3/30/13 Homework | VLSI CAD: Logic to Layout

Problem Set #1
Warning: You have already made the maximum number of submissions. Additional
submissions will not count for credit. You are welcome to try it as a learning exercise.

Question 1
Let Boolean function F (x, y, z, w) = (xy + x̄z) ⊕ w . Which of the following
equations involving Shannon Cofactors are correct?

Select all the correct solutions. There may be more than one correct solution.

¯ ¯
(Reminder: a ⊕ b = a b + ab ; think carefully about what equations like
(Boolean stuff)⊕1 and (Boolean stuff)⊕0 can simplify to.)

F x̄ȳ = w

F xy = w̄

¯¯¯¯¯¯¯¯¯¯¯
F w = xy + x̄z

Fx = y ⊕ w

F x̄ = z̄ ⊕ w

Question 2
There are other ways of representing the Shannon Expansion theorem. The
version we gave you –
¯i ∙ F (xi = 0)
F (x1 , … xi , … xn ) = xi ∙ F ( xi = 1) + x can be thought of

as a “sum of products” form, since the equation is an OR (sum) of two small AND

(product) terms. But there must be a "product of sums" form for the Shannon

expansion. Use Boolean algebra, and the basic properties of cofactors, and tell us

which of these equations is a correct alternate form of the Shannon expansion.

https://class.coursera.org/vlsicad-001/quiz/attempt?quiz_id=8 1/9
3/30/13 Homework | VLSI CAD: Logic to Layout

There may be more than one correct answer; if so, select each correct answer.

¯i + F (xi = 0)]
[xi + F (xi = 1)] ∙ [x

¯i + F (xi = 0)]
[xi + F (xi = 1) + x

¯i ] ∙ [F (xi = 0) + F (xi = 1)]

[xi + x

¯i + F (xi = 1)]
[xi + F (xi = 0)] ∙ [x

Question 3
Consider the small logic network shown below. Use the definition of the Boolean
Difference and determine which of these equations is correct.

There may be more than one correct answer; select all correct equations.

(Reminder: p ⊕ q ¯ + p̄ q
= pq . Think: p ⊕ p = what? p ⊕ p̄ = what?)

∂f
¯
= ā ⋅ b
∂x

∂f
¯¯¯
= ab
∂x

∂f
= 1
∂y

∂f ¯ ¯
= a + b + x
∂y

∂f
= 0
∂y

https://class.coursera.org/vlsicad-001/quiz/attempt?quiz_id=8 2/9
3/30/13 Homework | VLSI CAD: Logic to Layout

Question 4
Looking at the same gate network from Problem 3, which are correct Boolean
formulas for these quantification operations on F?

(∃x, f )[a, b, y] = 0

¯ ¯
(∀x, f )[a, b, y] = (a + b ) ⋅ y

(∀x, f )[a, b, y] = a + b

¯ ¯ ¯
(∃x, f )[a, b, y] = a + b + y

(∃x, f )[a, b, y] = 1

Question 5
Network repair

The carry output of a 1-bit adder has the Boolean equation

C out = a ⋅ b + (a + b) ⋅ C in , where a and b are the 1-bit numbers we want
to add, and C in is the input carry bit. Suppose we know that we have
implemented the gate-level logic incorrectly, the gate with the “??” label is the one
we suspect is incorrect.

Do this: use the logic network repair procedure discussed in lecture; replace the
“??” gate with the 4:1 multiplexor shown below, with inputs a ,b connected to the
,
s1 s0 select lines. (So, for example, s1s0 = 10 means q = d2 for this mux.)
Use the quantification procedure, and do the Boolean algebra to see what the
result tells you about how to fix it.

Select the right answer below.

https://class.coursera.org/vlsicad-001/quiz/attempt?quiz_id=8 3/9
3/30/13 Homework | VLSI CAD: Logic to Layout

There are two feasible repairs: replace “??” with either an OR or an EXOR
gate.

It’s a trick! There is NO feasible repair of this network!

The only feasible repair is to replace the “??” with a NOR gate.

The only feasible repair is replace the “??” with an EXOR gate.

The only feasible repair is to replace the “??” with an OR gate.

