First few thoughts:

• Free will is an illusion
• The concept does not need to exist, I guess?
• This becomes evident when we look at our actions and our inner monologue or what goes on in our mind from a 3rd POV and make a note.
• We can purely look at actions, responsibilities, and intentions without even considering free will and still have accountability without any issues

My Questions: - How does this tie to Karma and Dharma?

This rationale is rooted in a utilitarian perspective, which emphasizes actions that maximize overall happiness and well-being while minimizing harm and suffering.

when considering responsibility for actions, we often look beyond the immediate actions to the broader context of the individual’s mental state, background, and the intentions behind their actions.

we might need to pretend we have free will to maintain motivation and social order. Instead, it asserts that recognizing our real agenthood, even within a deterministic framework, is both truthful and sufficient for maintaining a useful sense of personal agency and responsibility.

while consciousness is significant, it may not have a causal impact beyond the physical processes of the brain.

Determinism: The view that all events, including human actions, are caused by previous events in accordance with the laws of nature, making free will illusory.

Libertarianism: This philosophy asserts that humans have free will and can make genuinely free choices that are not determined by previous events or causal laws. This often involves the belief in metaphysical concepts such as a soul that transcends physical causation.

Compatibilism: The belief that free will and determinism are compatible. It argues that freedom can exist even if our actions are determined by prior causes, typically by redefining free will to fit within deterministic constraints.

Example: Do not think of a pink elephant

Circling back, when \(t \rightarrow 0\) , \(freedom \rightarrow 0\)

We do not know what influences the influences behind our intentions to carry out the actions.

Another influence \( \rightarrow \) influence?

The connection between free will and the conscious-sub-conscious mind needs to be explored.

You are on the verge or falling asleep but you keep yourself awake. Is this a good example?

It seems to me that we have more control over our shaping of behaviors and habits than we have in our actions (free will). So, since we can only process one thought / perform one action at a time, we are training our habitual state of being with each sample.

It might also be the case that this is a way of saving cognitive load and saving energy for something more important at the time. I'd like to use survival as an explanation here.

"we could have made alternative decisions or taken different paths than the ones we actually took."

"we believe that we are the ones actively generating and controlling them. It suggests a sense of agency over our current mental processes and behaviors."

Latency: The signal travel time from your device to the server. This includes the time taken for your request to pass through various network nodes and arrive at the server. If this takes 100 milliseconds, then the latency is 100 milliseconds.

Response Time: The total time from when you click a link to when the webpage is fully loaded on your device. This includes the 100 milliseconds of latency, the time the server takes to process the request (say, 200 milliseconds to retrieve data and render the webpage), and the time to send the data back to your device (another 100 milliseconds). If these are the involved times, the total response time would be 400 milliseconds.

Hashing: Both the nodes and the data items are hashed onto a hash ring, a conceptual circle representing the hash space. The hashing function typically generates a fixed range of values (e.g., 0-255 for an 8-bit hash).

Placement on the Ring: Each node and data item is assigned a position on this hash ring based on its hash value.

Data Lookup: To find where a data item should be stored, you move clockwise around the ring starting from the hash value of that item until you find the first node. This node is where the data item is stored.

Scalability and Flexibility: When a new node is added, it gets placed into its position on the ring based on its hash value. It only takes over the responsibility for data items that are closest to it on the ring. Similarly, when a node is removed, only its immediate neighbors on the ring take over serving its data items. This limits the amount of data that needs to be moved around, which is a key advantage over traditional hashing, where adding or removing a node might reshuffle a large portion of the data.

Load Balancing: Consistent hashing tends to naturally balance the load because each node gets approximately the same range of hash values. Virtual nodes, or "vnodes," can also be used where each physical node is represented by multiple points on the hash ring. This technique can help balance the workload more evenly among nodes.

SSTables (Sorted Strings Tables)

Immutable: Once an SSTable is written and closed, it cannot be modified. This immutability helps in reducing the complexity of data management and ensures consistency.

