Intersection of two BDDs using CUDD - binary-decision-diagram

I would like to find intersection of two BDDs for the following two Boolean functions:
F=A'B'C'D'=1
G=A XOR B XOR C XOR D=1
Here is my code:
int main (int argc, char *argv[])
{
char filename[30];
DdManager *gbm; /* Global BDD manager. */
gbm = Cudd_Init(0,0,CUDD_UNIQUE_SLOTS,CUDD_CACHE_SLOTS,0); /* Initialize a new BDD manager. */
DdNode *bdd, *var, *tmp_neg, *tmp,*f,*g;
int i;
bdd = Cudd_ReadOne(gbm); /*Returns the logic one constant of the manager*/
Cudd_Ref(bdd); /*Increases the reference count of a node*/
for (i = 3; i >= 0; i--) {
var = Cudd_bddIthVar(gbm,i); /*Create a new BDD variable*/
tmp_neg = Cudd_Not(var); /*Perform NOT Boolean operation*/
tmp = Cudd_bddAnd(gbm, tmp_neg, bdd); /*Perform AND Boolean operation*/
Cudd_Ref(tmp);
Cudd_RecursiveDeref(gbm,bdd);
f = tmp;
}
for (i = 3; i >= 0; i--) {
var = Cudd_bddIthVar(gbm,i); /*Create a new BDD variable*/
tmp = Cudd_bddXor(gbm, var, bdd); /*Perform AND Boolean operation*/
Cudd_Ref(tmp);
Cudd_RecursiveDeref(gbm,bdd);
g = tmp;
}
bdd= Cudd_bddIntersect(gbm,f,g);/*Intersection between F and G */
bdd = Cudd_BddToAdd(gbm, bdd); /*Convert BDD to ADD for display purpose*/
print_dd (gbm, bdd, 2,4); /*Print the dd to standard output*/
sprintf(filename, "./bdd/graph.dot"); /*Write .dot filename to a string*/
write_dd(gbm, bdd, filename); /*Write the resulting cascade dd to a file*/
Cudd_Quit(gbm);
return 0;
}
And here is the result I got:
DdManager nodes: 7 | DdManager vars: 4 | DdManager reorderings: 0 | DdManager memory: 8949888
: 3 nodes 2 leaves 2 minterms
ID = 0xaa40f index = 0 T = 0 E = 1
0--- 1
As you can see here the intersection gives A=0 and don't cares for B,C and D. I was expecting values of A,B,C and D that satifies both F and G. But clearly A=0 is not the solution for both F and G. For example someone can choose A=0,B=1 which gives 0 for function F. What is wrong here?

This reply comes awfully late, but just to close the issue, the problem is that the last operand to both Cudd_bddAnd and Cudd_bddXor is bdd instead of f or g. Of course, both f and g should be properly initialized (the way bdd is currently initialized). Fixing the code this way will also take care of the multiple dereferences of bdd, which are going to cause grief should garbage collection kick in.
Also, Cudd_bddIntersect does not compute the AND of two BDDs, but a function that implies the AND. It's used when one wants a witness to the nonemptiness of the conjunction of two BDDs without computing the whole result (and then possibly extracting a witness from it).
Finally, bdd is used as both operand to Cudd_BddToAdd and as destination for the return value. This is guaranteed to "leak" BDD nodes.

Related

Why is there a loop in this division as multiplication code?