Question 6
Using quantification for finite state machine (FSM) analysis.

It is very natural to think about representing a finite state machine using some
kind of a graph: the nodes represent the states, and the edges represent the
transitions from state to state under input changes. The common state diagram
notation (see diagram) is exactly this kind of a graph:

This machine has four states AA BB CC DD, represented by 2 bits of state (2 flip
flops) whose values appear in brackets (e.g., [11] represents state CC) in each
state bubble. This machine has just 1 input x. We are ignoring any outputs for

https://class.coursera.org/vlsicad-001/quiz/attempt?quiz_id=8 4/9
3/30/13 Homework | VLSI CAD: Logic to Layout

this machine.

Unfortunately, these graphs can get very big, very easily. Imagine a state
machine with 20 bits of state (20 separate flip flops). It has 220 states ∼
1,000,000 states. Suppose also there are 20 input variables. Then each of these
1,000,000 states has 1,000,000 transition arrows leaving it. Our little machine
with 20 flip flops has one trillion state transitions!

It turns out there is a more elegant way of representing things here. Imagine that
we create a new Boolean function, called R, which we call the state transition
relation function. In general R has 3 kinds of inputs:

R(state variables for starting state, input variables, state variables for ending

state)

The way to think about the R function is that it answers a simple question: can I
get from a specific starting state to a specified ending state via specified input
value? For our little example, here are a few values of R:

R(AA, 0, BB) = 1 means “Yes, from state AA, an input x=0 takes you to state

BB”

R(BB, 1, AA) = 0 means “No, from state BB, input x=1 does NOT take you to

AA”

Of course, you can’t just input the state “names”, you have to use the state

assignment bits that represent each state. This means that R is a function of 5
variables for our little example:

R( S1, S0, x, E1,E0 ),

where S1 S0 is the state assignment for the starting state, and E1 E0 is the

assignment for the ending state. So, for example:

R(AA, 0, BB) = 1 really means that R(0, 0, 0, 0, 1) = 1, since state AA is

represented by the assignment S1 S0 = 00, and similarly the state BB has the

assignment 01.

You could easily build a truth table for R( ) for this little 4-state example, and

https://class.coursera.org/vlsicad-001/quiz/attempt?quiz_id=8 5/9
3/30/13 Homework | VLSI CAD: Logic to Layout

create a sum of products Boolean expression for R(S1, S0, x, E1, E0 ) using a 5-
variable Kmap, or just ordinary Boolean algebra. All the information you need to
build R( ) is available in our state machine design.

Now, let’s create a more interesting function, called G(E1, E0). G is again a
function of the states, but it answers a different question: is there ANY way to
reach this state E1 E0? For example, it is clear from our little state diagram that
G(CC) = 0, i.e., there is just NO way to start from some state and take a
transition that gets you to state CC. But G(AA)=1 because it is possible to get to
state AA (for example, by being in state CC and taking the x=1 transition).

For a trivial example like our little 4-state machine, you can just look at the right
answer. For a real machine, with perhaps billions of states+edges, you cannot
just look at the graph. So, answer this: Can we use computational Boolean
algebra and transform the R( ) function – which we can build – in the G( )
function, directly?

Select the correct answers below.

Yes, we can compute this as (∀S1S0x, R)[E1, E0] , because we need to

know “for all” end states (E1,E0) how the start state (S1,S0) and input
transition allow us to reach this end state.
∂R
Yes, we can compute this as ∂S1∂S0∂x
[E1, E0] , since we want to know the
sensitivity of the end state E1,E0 to the start state S1S0 and input x.

Yes, we can compute this as (∃S1S0x, R)[E1, E0] because all we really
need to know is that “there exists” some start state (S1S0) and some transition
(x) that does lead to this end state (E1E0), and we don’t need to know what
those are, just that “they exist”.

No, it’s a trick! There is no way to use the computational Boolean algebra
methods we just learned to do this!

Question 7
Unate functions

https://class.coursera.org/vlsicad-001/quiz/attempt?quiz_id=8 6/9
3/30/13 Homework | VLSI CAD: Logic to Layout

Consider this Boolean function F (x, y, z, w) = (xyw̄ + ȳ z̄ + xzw̄ + yzw̄)

Which of these are true? Select all correct answers.

