EBoard 34: Hash tables, continued

Warning This class is being recorded (and transcribed) (assuming Teams succeeds).

Approximate overview

• Administrivia
• Questions
• About MP9
• About hash tables
• Lab

Administrivia

• Please return cards, boards, and markers to the back of the room when you finish class today.
• Sorry for missing on Wednesday.
• Happy Friday!
• Technology hates me today.

Upcoming Token activities

Academic

• Tuesday, 2023-11-21, Noon, Day PDR, CS Table.

Cultural

Peer

• Language study! Talk to your colleague.
• Swim meet, Saturday at 1pm (we think).
• Tuesday, 2023-11-21, 4-6pm, 3rd floor HSSC, somewhere: Wilson Catalyst

Wellness

• We need more of these!

Misc

• Subject yourself to a study of types.
• Please fill in the peer educator evaluations

Other good things (no tokens)

Upcoming work

• MP9 assigned today (JSON)
• MP9 pre-assessment due Sunday

Friday PSA

• Please be moderate
• Consent is essential

Questions

Hash tables

Administrative

Are you ever going to talk to the graders?

Yes. Monday, I hope.

Is MP9 our last mini-project?

Yes.

When is it due?

Thursday, November 30

Can we have more time for redos?

Sure.

Why is the Sample LA there?

As a reminder to Sam to Please LAs expunge

Are you going to charge for late redos?

No.

Can I use tokens to turn in pre- and post-assessments really late?

No. The point of these is to reflect at the time you do the assignment.

MP9

Goal: Parse JSON into an appropriate Java structure.

Input { "a": "apple", "e": [2,7,1,8], "q": { "x" : "xerox } }

Output: JSONHash with three KVPairs.

• The first KVPair has a key of the JSONString “a” and a value of the JSONString “apple”.
• The second KVPair has a key of the JSONString “e” and a value of a JSONArray of length four
  • Element 0 of the JSONArray is the integer 2
  • Element 1 of the JSONArray is the integer 7
  • …
• The third KVPair has a key of “q” and a value of another JSONHash

We looked at the provided code together. Whee!

Hash table review

TPS: Five key ideas.

Cool. The screen stopped flickering. Have I mentioned that I hate computers? Of course, now my laptop is running out of power, so the recording will stop soon.

• A hash table is something that pairs keys and values. We generally call such things “Dictionaries” or “Maps”. Hash tables are a particular implementatoin of Dictionaries/Maps. In fact, they are popular enough that some people call Dictionaries/Maps “Hashes”.
• Hash tables use arrays to implement dictionaries/maps.
  • Array access is fast. We’d like to use array access when keys are not necessarily numeric.
  • We do so by converting the key to an integer (called hashing) in a consistent, reliable, fast way so that (a) we always get the same hash code for equal values, and (b) we are likely to get different hash codes for different values.
  • We look in the cell based on the hash code (mod table size)
• We can use the arrays to build hash tables in two ways:
  • Chained/bucketed hash tables put a list of values in each cell. If you have a different key that maps to the same cell as an existing key, you add to the list.
  • Probed hash tables may require you to look elsewhere in the hash table. If you have a key that maps to an already filled cell with a different key, you look elsewhere in the table.
• Side note: Badly designed hash functions tend to group things in clumps. For example, “sum the ASCII values” will group length-five strings around 5*109 (the ASCII value for “m”, the middle letter).
• UM: Hash tables use math (in computing good hash values).
• Hash tables are “better” than binary search trees.
  • Arrays are fast.
  • Our hash function tends to distribute things well, but we’ll still have collisions.
  • Are there things we can say about the expected number of collisions?
  • In a well-designed hash table, about half of the cells are empty. On average, we should find an empty cell fairly quickly when probing. [Statistics!]
  • We have to grow the hash table to maintain that property.
• As you will eventually learn, things that are nearby in memory perform better in real life, so hash tables (and all array-based structures) generally perform better than linked structures.
• Hashing is so important that Java expects each object to provide a well-designed hashCode method.
  • See the Osera chapter for designing a good method.

Lab

Make sure you share contact info with your HW9 partners.

Schedule a meeting with your HW9 partners.

Continue working with probed hash tables using the lab from last class.