I got the js code below from an archive of hackers delight (view the source)
The code takes in a value (such as 7) and spits out a magic number to multiply with. Then you bitshift to get the results. I don't remember assembly or any math so I'm sure I'm wrong but I can't find the reason why I'm wrong
From my understanding you could get a magic number by writing ceil(1/divide * 1<<32) (or <<64 for 64bit values, but you'd need bigger ints). If you multiple an integer with imul you'd get the result in one register and the remainder in another. The result register is magically the correct result of a division with this magic number from my formula
I wrote some C++ code to show what I mean. However I only tested with the values below. It seems correct. The JS code has a loop and more and I was wondering, why? Am I missing something? What values can I use to get an incorrect result that the JS code would get correctly? I'm not very good at math so I didn't understand any of the comments
#include <cstdio>
#include <cassert>
int main(int argc, char *argv[])
{
auto test_divisor = 7;
auto test_value = 43;
auto a = test_value*test_divisor;
auto b = a-1; //One less test
auto magic = (1ULL<<32)/test_divisor;
if (((1ULL<<32)%test_divisor) != 0) {
magic++; //Round up
}
auto answer1 = (a*magic) >> 32;
auto answer2 = (b*magic) >> 32;
assert(answer1 == test_value);
assert(answer2 == test_value-1);
printf("%lld %lld\n", answer1, answer2);
}
JS code from hackers delight
var two31 = 0x80000000
var two32 = 0x100000000
function magic_signed(d) { with(Math) {
if (d >= two31) d = d - two32// Treat large positive as short for negative.
var ad = abs(d)
var t = two31 + (d >>> 31)
var anc = t - 1 - t%ad // Absolute value of nc.
var p = 31 // Init p.
var q1 = floor(two31/anc) // Init q1 = 2**p/|nc|.
var r1 = two31 - q1*anc // Init r1 = rem(2**p, |nc|).
var q2 = floor(two31/ad) // Init q2 = 2**p/|d|.
var r2 = two31 - q2*ad // Init r2 = rem(2**p, |d|).
do {
p = p + 1;
q1 = 2*q1; // Update q1 = 2**p/|nc|.
r1 = 2*r1; // Update r1 = rem(2**p, |nc|.
if (r1 >= anc) { // (Must be an unsigned
q1 = q1 + 1; // comparison here).
r1 = r1 - anc;}
q2 = 2*q2; // Update q2 = 2**p/|d|.
r2 = 2*r2; // Update r2 = rem(2**p, |d|.
if (r2 >= ad) { // (Must be an unsigned
q2 = q2 + 1; // comparison here).
r2 = r2 - ad;}
var delta = ad - r2;
} while (q1 < delta || (q1 == delta && r1 == 0))
var mag = q2 + 1
if (d < 0) mag = two32 - mag // Magic number and
shift = p - 32 // shift amount to return.
return mag
}}
In the C CODE:
auto magic = (1ULL<<32)/test_divisor;
We get Integer Value in magic because both (1ULL<<32) & test_divisor are Integers.
The Algorithms requires incrementing magic on certain conditions, which is the next conditional statement.
Now, multiplication also gives Integers:
auto answer1 = (a*magic) >> 32;
auto answer2 = (b*magic) >> 32;
C CODE is DONE !
In the JS CODE:
All Variables are var ; no Data types !
No Integer Division ; No Integer Multiplication !
Bitwise Operations are not easy and not suitable to use in this Algorithm.
Numeric Data is via Number & BigInt which are not like "C Int" or "C Unsigned Long Long".
Hence the Algorithm is using loops to Iteratively add and compare whether "Division & Multiplication" has occurred to within the nearest Integer.
Both versions try to Implement the same Algorithm ; Both "should" give same answer, but JS Version is "buggy" & non-standard.
While there are many Issues with the JS version, I will highlight only 3:
(1) In the loop, while trying to get the best Power of 2, we have these two statements :
p = p + 1;
q1 = 2*q1; // Update q1 = 2**p/|nc|.
It is basically incrementing a counter & multiplying a number by 2, which is a left shift in C++.
The C++ version will not require this rigmarole.
(2) The while Condition has 2 Equality comparisons on RHS of || :
while (q1 < delta || (q1 == delta && r1 == 0))
But both these will be false in floating Point Calculations [[ eg check "Math.sqrt(2)*Math.sqrt(0.5) == 1" : even though this must be true, it will almost always be false ]] hence the while Condition is basically the LHS of || , because RHS will always be false.
(3) The JS version returns only one variable mag but user is supposed to get (& use) even variable shift which is given by global variable access. Inconsistent & BAD !
Comparing , we see that the C Version is more Standard, but Point is to not use auto but use int64_t with known number of bits.
First I think ceil(1/divide * 1<<32) can, depending on the divide, have cases where the result is off by one. So you don't need a loop but sometimes you need a corrective factor.
Secondly the JS code seems to allow for other shifts than 32: shift = p - 32 // shift amount to return. But then it never returns that. So not sure what is going on there.
Why not implement the JS code in C++ as well and then run a loop over all int32_t and see if they give the same result? That shouldn't take too long.
And when you find a d where they differ you can then test a / d for all int32_t a using both magic numbers and compare a / d, a * m_ceil and a * m_js.