F is positive unate in variables (x, y ) and negative unate in variables (w, z )

F is unate in all variables

F is binate in variables (y, z ) and unate in all other variables

F is positive unate in variable (y ) and negative unate in variable (z )

F is positive unate in variable (x) and negative unate in variable (w)

Question 8
This question has no correct answer
See this thread on the forum for more context.

Thus, we are going to nullify this question. Don't worry: if you have submitted your
quiz before (with whatever answer choice), you will receive full points for this
part. You might not see the updated score right away but we will issue a regrade
and all will be well again.

What you should do, in place of this problem, is take Problem Set #1a that will
have a fixed version of this problem. We are leaving the question as-is here just
for your reference.

Positional Cube Notation and manipulations.

Consider this Boolean equation F (x, y, z, w) = xyz̄ w̄ + y + x̄ȳ z̄ w + xz̄ w̄

If we transform this into PCN form, where each cube is [xx yy zz ww], a set of 2-
bit entries for each variable in order, it will become the following list:

[01 01 10 10], [11 01 11 11], [10 10 10 01], [01 11 10 10]

Which of the following is true, starting with this PCN list, using the methods and
rules from URP tautology lecture?

https://class.coursera.org/vlsicad-001/quiz/attempt?quiz_id=8 7/9
3/30/13 Homework | VLSI CAD: Logic to Layout

F( w = 1 ) = [11 01 11 01], [10 10 10 11] and this cofactor is binate.

F( z = 1 ) = [11 01 01 11]

F( x = 0 ) = [10 01 11 11], [11 10 10 01] and this cofactor is unate.

F( y = 1) = [01 11 10 10], [11 11 11 11], [01 01 10 10] and this cofactor is

unate.

F( w = 0 ) = [11 01 11 10], [10 10 10 11]

F( y = 0 ) = [01 11 10 10], [11 11 11 11], [10 10 10 01], [01 11 10 10]

F( z = 0 ) = [01 01 11 10], [11 01 10 11], [10 10 11 01], [01 11 11 10] and

this cofactor is unate.

F( x = 1 ) = [11 01 10 10], [01 01 11 11], [11 11 10 10]

Question 9
Unate Recursive Paradigm algorithm

Suppose we have these cubes at one node of our URP tautology recursion, and
we need to decide on the splitting variable to use to cofactor and recurse. Which
variable do we pick, and why? Select the right answer.

Variable y is the most binate variable since it appears in 5 cubes.

Variable x is the only binate variable, and so we select it.

Variable z is the most binate variable since it appears in 5 cubes.

Variables y and z are tied for most-binate, since they each appear in 5 cubes,
but we select y because it has the smallest value of |True cubes –
Complement cubes|. This value is 0 for variable y and 1 for variable z , so we
pick y .

https://class.coursera.org/vlsicad-001/quiz/attempt?quiz_id=8 8/9
3/30/13 Homework | VLSI CAD: Logic to Layout

Question 10
Unate Recursive Paradigm.

Consider this new cube list over variables x, y, z , which appears at one node of
our URP tautology algorithm. We select variable x to split on. Let us consider
what happens when we recurse, and call Tautology on the positive x cofactor.

Which of the following are true statements for the cubelist that should appear
inside the f(x=1) recursive call, highlighted in our figure?

The cube list is [01 11 01], [01 11 10], [01 10 11] and it is not a tautology.

The cube list is [11 11 01], [11 11 10], [11 10 11] and it is a tautology

The cube list is [01 11 01], [01 11 10], [01 10 11] and we cannot tell if it is
tautology, we must recurse.

The cube list is [11 11 01], [11 11 10], [11 10 11] and it is not a tautology

The cube list is [11 11 01], [11 11 10], [11 10 11] and we cannot tell if it is a
tautology and we must recurse some more

In accordance with the Honor Code, I certify that my answers here are my own
work.

Submit Answers Save Answers

https://class.coursera.org/vlsicad-001/quiz/attempt?quiz_id=8 9/9