VIEWS: 15 PAGES: 22 CATEGORY: Childrens Literature POSTED ON: 12/1/2009 Public Domain
Wild Circuits Investigating the Limits of MIN/MAX/AVG Circuits Brendan Juba Faculty Advisor: Manuel Blum Graduate Mentor: Ryan Williams Definitions: MIN/MAX/AVG Circuits unsatisfied satisfied We are given a circuit, C, with feedback, operating on real numbers 1 0 from the closed interval [0,1]. C contains 0 MIN MIN, MAX, or AVG gates with two inputs “Inputs” to the circuit that are hard-wired to AVG either 0 or 1. 0 |C| denotes the number of gates of C MAX Here, |C| = 3 When the output of a gate is the 0 appropriate function of its inputs, we say that the gate is satisfied satisfied Definitions: MIN/MAX/AVG Circuits Settings of the gate outputs from the interval [0,1] are value vectors 1 0 A value vector for C, v [0,1]|C| The ith entry, vi, is the output of the ith gate. We may also consider an update function, MIN F: [0,1]|C| [0,1]|C| We are interested in two varieties: single-gate AVG update functions and circuit-wide update functions: MAX A single gate update function replaces the output of a single designated gate with the correct output value. • We will call iterating over the single gate update functions “gate-by-gate update” The circuit-wide update function simultaneously replaces the output of every gate with a value that is correct with respect to the old values Definition: Stable Circuit Problem A vector v is stable iff every gate is satisfied. (F(v) = v) We wish to find these stable vectors. Gate-by-gate update from 0 clearly obtains a stable vector in the limit. This is the minimum stable solution 1 0 1 0 stable unstable MIN MIN 0 0 AVG AVG 1/2 1/2 MAX MAX 1/2 0 Stable Circuit Decision Problem We designate some ith gate of C in the minimum stable solution, s, and ask, “is si ≥ 1/2?” We can reduce the function problem to this decision problem. We can find 2|C| bits of any si, which may be shown to be sufficient. Suppose there is a gate xk: sxk ≥ 1/2 iff the kth bit of si is a 1 If sxk ≥ 1/2, we assume its binary decimal representation is .100…0 with 0s in positions 2 through 2k-1 Otherwise, it is .011…1 with 1s in positions 2 through 2k-1 Depending on whether sxk ≥ 1/2, we add a new gate, xk+1: AVG(xk,1/2-1/22k), if sxk = .100…0 AVG(xk,1/2+1/22k), if sxk = .011…1 In the former case, sxk+1=(.1)(.100…0 + .011…11) In the latter, sxk+1=(.1)(.011…1 + .100…01) In the solution for the modified circuit, xk+1 clearly has the desired properties. The largest construction is O(|C|2) gates. Unique Solution Circuits x 0 Replace any wire from x to y in the circuit with the construction on the right using m AVG gates AVG This circuit has a unique solution (Shapley, 1953) Suppose the original circuit-wide update AVG function is F, stable solutions are u and v If ||u-v||∞= c, then it is easy to see c = ||u-v|| = (1-1/2m)||F(u)-F(v)|| ≤ (1-1/2m)c Clearly, c = 0. AVG These solutions turn out to be arbitrarily close to the minimum y stable solutions (for appropriate m). Stable Circuit is in NPco-NP (Condon, 1992) A nondeterministic machine M can, in polynomial time, on input circuit C, for gate i Build a suitably close unique solution circuit C’ Guess the solution to C’ Verify the guessed vector is a solution Accept or reject, respectively, precisely when the value of gate i is above 1/2 (since C’ was close to C, either i is above 1/2 in neither, or it is above 1/2 in both) Stable Circuit is P-hard If we use no AVG gates, the wires of the circuit will only carry 0 or 1 It is immediate that we may use MIN as AND, MAX as OR For any circuit with fixed inputs, we can construct a “complement” circuit Switch 0 inputs with 1 inputs Switch MIN gates with MAX gates We can now negate by crossing a wire between the original and complement circuits (In this AVG-free case, deciding the output is in P, too) Observation and Motivation If we apply gate-by-gate or circuit-wide update on arbitrary starting value vectors, we can obtain “interesting” circuits One such “interesting” circuit is a binary counter We do not necessarily obtain the stable configurations of our circuits this way -- this is not Stable Circuit Can we obtain such circuits under gate-by-gate update from 0? If so, the minimum stable solution is the configuration of the device after an unbounded amount of time! “Leapfrog” circuits We assign each wire a “threshold” wire and interpret its value relative to that threshold Above threshold: T Below threshold: F It is already clear that we still have AND and OR There is also a construction for NOT (next slide) If there are W wires which we wish to interpret relative to the same threshold, this gadget takes Θ(W) gates NB: The