Chapel: Can you re-index a domain in place?

A great man once said, I have a matrix A. But this time she has a friend B. Like the Montagues and Capulets, they have different domains.
// A.domain is { 1..10, 1..10 }
// B.domain is { 0.. 9, 0.. 9 }
for ij in B.domain {
if B[ij] <has a condition> {
// poops
A[ij] = B[ij];
}
}
My guess is I need to reindex so that the B.domain is {1..10, 1..10}. Since B is an input, I get push back from the compiler. Any suggestions?
There's a reindex array method to accomplish exactly this, and you can create a ref to the result to prevent creating a new array:
var Adom = {1..10,1..10},
Bdom = {0..9, 0..9};
var A: [Adom] real,
B: [Bdom] real;
// Set B to 1.0
B = 1;
// 0-based reference to A -- note that Bdom must be same shape as Adom
ref A0 = A.reindex(Bdom);
// Set all of A's values to B's values
for ij in B.domain {
A0[ij] = B[ij];
}
// Confirm A is now 1.0 now
writeln(A);
chapel compiler must object,documentation is explicit and clear on this:
Note that querying an array's domain via the .domain method or the function argument query syntax does not result in a domain expression that can be reassigned. In particular, we cannot do:
VarArr.domain = {1..2*n};
In case the <has_a_condition> is non-intervening and without side-effects, a solution to the expressed will may use domain-operators similar to this pure, contiguous-domain, index translation:
forall ij in B.domain do {
if <has_a_condition> {
A[ ij(1) + A.domain.dims()(1).low,
ij(2) + A.domain.dims()(2).low
] = B[ij];
}
}

Creating a DFA at run time. How many states?

I am currently working on a program that will take a English description of a language and then use the description to create a DFA for those specs. I allow certain operations, example {w | w has the sub string 01 in the beginning} and other options such as even odd sub string, more less or exactly than k sub string, etc. The user also chooses the alphabet.
My question is how would i know how many states I will need? since the user gives me my alphabet and rules I don't know anything until run time. I have created DFA's/transition tables before, but in those cases I knew what my DFA was and could declare it in a class or have it static. Should I be using the 5 tupil (Q, ∑, δ, q0, F)? or taking a different approach? Any help wrapping my head around the problem is appreciated.
My question is how would i know how many states I will need?
You won't.
You can represent the transition function as std::unordered_map of pairs of (q, σ) ∈ (Q, Σ) to q ∈ Q
using state = int;
using symbol = char;
using tranFn = std::unordered_map<std::pair<state, symbol>, state>;
// sets up transition function, the values can be read at runtime
tranFn fn;
fn[{0, 'a'}] = 1;
fn[{0, 'b'}] = 2;
fn[{1, 'a'}] = 1;
fn[{1, 'b'}] = 0;
fn[{2, 'a'}] = 2;
fn[{2, 'b'}] = 2;
// run the dfa
state s = 0;
symbol sym;
while(std::cin >> sym)
s = fn[{s, sym}];
std::cout << s;
btw, if 1 is the accepting state in the dfa above, it accepts exactly a(a + ba)*

