Tensors for Beginners 14: Tensors are general vector/covector combinations

eigenchris
24 Feb 201808:50
EducationalLearning
32 Likes 10 Comments

TLDRThe video delves into the concept of general tensors, explaining how they are constructed from combinations of vectors and covectors using the tensor product. The speaker introduces new types of tensors, explores transformation rules, multiplication formulas, and array shapes, and emphasizes the complexities involved as tensors become larger. The importance of Einstein notation for simplifying tensor operations is highlighted, particularly for high-rank tensors where array representation becomes cumbersome. Ultimately, tensors are defined as collections of vectors and covectors, with their properties and transformations determined by their basis.

Takeaways
  • πŸ“š The video discusses generalizing the concept of the tensor product to create general tensors using combinations of vectors and covectors.
  • πŸ”„ The script introduces a non-standard tensor product notation, different from the circle-times symbol typically found in textbooks.
  • πŸ—Ί Linear maps are explained as linear combinations of vector-covector pairs, which simplifies understanding transformation rules, multiplication formulas, and array shapes.
  • πŸ“ˆ Bilinear forms are also described as linear combinations of covector-covector pairs, following a similar pattern to linear maps for ease of understanding.
  • πŸ†• The video introduces new tensors, 'D' and 'Q', with 'D' being a (2,0)-tensor made of vector pairs and 'Q' a (1,2)-tensor with one vector and two covector parts.
  • ❓ It's highlighted that there isn't a single way for tensor 'Q' to act on tensor 'D', emphasizing the complexity of higher-dimensional tensors.
  • πŸ”§ The transformation rules for tensors are simplified by using linear combinations and the transformation rules for basis vectors and covectors.
  • 🧩 The script explains that tensor components can be visualized as arrays, similar to the Kronecker product, which helps in understanding their shapes and multiplication.
  • πŸ“‰ For high type tensors, the array representation becomes less useful due to the multiple possible multiplication rules, making Einstein notation more preferable.
  • πŸ“ The importance of basis choice in tensor representation is emphasized, noting that different bases can yield different components for the same tensor.
  • πŸ”‘ The script concludes by defining tensors as collections of vectors and covectors combined using the tensor product, and the importance of understanding transformation rules and Einstein notation for complex tensors.
Q & A
  • What is the main topic of the video?

    -The video discusses the generalization of tensor products to general tensors, which can be made using any combination of vectors and covectors.

  • What is the significance of the tensor product in the context of tensors?

    -The tensor product is crucial for combining vectors and covectors to create tensors of various types, which are then used to define linear maps, bilinear forms, and more complex tensors.

  • How are linear maps related to tensors?

    -Linear maps can be understood as linear combinations of vector-covector pairs, and this perspective helps in deriving transformation rules, multiplication formulas, and array shapes for tensors.

  • What is a (2,0)-tensor and how is it constructed?

    -A (2,0)-tensor, like tensor D mentioned in the video, is constructed from pairs of vectors, indicating it has two contravariant parts and no covariant parts.

  • What are the transformation rules for tensors?

    -Transformation rules for tensors involve applying the transformation rules for basis vectors and covectors. For example, a (2,0)-tensor would require applying the vector basis transformation twice.

  • Why is the multiplication of larger tensors, like Q and D, ambiguous?

    -The multiplication of larger tensors is ambiguous because there are multiple ways to pass input vectors to covectors, leading to different possible results for the same tensor product.

  • What is the advantage of using Einstein notation for tensor multiplication?

    -Einstein notation simplifies tensor multiplication by making the summation rules explicit, which is especially useful for high-type tensors where multiple multiplication paths exist.

  • How can tensor arrays be visualized, and what are the limitations of this visualization?

    -Tensor arrays can be visualized as a Kronecker product of vectors and covectors, resulting in structures like 'columns of columns' or 'rows of rows of columns.' However, visualizing tensors this way can obscure important information about their covariant and contravariant parts.

  • Why might array notation be less useful for high-type tensors?

    -Array notation might be less useful for high-type tensors because the multiple possible multiplication rules make it difficult to perform straightforward array multiplication without ambiguity.

  • What is the best definition of tensors according to the video?

    -Tensors are best defined as collections of vectors and covectors combined using the tensor product, and their transformation rules can always be determined based on this combination.

Outlines
00:00
πŸ”’ Generalizing Tensors and Transformation Rules

This paragraph introduces the concept of general tensors, building on the previous understanding of tensor products. It discusses the generalization to tensors that can be formed using various combinations of vectors and covectors. The explanation highlights the importance of understanding transformation rules, multiplication formulas, and array shapes for these tensors. The speaker introduces new tensors D (a (2,0)-tensor) and Q (a (1,2)-tensor) and emphasizes the ease of determining transformation rules by expressing tensors as linear combinations of basis vectors and covectors.

05:01
🧠 Understanding Tensor Multiplication Ambiguities

This section delves into the complexities of tensor multiplication, particularly when dealing with higher-order tensors like Q and D. Unlike simpler cases where the multiplication is straightforward, the speaker explains that multiple valid methods exist to apply one tensor to another. The example of linear maps versus the tensors Q and D highlights how the number of vectors and covectors introduces multiple possible summation rules, leading to different results. The discussion underscores the ambiguity and the need for careful notation, such as Einstein notation, to specify operations clearly.

