materialize

Materializes this tensor view into a standalone tensor using the default strategy.

This method creates a new tensor that contains a copy of all data from the view, making it independent of the parent tensor. The default strategy uses immediate copying (CopyMaterializationStrategy) which provides predictable performance and memory usage characteristics.

Usage Examples

val parentTensor = tensorOf(shape = Shape(4, 4)) { /* initialization */}
val view = parentTensor.sliceView { range(1, 3) }
val materialized = view.materialize() // Independent copy of the view data

Materialization Benefits

• Independence: Result tensor has no dependency on parent tensor

• Performance: Direct memory access without coordinate transformation

• Compatibility: Works with all tensor operations and external libraries

• Memory Management: Allows parent tensor to be garbage collected

Parameters

Throws

Materializes this tensor view using a specific materialization strategy.

This method allows explicit control over the materialization process by selecting a specific strategy. Different strategies provide different trade-offs between memory usage, performance, and access patterns.

Available Strategies

• CopyMaterializationStrategy: Immediate copying with full memory allocation

• LazyMaterializationStrategy: Deferred copying with sparse element caching

Usage Examples

val view = parentTensor.sliceView { /* slicing definition */}

// Use lazy materialization for sparse access patterns
val lazy = view.materialize(LazyMaterializationStrategy())

// Use copy materialization for frequent access
val copy = view.materialize(CopyMaterializationStrategy())

Parameters

Throws