Fast inner product of ternary vectors

Consider two vectors, A and B, of size n, 7 <= n <= 23. Both A and B consists of -1s, 0s and 1s only.
I need a fast algorithm which computes the inner product of A and B.
So far I've thought of storing the signs and values in separate uint32_ts using the following encoding:
sign 0, value 0 → 0
sign 0, value 1 → 1
sign 1, value 1 → -1.
The C++ implementation I've thought of looks like the following:
struct ternary_vector {
uint32_t sign, value;
};
int inner_product(const ternary_vector & a, const ternary_vector & b) {
uint32_t psign = a.sign ^ b.sign;
uint32_t pvalue = a.value & b.value;
psign &= pvalue;
pvalue ^= psign;
return __builtin_popcount(pvalue) - __builtin_popcount(psign);
}
This works reasonably well, but I'm not sure whether it is possible to do it better. Any comment on the matter is highly appreciated.
I like having the 2 uint32_t, but I think your actual calculation is a bit wasteful
Just a few minor points:
I'm not sure about the reference (getting a and b by const &) - this adds a level of indirection compared to putting them on the stack. When the code is this small (a couple of clocks maybe) this is significant. Try passing by value and see what you get
__builtin_popcount can be, unfortunately, very inefficient. I've used it myself, but found that even a very basic implementation I wrote was far faster than this. However - this is dependent on the platform.
Basically, if the platform has a hardware popcount implementation, __builtin_popcount uses it. If not - it uses a very inefficient replacement.
The one serious problem here is the reuse of the psign and pvalue variables for the positive and negative vectors. You are doing neither your compiler nor yourself any favors by obfuscating your code in this way.
Would it be possible for you to encode your ternary state in a std::bitset<2> and define the product in terms of and? For example, if your ternary types are:
1 = P = (1, 1)
0 = Z = (0, 0)
-1 = M = (1, 0) or (0, 1)
I believe you could define their product as:
1 * 1 = 1 => P * P = P => (1, 1) & (1, 1) = (1, 1) = P
1 * 0 = 0 => P * Z = Z => (1, 1) & (0, 0) = (0, 0) = Z
1 * -1 = -1 => P * M = M => (1, 1) & (1, 0) = (1, 0) = M
Then the inner product could start by taking the and of the bits of the elements and... I am working on how to add them together.
Edit:
My foolish suggestion did not consider that (-1)(-1) = 1, which cannot be handled by the representation I proposed. Thanks to #user92382 for bringing this up.
Depending on your architecture, you may want to optimize away the temporary bit vectors -- e.g. if your code is going to be compiled to FPGA, or laid out to an ASIC, then a sequence of logical operations will be better in terms of speed/energy/area than storing and reading/writing to two big buffers.
In this case, you can do:
int inner_product(const ternary_vector & a, const ternary_vector & b) {
return __builtin_popcount( a.value & b.value & ~(a.sign ^ b.sign))
- __builtin_popcount( a.value & b.value & (a.sign ^ b.sign));
}
This will lay out very well -- the (a.value & b.value & ... ) can enable/disable an XOR gate, whose output splits into two signed accumulators, with the first pathway NOTed before accumulation.

