³ò YÄGJc@sGdZddkZddkZdeifd„ƒYZdd„ZdS(s/ ffit.py Functions for Fitness Lev 1 Feb 2009 iÿÿÿÿNtCONSTcBseZdZRS(sI This class is intended be used to fix a parameter during a fit. (t__name__t __module__t__doc__(((sc:\py\lib\ffit.pyR sc s|d jo/g}|D]}|t|tƒ q~}nti|dtƒ}ti|dtidtƒ}|i djot dƒ‚n|i |i jot dƒ‚nti |ƒpt dƒ‚n|d jo@‡fd†}t ii|||||||f|} n@‡fd†}t ii||||||||f|} | d ||>> def func(x, A, B): return A*x + B ... >>> x = numpy.array([1,2,3]) >>> y = numpy.array([1,2,3]) >>> res = ffit.leastsq(x, y, func, [1, 1], [True, True]) >>> print res[0] [ 1.00000000e+00 -4.06333240e-09] >>> res = ffit.leastsq(x, y, func, [1, 0.25], [True, False]) >>> print res[0] [ 0.89285714 0.25 ] >>> res = ffit.leastsq(x, y, func, [ffit.CONST(0.25), 1]) >>> print res[0] [ 0.25 1.5 ] In the second example the intercept of the linear fit is kept fixed at 1, the best fit being y = 0.89*x + 1, rather than y = 1*x + 0.25. Similarily in the third one the slope is fixed at 0.25, and the best ift is y = 0.25*x + 1.5. tdtypetcopyis'params' must be a 1D arrays)'params' and 'mask' must be the same sizes=all 'params' are constrained. no degrees of freedom for a fitcs7tit|ƒƒ}|||<|||<ˆ||ŒS(N(tnumpytzerostlen(t fitparamstmasktxtconststparams(tfunc(sc:\py\lib\ffit.pyt residualsTs  csAtit|ƒƒ}|||<|||s