πŸ“ Exploring Array Shapes of Tensors

This paragraph examines the array shapes that result from tensor products, using specific examples to illustrate the point. The speaker compares tensor products to Kronecker products of vectors, showing how the shape of the resulting arrays can be visualized. The speaker demonstrates this with the tensors D and Q, noting that while some might prefer visualizing tensors as 3D cubes, it can obscure important information about the tensor's structure. Instead, they advocate for a row of rows of columns approach to maintain clarity on the tensor's type and its vector-covector parts.

πŸ› οΈ Practical Challenges of High-Order Tensors

This final paragraph addresses the practical difficulties of working with high-order tensors, especially when it comes to array representations and tensor multiplication. The speaker explains that while array notation is useful for low-order tensors, it becomes less effective for higher-order tensors like Q and D, where multiple valid multiplication rules exist. As a result, the speaker suggests using Einstein component notation to express multiplication rules more clearly. The discussion concludes by emphasizing the importance of remembering that tensor components depend on the chosen basis and can vary with different bases.

Mindmap
Keywords
πŸ’‘Tensor
A tensor is a mathematical object that generalizes vectors and matrices to higher dimensions. It can be made using any combination of vectors and covectors. In the video, tensors like D (a (2,0)-tensor) and Q (a (1,2)-tensor) are introduced to illustrate how different tensors can be constructed and manipulated, highlighting their transformation rules, multiplication formulas, and array shapes.
πŸ’‘Covector
A covector, also known as a dual vector, is a linear functional that maps a vector to a scalar. In the video, covectors are paired with vectors to create tensors. For example, bilinear forms are described as linear combinations of covector-covector pairs, which helps in deriving transformation rules and multiplication formulas for these tensors.
πŸ’‘Transformation Rules
Transformation rules describe how the components of tensors change when transitioning between different coordinate bases. In the video, these rules are used to show how tensors like D and Q transform, depending on the basis vectors and covectors involved. The ease of transformation is emphasized when using linear combinations in tensor expressions.
πŸ’‘Bilinear Form
A bilinear form is a function that maps two vectors to a scalar, and it is represented as a linear combination of covector-covector pairs. The video explains that bilinear forms are essential for understanding the structure of tensors, and they provide a straightforward way to derive transformation rules and array shapes for these forms.
πŸ’‘Kronecker Product
The Kronecker product is an operation on two matrices that results in a block matrix. It is used in the video to explain the structure of tensors. For instance, the array shape of the tensor D is compared to the Kronecker product of two column vectors, showing how the product forms a column of columns.
πŸ’‘Einstein Notation
Einstein notation, or index notation, is a concise way to represent tensor operations where repeated indices imply summation. The video emphasizes that as tensors become more complex, using Einstein notation becomes more practical for expressing tensor multiplication and summation rules, as it avoids ambiguity in operations like Q acting on D.
πŸ’‘Linear Map
A linear map is a function between two vector spaces that preserves vector addition and scalar multiplication. In the video, linear maps are introduced as linear combinations of vector-covector pairs. The example of a linear map L acting on a vector v is used to show how tensors can act on vectors in a unique and well-defined way.
πŸ’‘Array Shape
The array shape refers to the multi-dimensional structure of a tensor when it is represented as an array of numbers. The video explains the array shapes of tensors like D (a column of columns) and Q (a row of rows of columns), showing how these shapes relate to the type and structure of the tensor.
πŸ’‘Contravariant and Covariant Parts
Contravariant parts refer to the vector components of a tensor, while covariant parts refer to the covector components. The video uses these terms to classify and describe tensors like D (which has two contravariant parts) and Q (which has one contravariant part and two covariant parts), helping to determine their transformation rules and array shapes.
πŸ’‘Tensor Multiplication
Tensor multiplication involves combining tensors in ways that respect their covariant and contravariant parts. The video discusses how, for high-type tensors like Q and D, there can be multiple ways to perform the multiplication, leading to different results. This complexity is why the video suggests using Einstein notation to clearly define the intended operations.
Highlights

Introduction to general tensors made from any combination of vectors and covectors.

Use of non-standard tensor product notation without the circle-times symbol.

Linear maps as linear combinations of vector-covector pairs for deriving transformation rules and multiplication formulas.

Bilinear forms as linear combinations of covector-covector pairs with easy derivation of properties.

Introduction of new tensors D (2,0) and Q (1,2) with unique combinations of vector and covector parts.

Exploration of coordinate transformation rules for tensors using basis vectors and covectors.

Multiplication formula for Q acting on D with multiple possible outcomes due to tensor complexity.

Discussion on the ambiguity of tensor action without specifying summation indices in Einstein notation.

Array shapes for tensors D and Q derived from Kronecker products of vectors and covectors.

Comparison of visualizing tensor arrays as 3D cubes versus rows of rows of columns for clarity.

Mechanical way of component multiplication using standard row-column multiplication rules for simpler tensors.

Limitations of array representation for high type tensors and the preference for Einstein notation.

Abstract notation for tensors as collections of vectors and covectors using tensor products.

Tensor components always come from a choice of basis and can vary with different basis representations.

Importance of understanding tensor transformation rules for both known and new tensors like D and Q.

Practical applications of Einstein component notation for clarity in tensor multiplication formulas.

Simplification of tensor representation focusing on components without basis vectors and covectors.

Transcripts
Rate This

5.0 / 5 (0 votes)

Thanks for rating: