What is the main difference in the output of functions created with `numpy.frompyfunc` versus `numpy.vectorize`?
Explanation
This question differentiates between two ways of creating custom ufuncs in Python, focusing on the practical difference in their output which affects usability and performance.
Other questions
What are the four internal components of a NumPy ndarray object?
For a typical C order 3 x 4 x 5 array of float64 (8-byte) values, what are the strides?
Which NumPy function is used to check if an array's data type is a subclass of a general type like np.integer or np.floating?
What is the primary difference between the NumPy `ravel` and `flatten` array methods?
What is the result of reshaping the array `arr = np.arange(12).reshape((3, 4))` using `arr.ravel('F')`?
Which pair of NumPy functions are convenience functions for concatenating arrays by rows (axis 0) and by columns (axis 1) respectively?
According to the Broadcasting Rule in NumPy, when are two arrays compatible for broadcasting?
To subtract the row means from a 4x3 array named `arr`, the 1D array of row means, `row_means`, must be reshaped. What is the correct shape for `row_means` to enable broadcasting over axis 1?
What is the purpose of the `np.newaxis` attribute in NumPy?
What does the `np.add.reduce` method do on a 1D array?
Given the array `arr = np.arange(10)`, what is the output of `np.add.reduceat(arr, [0, 5, 8])`?
What is the primary function of the `outer` ufunc method, for example, `np.multiply.outer(x, y)`?
How is a typical structured data type specified when creating a NumPy structured array?
When you access a single field of a structured array, like `sarr['x']`, what is returned?
What is a key advantage of using NumPy structured arrays compared to pandas DataFrames for certain applications?
What is the fundamental difference between `arr.sort()`, an instance method, and `np.sort(arr)`, a top-level function?
Which sorting algorithm in NumPy is the only one guaranteed to be stable?
What is the purpose of the `numpy.partition` function?
What is the primary purpose of the Numba library in the context of NumPy?
In NumPy, what is a memory-mapped file?
When you assign data to a slice of a NumPy `memmap` object, when are the changes guaranteed to be written to the on-disk file?
What does it mean for a NumPy array's memory layout to be 'C-contiguous'?
How can you check the memory layout flags, such as `C_CONTIGUOUS` and `F_CONTIGUOUS`, for a NumPy array named `arr`?
If `arr_f` is a FORTRAN-contiguous array, what will be the boolean value of `arr_f.copy('C').flags.C_CONTIGUOUS`?
What is the purpose of the strides tuple in a NumPy ndarray's internal structure?
When using the `reshape` method on a NumPy array, what does passing -1 as one of the dimension values signify?
Given `arr = np.arange(15)`, what is the shape of the resulting array after executing `arr.reshape((5, -1))`?
In the context of NumPy's advanced array manipulation, what is the key difference in how C/row-major order and FORTRAN/column-major order traverse dimensions?
Which ufunc method produces an array of intermediate 'accumulated' values, similar to how `cumsum` is related to `sum`?
Given a 2D array `arr = np.arange(15).reshape((3, 5))`, what is the result of `np.add.accumulate(arr, axis=1)`?
What is the primary purpose of `numpy.lexsort`?
When using `numpy.lexsort((first_name, last_name))`, which array is used as the primary sort key?
Given `values = np.array(['2:first', '2:second', '1:first', '1:second', '1:third'])` and `key = np.array([2, 2, 1, 1, 1])`, what is the output of `values.take(key.argsort(kind='mergesort'))`?
Which two Python projects are mentioned as providing NumPy-friendly interfaces for storing array data in the HDF5 format?
Which of the following is NOT a performance tip mentioned for getting the best performance out of NumPy?
What is the method resolution order (MRO) for the `np.float64` data type shown in the text?
Which special objects in the NumPy namespace provide a concise way to stack arrays, for example, vertically like `vstack`?
How does the `repeat` method work on a multidimensional array if you pass an array of integers instead of a single integer?
What is the key difference between `tile` and `repeat` in NumPy?
What is the output of `np.tile(arr, (2, 1))` where `arr` is a 2x2 array?
What are the NumPy equivalents of fancy indexing `arr[inds]` for getting and setting values on a single axis?
What does `numpy.searchsorted` return when searching for a value in a sorted array?
Given `arr = np.array([0, 1, 7, 12, 15])`, what is the output of `arr.searchsorted([0, 8, 11, 16])`?
In the context of Numba, what is the purpose of the `nopython=True` option in the `jit` function?
When opening an existing memory map file with `np.memmap`, what information must you still specify?
If an array is C-contiguous, what is generally the fastest way to perform an operation like summing its elements?
When you create a view on a C-contiguous array using a slice like `arr_c[:, :50]`, is the resulting view guaranteed to be contiguous?
In a NumPy structured array, if a field is defined with a shape, such as `('x', np.int64, 3)`, what does accessing that field on a single record, like `arr[0]['x']`, return?
If you have a 3D array `arr` with shape (3, 4, 5) and a 2D array `means` with shape (3, 4) representing the means over axis 2, how must you reshape `means` to subtract it from `arr` using broadcasting?