Sorted: The entries in an SSTable are stored sorted by key. This sorting is beneficial for read performance, as it allows for efficient range queries.

Consider an example based on a social media application where we manage posts made by users. We'll use Apache Cassandra and its hybrid partitioning strategy, involving a compound primary key that includes a partition key and clustering columns.

Setup

• Objective: Store and retrieve user posts efficiently.
• Table Name: user_posts
• Schema:
• user_id: The unique identifier for a user (Partition Key)
• post_timestamp: Timestamp of when the post was made (Clustering Column)
• post_content: Content of the post

Cassandra Table Creation

Here’s an example of how you might define this table in Cassandra:

cql CREATE TABLE user_posts ( user_id uuid, post_timestamp timestamp, post_content text, PRIMARY KEY (user_id, post_timestamp) );

Data Distribution and Storage

• Partitioning: user_id is used as the partition key. All posts by a single user are stored together in the same partition. This is beneficial because it localizes all data related to a single user to one partition, making data management and retrieval efficient.
• Ordering within a Partition: post_timestamp is used as the clustering column. Within a partition, posts are automatically sorted by timestamp. This ordering supports efficient queries to fetch posts over a range of times.

Example Data

Assuming the following posts:

1. User 1 posts at 10:00 AM - "Good morning!"
2. User 1 posts at 10:05 AM - "Another post!"
3. User 2 posts at 09:50 AM - "Hello there!"

These posts would be stored as follows:

• Partition for User 1
• (10:00 AM, "Good morning!")
• (10:05 AM, "Another post!")
• Partition for User 2
• (09:50 AM, "Hello there!")

Example Queries

1. Fetch all posts by User 1: cql SELECT * FROM user_posts WHERE user_id = <uuid_of_user_1>; This query will efficiently return all posts by User 1, sorted by the timestamp.

2. Fetch posts by User 1 within a specific timeframe: cql SELECT * FROM user_posts WHERE user_id = <uuid_of_user_1> AND post_timestamp >= '2024-04-11 10:00' AND post_timestamp <= '2024-04-11 11:00'; This query utilizes the order provided by the clustering column to efficiently retrieve posts made by User 1 between 10:00 and 11:00 on April 11, 2024.

In summary, setting up a new follower in a database system involves creating a consistent snapshot of the leader's data, copying this snapshot to the follower, and then having the follower catch up with the leader by processing all the data changes that occurred since the snapshot was taken. This approach ensures data consistency without the need to lock the entire database, thus maintaining high availability.

A materialized view is different from a regular view in that it is a physical copy of the data, while a view is a virtual table that does not store data. When a query is run on a view, it retrieves the data from the original tables, whereas a materialized view stores the results of the query, allowing for faster data retrieval. Materialized views are particularly useful when data needs to be accessed frequently and the underlying data does not change often, as they can improve performance by providing precomputed results.

**Permutations of a string **

image processing techniques that alter the colors of an image in a non-linear manner.

. For example, increasing the contrast of an image can make the darker colors darker and the brighter colors brighter, which can change the perceived illumination color. Similarly, increasing the saturation of an image can make the colors more vivid, which can also change the perceived illumination color. Therefore, it is important to use linear color space when performing the color constancy task

Photometric transformations are operations that change the brightness, contrast, color, or intensity of an image, without affecting its geometry

So light can impact color constancy

When you design by the interface, code duplication is most likely. To avoid this, we use abstract classes which - Implements the interface - This the base case for other class that used to implement this interface - Can contain methods with/without definition that the child classes can reuse

high-frequency information refers to the rate of change of intensity values from one pixel to the next. A high-frequency image is one where the intensity values change quickly from one pixel to the next, such as edges in an image.

General Comments - Not the first transformer + SISR - Not SOTA - A more efficient approach: the best trade-off between model performance and computation cost.

Critique - Choice of metrics, subjective. Lack of other comparison with other metrics. - Does this compare to the state-of-the-art? Why only compare it to the lightweight ones? How will I know what I am trading off for this?

How doe IPI vary within each coda click?