Saturday, May 12, 2012
Judea Pearl
Judea Pearl is the newest winner of the Turing Award.
Judea Pearl's studies and advancements range from in artificial intelligence to philosophy
I recently listened to 1hr 41min interview by Stephen Ibaraki on behalf of ACM here are the highlights
Education:
Counterfactuals:
Success:
Friday, May 11, 2012
3: First Order Recurrence Relations
Check out the project description here
First Order what?
First order recurrence is a type of a recursive equation that can be . The following equations have first order recurrence relations:
The Code
Due to the fact that my professor usually gives us read-in-files contain similar formats I decided to make a master parsing class that you can look at on github if you want.
First Order what?
First order recurrence is a type of a recursive equation that can be . The following equations have first order recurrence relations:
- The coefficients are constants and only to the first power
- The nth term depends only on term ( n - 1 )
- General Form: S(n) = cS( n - 1 ) + g(n)
- S(n) = Cn-1S(1) + nΣi=2 [ Cn-i * g(n) ]
- Original: S(n) = 2*S( n - 1 ) + 3
- Becomes: S(n) = 2n-1 * S(1) + nΣi=2 [ 2n-i * 3 ]
The Code
Due to the fact that my professor usually gives us read-in-files contain similar formats I decided to make a master parsing class that you can look at on github if you want.
- The code doesn't contain anything fancy.
- I love ruby's exponent operator **
- I love ruby's for loops
Monday, May 7, 2012
2: Second Order Recurrence Relations
Second Order What?
Second Order Recurrence Relations is a fancy way of saying, if recursive equation meets a specific criteria, it can be reduced to a no recursive equation.
Requirements:
Example:
S(n) = 2S( n - 1 ) + 3S( n - 2 )
S(1) = 3
S(2) = 1
Second Order Recurrence Relations is a fancy way of saying, if recursive equation meets a specific criteria, it can be reduced to a no recursive equation.
Requirements:
- The nth term depends on the two previous terms
- Constant coefficients with exponents no greater than 1
- Must be homogeneous ( g(n) == 0 )
- General form: S(n) = C1S( n - 1 ) + C2S( n - 2 )
- Find the roots of t2 - C1t - C2 = 0
- r1 & r2
- Solve for p & q
- S(1) = p + q
- S(2) = p( r1 ) + q( r2 )
- Plug answers in the solution formula
- S(n) = p( r1 )n - 1 + q( r2 )n - 1
Example:
S(n) = 2S( n - 1 ) + 3S( n - 2 )
S(1) = 3
S(2) = 1
- t2 - 2t - 3 = 0
- r1 = 3
- r2 = -1
- Solve for p & q
- 3 = q + p
- p( 3 ) + q( -1 ) = 1
- p = 1
- q = 2
- Substitute
- S(n) = 1( 3 )n-1 + 2( -1 )n-1
- line 10: Love how you can simultaneously assign variables from an array
- Line 36: This isn't a true quadratic formula. It takes advantage of the fact that a == 1in all cases
- I added a master parse file
- I love the ||= assignment, this is amazing. It will only assign the variable if it doesn't exist
Subscribe to:
Posts (Atom)