provisdom.math.tensor — provisdom/math 2.67

=== clj

(=== tensor)

(=== tensor1 tensor2)

(=== tensor1 tensor2 & more)

Tensor equality that works with NaN.

add clj

(add)

(add tensor)

(add tensor1 tensor2)

(add tensor1 tensor2 & more)

Performs element-wise addition for one or more tensors.

average clj

(average tensor)

The average of the values of the elements.

compute-tensor clj

(compute-tensor shape indices->number)

indices->number is a function that takes a vector of indices and returns a number.

divide clj

(divide)

(divide tensor)

(divide tensor & more)

Performs element-wise division for one or more tensors.

ecount clj

(ecount tensor)

Returns the total count of elements.

emap clj

(emap f tensor)

(emap f tensor1 tensor2)

(emap f tensor1 tensor2 tensor3)

(emap f tensor1 tensor2 tensor3 & more)

Element-wise mapping over all elements of one or more tensors. Returns nil if tensors can't be expanded as necessary.

emap-kv clj

(emap-kv f tensor)

(emap-kv f tensor1 tensor2)

(emap-kv f tensor1 tensor2 tensor3)

(emap-kv f tensor1 tensor2 tensor3 & more)

Element-wise mapping over all elements of one or more tensors. Function f takes the indices and one number per tensor, and returns a number.

every-kv? clj

(every-kv? indices+number->bool tensor)

Returns true if function indices+number->bool is true for every element in tensor, else false.

fill-tensor clj

(fill-tensor shape numbers)

Creates a new tensor with a given shape from a sequence of numbers. If numbers doesn't fill tensor, will fill remaining elements with 0.0.

filter-kv clj

(filter-kv index+tensor->bool tensor-v)

Returns a vector of tensors of the tensors in tensor-v for which function index+tensor->bool returns true. index+tensor->bool must be free of side-effects.

first-number clj

(first-number tensor)

Returns the first number in the tensor.

inner-product clj

(inner-product tensor1 tensor2)

The inner product is the generalization of the [[dot-product]]. It is the tensor sum of the tensor products of the corresponding entries of two tensors. The count of both tensors must be the same.

mdl clj

multiply clj

(multiply)

(multiply tensor)

(multiply tensor1 tensor2)

(multiply tensor1 tensor2 & more)

Performs element-wise multiplication for one or more tensors.

norm clj

(norm tensor)

The square-root of the sum of the squared values of the elements.

norm-p clj

(norm-p tensor p)

The 1/p power of the sum of the element values to the power p. To be a norm, p must be <= 1.0.

norm1 clj

(norm1 tensor)

The sum of the absolute values of the elements.

norm2 clj

See norm

normalize clj

(normalize tensor)

Returns as length one in norm2.

normalize-p clj

(normalize-p tensor p)

Returns as length one in norm-p. To be a norm, p must be <= 1.0.

normalize1 clj

(normalize1 tensor)

Returns as length one in norm1.

normalize2 clj

See normalize2

partition-recursively clj

(partition-recursively n tensor)

Partitions recursively in sets of 'n' elements. There may be unused elements. For example, a 1000-element tensor could be partitioned into 10x10x10.

rank clj

(rank tensor)

Returns the rank of a tensor. The rank is equal to the number of dimensions in the tensor's shape.

repeat-tensor clj

(repeat-tensor shape)

(repeat-tensor shape seed-tensor)

Constructs a new tensor of zeros (doubles) with the given shape, or constructs a new tensor with a new shape that is the concat of shape and the shape of the seed-tensor by placing copies of seed-tensor into shape.

rnd-tensor! clj

(rnd-tensor! shape)

Creates a new tensor with a given shape with random doubles.

roughly-distinct clj

(roughly-distinct tensor accu)

Returns a tensor with later duplicate top-level rows (or elements) removed.

roughly? clj

(roughly? tensor1 tensor2 accu)

Returns true if every element compared across two similarly-shaped tensors are within accu of each other.

shape clj

(shape tensor)

Returns the shape of the tensor.

subtract clj

(subtract)

(subtract tensor)

(subtract tensor1 tensor2)

(subtract tensor1 tensor2 & more)

Performs element-wise subtraction for one or more tensors.

tensor? clj

(tensor? x)

Returns true if a tensor (contains numbers only, and each dimension has rows of equal lengths).

to-tensor clj

(to-tensor x)

Tries to convert any x to tensor, otherwise returns nil.

transpose clj