circuits are still monotone! As we update, a value may seem to rise or fall, as we follow it across different wires through the circuit The value on any particular wire only rises as the gates of the circuit are updated NOT Gadget th x0 x1 x2 AVG MAX MIN MIN MIN AVG MAX MAX ~x0 x1 x2 th x0 x1 x2 th Caveats Assumptions: 1. All values above [below] threshold are equal 2. The values th, T, and F are all different 3. We may specify the update order for the gates of the circuit Take each in turn: 1. Everything starts from zero, the property is preserved by all three gates 2. We can push th above zero by means of an AVG gate With feedback, we must pass the other wires to be interpreted relative to that threshold through similar constructions so as to maintain relative values 3. Update order doesn’t change the solution we approach Two-bit Counter Circuit 1 AVG 1 x0 x1 th NOT NOT MIN MIN MAX 1 1 AVG AVG 0 x0 x1 th Two-bit Counter Circuit 1 AVG 17/32 x0 x1 th NOT NOT MIN MIN MAX 1 1 AVG AVG 1/2 x0 x1 th Two-bit Counter Circuit 1 AVG 781/ 1024 x0 x1 th NOT NOT MIN MIN MAX 1 1 AVG AVG 195/ 256 x0 x1 th Two-bit Counter Circuit 1 7217/ AVG 8192 x0 x1 th NOT NOT MIN MIN MAX 1 1 AVG AVG 28867/ 32768 x0 x1 th Serving Suggestions carry-in The counter generalizes to xi n bits easily NOT The n-bit counter takes Θ(n2) NOT gates, due to the size of the NOT gadgets MIN MIN MIN carry- We may also build a MAX out gadget such that, if its xi input is ever above threshold, a wire in the MAX 1 input gadget stays above AVG threshold forever: (the internal wire must also pass through the NOT gadgets) NP-hardness x1 x2 x3 th, etc. Let any boolean formula be given… Ex: (x1~x2x3) (~x1~x2x3) NOT Since we have AND, OR, and NOT NOT gates, we can easily translate any formula into a circuit which has an MAX MAX output above its threshold iff the formula is satisfied by the MAX MAX assignment from the input wires, as we have on the right. MIN (x1~x2x3)(~x1~x2x3) By attaching xi to the ith bit of the counter, we try all possible assignments, allowing us to encode answers to SAT on a wire. The number of gates in this circuit is quadratic in the length of the formula. Entering PSPACE We can still do better: using the counter, we will decide whether quantified boolean formulas are x1 valid Assume WLOG that the quantifiers alternate: odd variables are universal, even existential Observe that the counter “walks” along the x0 x0 leaves of a tree of assignments, left to right. Suppose that at the bottom we evaluate the quantifier-free part of the formula on the specified 00 01 10 11 assignment. Now suppose at every level of the tree, we have one bit of memory for the left branch Set it to T when the branch is T, reset it to F when leaving that subtree. Pass T up the tree when We see T at either branch at an level We see T at the right branch of a level with the left branch bit already set to T. T is passed up from the top of the tree iff we have a TQBF. Quantifier Circuit: xi xi-1 A xi Carry-out: xi v i0 A • IH: the wire labeled A will carry T iff the shorter boolean formula with alternating NOT quantifiers, A, is satisfied by the current assignment to xi-1,…,xn from the counter MIN • vi0 is a register that holds the evaluation of NOT xi-1A when xi is F, while xi+1,…,xn remain fixed • When there is a carry out of xi, xi+1 has altered, (new branch) so we reset vi0 to F MIN MIN • If vi0 holds T, and when xi is T, for some xi-1 A carries T, then the wire labeled xixi-1A MAX carries T. Otherwise, the wire remains F. MIN • Observe that the wire xixi-1A will carry T iff xi v i0 xixi-1A is satisfied by the current xixi-1A assignment to xi+1,…,xn, so the IH is satisfied End of the Line: Thwarted by PSPACE In the limit, the separation between T and F shrinks as the internal wires approach 1. It is not immediately clear how to recover the values of any wire from the minimum stable solution Recall: finding values in the limit (the minimum stable solution) is known to be in NPco-NP Answers to PSPACE-hard problems (QSAT) may be encoded on the wires as we update Since the “space” in Leapfrog circuits is bounded by the number of gates, it is doubtful that such circuits can solve anything harder In the limit, it is impossible to distinguish the values in Leapfrog circuits unless NP = Stoppable NOT Gadget th x0 check x1 x2 This gadget behaves AVG identically to the regular NOT, unless check is set MAX high, in which case, all outputs are set high. MIN MIN MIN MIN Gadgets such as this MAX suggest that the problem AVG with our Leapfrog counter MAX MAX MAX was in the AVG gates we th check x1 x2 used to “power” it from 0. ~x0 GAME OVER Thank you for playing CAPCOM