pow = staticmethod(lagpow)
Chris@87 1770 _val = staticmethod(lagval)
Chris@87 1771 _int = staticmethod(lagint)
Chris@87 1772 _der = staticmethod(lagder)
Chris@87 1773 _fit = staticmethod(lagfit)
Chris@87 1774 _line = staticmethod(lagline)
Chris@87 1775 _roots = staticmethod(lagroots)
Chris@87 1776 _fromroots = staticmethod(lagfromroots)
Chris@87 1777
Chris@87 1778 # Virtual properties
Chris@87 1779 nickname = 'lag'
Chris@87 1780 domain = np.array(lagdomain)
Chris@87 1781 window = np.array(lagdomain)