Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/hst/training/events/Python/usignOOFeatures_intro.pdf Дата изменения: Wed Jun 15 22:47:09 2005 Дата индексирования: Sun Dec 23 08:37:03 2007 Кодировка: Поисковые слова: с р р с с п п р п п с с с р р р р р п р п п п п р п р п п

Integration
· Numeric integration much studied
­ But hard to make robust for all functions ­ Problem is finding where function changes rapidly

· Use wavesets customized to each component
­ Indicates where function has rapidly changing detail

· Integration machinery is generic


Integration (cont.)
>>> source = GaussianAbsorption(4000,20,0.8) * \ ... (Gaussian(3000,100) + 1.e10*BlackBody(6000)) >>> source.integrate() # returns integrated flux >>> source = BlackBody(6000).normalize(VMag(23.3)) >>> source = (Gaussian(3000,50) + ... 1.e10*BlackBody(6000)).normalize(Jy(3.1), ... trans=JohnsonV)


Empirical Spectra
· Can be treated exactly like analytical functions
­ Just has a lot of parameters! ­ Can use customized interpolation

>>> data = pyfits.getdata('fuse.fits') >>> source = TableSource(flux=data.field('flux'), ... wavelength=data.field('wave'))


Valid Range
· Need limits to integration · Info on valid range of combined components
­ Intersection of component ranges >>> source.waverange() [985., 1085.] >>> BlackBody(6000).waverange() [None, None]


Normalization
· Need to scale components to various measures of intensity · Optional bandpass for normalization · Generic machinery (part of abstract class)

>>> source = BlackBody(6000).normalize(VMag(23.3)) >>> source = (Gaussian(3000,50) + ... 1.e10*Blackbody(6000)).normalize(Jy(3.1), ... trans=JohnsonV)


Redshift
class SourceFunction: [...] def __call__(self, wavelengths): return self.resteval(invZ(wavelengths, self.z))/(1.+self.z) def waveset(self): return Z(self.restwaveset(), self.z)

The following shows how redshift would be used.
>>> >>> >>> >>> >>> ... source source source source source = = = = = BlackBody(6000,z=1.3) # or BlackBody(6000).z(1.3) (Gaussian(3000,20)+Blackbody(6000)).z(1.3) # or Z(Gaussian(3000,20)+Blackbody(6000), 1.3) (GaussianAbsorption(3000,20,0.8,z=.3)* \ BlackBody(6000,.5)).z(.8)


Units
· Complication of transformation between and
>>> ... >>> ... >>> >>> asJy(BlackBody(6000)(NM(300)) # evaluate at 300 nm, # return in Janskys source = GaussianSource(EV(0.5),EV(0.1)) # centered at #.5 electron volts source(NM(10), funits=Fnu) # or asFnu(source(NM(10), NM(10)))


Advantages
· · · · · Polymorphism Data and Information Hiding Decoupling/minimal interfaces Flexibility Takes advantage of language syntax


Disadvantages
· Harder to follow program flow as compared to procedural flow · Harder to get mindset if experienced in procedural programming · Not appropriate for all problems