Kev tso tawm ntawm Python lub tsev qiv ntawv rau kev tshawb fawb suav NumPy 2.0.0 yog muaj, tsom mus rau kev ua haujlwm nrog ntau qhov sib txawv thiab cov matrices, thiab tseem muab cov kev sau loj ntawm kev ua haujlwm nrog kev siv ntau yam algorithms ntsig txog kev siv matrices. NumPy yog ib lub tsev qiv ntawv nrov tshaj plaws siv rau kev suav lej. Txoj haujlwm code yog sau rau hauv Python siv optimizations hauv C thiab muab faib raws li BSD daim ntawv tso cai.
NumPy 2.0.0 yog thawj qhov tseem ceeb tshaj tawm txij li xyoo 2006.
Ntxiv nrog rau cov yam ntxwv tshiab thiab kev ua tau zoo, cov ntawv tshiab suav nrog kev hloov pauv rau ABI, Python API thiab C-API uas cuam tshuam rov qab sib raug zoo. Piv txwv li, lub tsev qiv ntawv SciPy muab tso ua ke nrog NumPy 1.x yuav tsum tau rov ua dua los ua haujlwm nrog NumPy 2.0. Qee zaum, cov lej hloov pauv yuav raug siv los siv NumPy 2.0 hauv cov ntawv thov.
Qhov kab laum tseem ceeb tshaj plaws yog cuam tshuam nrog kev khaws cia qhov tseeb ntawm cov lus qhia scalar, piv txwv li, "np.float32(3) + 3" tam sim no yuav rov qab tus nqi ntawm hom float32, tsis yog float64, thiab hauv cov lus qhia nrog ntau hom, hom nrog qhov tseeb siab tshaj plaws yuav raug siv rau qhov tshwm sim, piv txwv li "np.array([3], dtype=np.float32) + np.float64(3)" yuav rov qab tus nqi ntawm hom float64. Cov hom integer siv rau ntawm lub platform tau hloov pauv lawm. Windows — ntawm 64-ntsis systems, hom integer 64-ntsis tam sim no siv, thiab ntawm 32-ntsis systems, hom 32-ntsis (yav dhau los, C-hom analogue ntev tau siv, tab sis tam sim no qhov sib npaug np.intp).
Qee cov lus txhais hauv C-API tau hloov lossis tshem tawm, piv txwv li, PyArray_Descr qauv tau hloov pauv. Qhov siab tshaj plaws ntawm qhov ntev thiab cov lus sib cav tau teeb tsa los ntawm NPY_MAXDIMS thiab NPY_MAXARGS macros tau nce mus rau 64. Txhua hom kev nyuaj tau hloov mus rau kev siv hom qauv los ntawm C99 specification (cfloat_t, cdouble_t, clongdouble_t). Ntxiv C API tshiab rau tsim kev cai dtypes. Tshiab simplified pib ua haujlwm PyArray_ImportNumPyAPI thiab PyUFunc_ImportUFuncAPI tau thov.
Python API muab kev sib cais meej ntxiv ntawm pej xeem thiab ntiag tug APIs thiab qhia txog cov qauv qauv tshiab. Txog 100 lub luag haujlwm, cov qauv, thiab qhov tsis tu ncua tau raug tshem tawm ntawm lub ntsiab "np" lub npe, tsis pom zoo, lossis tshem tawm.
Cleaned np.lib namespace. Tus naj npawb ntawm cov khoom hauv lub npe lub ntsiab tau raug txo los ntawm 10%, thiab hauv numpy.lib namespace los ntawm 80%. Lub numpy.core namespace tau raug tsiv mus rau tus kheej. Tshem tawm qee txoj kev los ntawm np.ndarray thiab np.generic chav kawm. Tsim ib lub npe tshiab numpy.stringsf nrog txoj hlua ua haujlwm.
Cov yam ntxwv tshiab suav nrog kev txhawb nqa rau float32 thiab longdouble hom nyob rau hauv tag nrho cov numpy.fft lub zog, kev txhawb nqa rau tus qauv Array API nyob rau hauv lub ntsiab namespace, lub npe tshiab thiab hom rau cov hlua ntev sib txawv. Kev ua tau zoo ntawm kev ua tau zoo tau ua tiav siv Intel x86-simd-sort thiab Google Highway cov tsev qiv ntawv rau txoj hlua kev ua haujlwm, argsort, muab faib thiab argpartition, kev ua haujlwm nrog cov hlua ruaj hauv numpy.char module tau nrawm dua. API tshiab ntxiv
opt_func_info rau tracing thiab introspection.
Tau qhov twg los: opennet.ru