Wildcard String Search Algorithm

In my program I need to search in a quite big string (~1 mb) for a relatively small substring (< 1 kb).
The problem is the string contains simple wildcards in the sense of "a?c" which means I want to search for strings like "abc" or also "apc",... (I am only interested in the first occurence).
Until now I use the trivial approach (here in pseudocode)
algorithm "search", input: haystack(string), needle(string)
for(i = 0, i < length(haystack), ++i)
if(!CompareMemory(haystack+i,needle,length(needle))
return i;
return -1; (Not found)
Where "CompareMemory" returns 0 iff the first and second argument are identical (also concerning wildcards) only regarding the amount of bytes the third argument gives.
My question is now if there is a fast algorithm for this (you don't have to give it, but if you do I would prefer c++, c or pseudocode). I started here
but I think most of the fast algorithms don't allow wildcards (by the way they exploit the nature of strings).
I hope the format of the question is ok because I am new here, thank you in advance!
A fast way, which is kind of the same thing as using a regexp, (which I would recommend anyway), is to find something that is fixed in needle, "a", but not "?", and search for it, then see if you've got a complete match.
j = firstNonWildcardPos(needle)
for(i = j, i < length(haystack)-length(needle)+j, ++i)
if(haystack[i] == needle[j])
if(!CompareMemory(haystack+i-j,needle,length(needle))
return i;
return -1; (Not found)
A regexp would generate code similar to this (I believe).
Among strings over an alphabet of c characters, let S have length s and let T_1 ... T_k have average length b. S will be searched for each of the k target strings. (The problem statement doesn't mention multiple searches of a given string; I mention it below because in that paradigm my program does well.)
The program uses O(s+c) time and space for setup, and (if S and the T_i are random strings) O(k*u*s/c) + O(k*b + k*b*s/c^u) total time for searching, with u=3 in program as shown. For longer targets, u should be increased, and rare, widely-separated key characters chosen.
In step 1, the program creates an array L of s+TsizMax integers (in program, TsizMax = allowed target length) and uses it for c lists of locations of next occurrences of characters, with list heads in H[] and tails in T[]. This is the O(s+c) time and space step.
In step 2, the program repeatedly reads and processes target strings. Step 2A chooses u = 3 different non-wild key characters (in current target). As shown, the program just uses the first three such characters; with a tiny bit more work, it could instead use the rarest characters in the target, to improve performance. Note, it doesn't cope with targets with fewer than three such characters.
The line "L[T[r]] = L[g+i] = g+i;" within Step 2A sets up a guard cell in L with proper delta offset so that Step 2G will automatically execute at end of search, without needing any extra testing during the search. T[r] indexes the tail cell of the list for character r, so cell L[g+i] becomes a new, self-referencing, end-of-list for character r. (This technique allows the loops to run with a minimum of extraneous condition testing.)
Step 2B sets vars a,b,c to head-of-list locations, and sets deltas dab, dac, and dbc corresponding to distances between the chosen key characters in target.
Step 2C checks if key characters appear in S. This step is necessary because otherwise a while loop in Step 2E will hang. We don't want more checks within those while loops because they are the inner loops of search.
Step 2D does steps 2E to 2i until var c points to after end of S, at which point it is impossible to make any more matches.
Step 2E consists of u = 3 while loops, that "enforce delta distances", that is, crawl indexes a,b,c along over each other as long as they are not pattern-compatible. The while loops are fairly fast, each being in essence (with ++si instrumentation removed) "while (v+d < w) v = L[v]" for various v, d, w. Replicating the three while loops a few times may increase performance a little and will not change net results.
In Step 2G, we know that the u key characters match, so we do a complete compare of target to match point, with wild-character handling. Step 2H reports result of compare. Program as given also reports non-matches in this section; remove that in production.
Step 2I advances all the key-character indexes, because none of the currently-indexed characters can be the key part of another match.
You can run the program to see a few operation-count statistics. For example, the output
Target 5=<de?ga>
012345678901234567890123456789012345678901
abc1efgabc2efgabcde3gabcdefg4bcdefgabc5efg
# 17, de?ga and de3ga match
# 24, de?ga and defg4 differ
# 31, de?ga and defga match
Advances: 'd' 0+3 'e' 3+3 'g' 3+3 = 6+9 = 15
shows that Step 2G was entered 3 times (ie, the key characters matched 3 times); the full compare succeeded twice; step 2E while loops advanced indexes 6 times; step 2I advanced indexes 9 times; there were 15 advances in all, to search the 42-character string for the de?ga target.
/* jiw
$Id: stringsearch.c,v 1.2 2011/08/19 08:53:44 j-waldby Exp j-waldby $
Re: Concept-code for searching a long string for short targets,
where targets may contain wildcard characters.
The user can enter any number of targets as command line parameters.
This code has 2 long strings available for testing; if the first
character of the first parameter is '1' the jay[42] string is used,
else kay[321].
Eg, for tests with *hay = jay use command like
./stringsearch 1e?g a?cd bc?e?g c?efg de?ga ddee? ddee?f
or with *hay = kay,
./stringsearch bc?e? jih? pa?j ?av??j
to exercise program.
Copyright 2011 James Waldby. Offered without warranty
under GPL v3 terms as at http://www.gnu.org/licenses/gpl.html
*/
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <limits.h>
//================================================
int main(int argc, char *argv[]) {
char jay[]="abc1efgabc2efgabcde3gabcdefg4bcdefgabc5efg";
char kay[]="ludehkhtdiokihtmaihitoia1htkjkkchajajavpajkihtijkhijhipaja"
"etpajamhkajajacpajihiatokajavtoia2pkjpajjhiifakacpajjhiatkpajfojii"
"etkajamhpajajakpajihiatoiakavtoia3pakpajjhiifakacpajjhkatvpajfojii"
"ihiifojjjjhijpjkhtfdoiajadijpkoia4jihtfjavpapakjhiifjpajihiifkjach"
"ihikfkjjjjhijpjkhtfdoiajakijptoik4jihtfjakpapajjkiifjpajkhiifajkch";
char *hay = (argc>1 && argv[1][0]=='1')? jay:kay;
enum { chars=1<<CHAR_BIT, TsizMax=40, Lsiz=TsizMax+sizeof kay, L1, L2 };
int L[L2], H[chars], T[chars], g, k, par;
// Step 1. Make arrays L, H, T.
for (k=0; k<chars; ++k) H[k] = T[k] = L1; // Init H and T
for (g=0; hay[g]; ++g) { // Make linked character lists for hay.
k = hay[g]; // In same loop, could count char freqs.
if (T[k]==L1) H[k] = T[k] = g;
T[k] = L[T[k]] = g;
}
// Step 2. Read and process target strings.
for (par=1; par<argc; ++par) {
int alpha[3], at[3], a=g, b=g, c=g, da, dab, dbc, dac, i, j, r;
char * targ = argv[par];
enum { wild = '?' };
int sa=0, sb=0, sc=0, ta=0, tb=0, tc=0;
printf ("Target %d=<%s>\n", par, targ);
// Step 2A. Choose 3 non-wild characters to follow.
// As is, chooses first 3 non-wilds for a,b,c.
// Could instead choose 3 rarest characters.
for (j=0; j<3; ++j) alpha[j] = -j;
for (i=j=0; targ[i] && j<3; ++i)
if (targ[i] != wild) {
r = alpha[j] = targ[i];
if (alpha[0]==alpha[1] || alpha[1]==alpha[2]
|| alpha[0]==alpha[2]) continue;
at[j] = i;
L[T[r]] = L[g+i] = g+i;
++j;
}
if (j != 3) {
printf (" Too few target chars\n");
continue;
}
// Step 2B. Set a,b,c to head-of-list locations, set deltas.
da = at[0];
a = H[alpha[0]]; dab = at[1]-at[0];
b = H[alpha[1]]; dbc = at[2]-at[1];
c = H[alpha[2]]; dac = at[2]-at[0];
// Step 2C. See if key characters appear in haystack
if (a >= g || b >= g || c >= g) {
printf (" No match on some character\n");
continue;
}
for (g=0; hay[g]; ++g) printf ("%d", g%10);
printf ("\n%s\n", hay); // Show haystack, for user aid
// Step 2D. Search for match
while (c < g) {
// Step 2E. Enforce delta distances
while (a+dab < b) {a = L[a]; ++sa; } // Replicate these
while (b+dbc < c) {b = L[b]; ++sb; } // 3 abc lines as many
while (a+dac > c) {c = L[c]; ++sc; } // times as you like.
while (a+dab < b) {a = L[a]; ++sa; } // Replicate these
while (b+dbc < c) {b = L[b]; ++sb; } // 3 abc lines as many
while (a+dac > c) {c = L[c]; ++sc; } // times as you like.
// Step 2F. See if delta distances were met
if (a+dab==b && b+dbc==c && c<g) {
// Step 2G. Yes, so we have 3-letter-match and need to test whole match.
r = a-da;
for (k=0; targ[k]; ++k)
if ((hay[r+k] != targ[k]) && (targ[k] != wild))
break;
printf ("# %3d, %s and ", r, targ);
for (i=0; targ[i]; ++i) putchar(hay[r++]);
// Step 2H. Report match, if found
puts (targ[k]? " differ" : " match");
// Step 2I. Advance all of a,b,c, to go on looking
a = L[a]; ++ta;
b = L[b]; ++tb;
c = L[c]; ++tc;
}
}
printf ("Advances: '%c' %d+%d '%c' %d+%d '%c' %d+%d = %d+%d = %d\n",
alpha[0], sa,ta, alpha[1], sb,tb, alpha[2], sc,tc,
sa+sb+sc, ta+tb+tc, sa+sb+sc+ta+tb+tc);
}
return 0;
}
Note, if you like this answer better than current preferred answer, unmark that one and mark this one. :)
Regular expressions usually use a finite state automation-based search, I think. Try implementing that.