Révision 219 pobysoPythonSage/src/sageSLZ/sageRunSLZ.sage
sageRunSLZ.sage (revision 219) | ||
---|---|---|
1950 | 1950 |
Changes from V3: |
1951 | 1951 |
Root search is changed again: |
1952 | 1952 |
- only resultants in i are computed; |
1953 |
- root are searched for; |
|
1953 |
- roots in i are searched for;
|
|
1954 | 1954 |
- if any, they are tested for hardness-to-round. |
1955 | 1955 |
Changes from V2: |
1956 | 1956 |
Root search is changed: |
... | ... | |
2396 | 2396 |
## Output counters |
2397 | 2397 |
# End srs_runSLZ-v04 |
2398 | 2398 |
|
2399 |
def srs_run_SLZ_v05(inputFunction, |
|
2400 |
inputLowerBound, |
|
2401 |
inputUpperBound, |
|
2402 |
alpha, |
|
2403 |
degree, |
|
2404 |
precision, |
|
2405 |
emin, |
|
2406 |
emax, |
|
2407 |
targetHardnessToRound, |
|
2408 |
debug = False): |
|
2409 |
""" |
|
2410 |
Changes from V4: |
|
2411 |
Approximation polynomial has coefficients rounded. |
|
2412 |
Changes from V3: |
|
2413 |
Root search is changed again: |
|
2414 |
- only resultants in i are computed; |
|
2415 |
- roots in i are searched for; |
|
2416 |
- if any, they are tested for hardness-to-round. |
|
2417 |
Changes from V2: |
|
2418 |
Root search is changed: |
|
2419 |
- we compute the resultants in i and in t; |
|
2420 |
- we compute the roots set of each of these resultants; |
|
2421 |
- we combine all the possible pairs between the two sets; |
|
2422 |
- we check these pairs in polynomials for correctness. |
|
2423 |
Changes from V1: |
|
2424 |
1- check for roots as soon as a resultant is computed; |
|
2425 |
2- once a non null resultant is found, check for roots; |
|
2426 |
3- constant resultant == no root. |
|
2427 |
""" |
|
2428 |
|
|
2429 |
if debug: |
|
2430 |
print "Function :", inputFunction |
|
2431 |
print "Lower bound :", inputLowerBound |
|
2432 |
print "Upper bounds :", inputUpperBound |
|
2433 |
print "Alpha :", alpha |
|
2434 |
print "Degree :", degree |
|
2435 |
print "Precision :", precision |
|
2436 |
print "Emin :", emin |
|
2437 |
print "Emax :", emax |
|
2438 |
print "Target hardness-to-round:", targetHardnessToRound |
|
2439 |
|
|
2440 |
## Important constants. |
|
2441 |
### Stretch the interval if no error happens. |
|
2442 |
noErrorIntervalStretch = 1 + 2^(-5) |
|
2443 |
### If no vector validates the Coppersmith condition, shrink the interval |
|
2444 |
# by the following factor. |
|
2445 |
noCoppersmithIntervalShrink = 1/2 |
|
2446 |
### If only (or at least) one vector validates the Coppersmith condition, |
|
2447 |
# shrink the interval by the following factor. |
|
2448 |
oneCoppersmithIntervalShrink = 3/4 |
|
2449 |
#### If only null resultants are found, shrink the interval by the |
|
2450 |
# following factor. |
|
2451 |
onlyNullResultantsShrink = 3/4 |
|
2452 |
## Structures. |
|
2453 |
RRR = RealField(precision) |
|
2454 |
RRIF = RealIntervalField(precision) |
|
2455 |
## Converting input bound into the "right" field. |
|
2456 |
lowerBound = RRR(inputLowerBound) |
|
2457 |
upperBound = RRR(inputUpperBound) |
|
2458 |
## Before going any further, check domain and image binade conditions. |
|
2459 |
print inputFunction(1).n() |
|
2460 |
output = slz_fix_bounds_for_binades(lowerBound, upperBound, inputFunction) |
|
2461 |
if output is None: |
|
2462 |
print "Invalid domain/image binades. Domain:",\ |
|
2463 |
lowerBound, upperBound, "Images:", \ |
|
2464 |
inputFunction(lowerBound), inputFunction(upperBound) |
|
2465 |
raise Exception("Invalid domain/image binades.") |
|
2466 |
lb = output[0] ; ub = output[1] |
|
2467 |
if lb != lowerBound or ub != upperBound: |
|
2468 |
print "lb:", lb, " - ub:", ub |
|
2469 |
print "Invalid domain/image binades. Domain:",\ |
|
2470 |
lowerBound, upperBound, "Images:", \ |
|
2471 |
inputFunction(lowerBound), inputFunction(upperBound) |
|
2472 |
raise Exception("Invalid domain/image binades.") |
|
2473 |
# |
|
2474 |
## Progam initialization |
|
2475 |
### Approximation polynomial accuracy and hardness to round. |
|
2476 |
polyApproxAccur = 2^(-(targetHardnessToRound + 1)) |
|
2477 |
polyTargetHardnessToRound = targetHardnessToRound + 1 |
|
2478 |
### Significand to integer conversion ratio. |
|
2479 |
toIntegerFactor = 2^(precision-1) |
|
2480 |
print "Polynomial approximation required accuracy:", polyApproxAccur.n() |
|
2481 |
### Variables and rings for polynomials and root searching. |
|
2482 |
i=var('i') |
|
2483 |
t=var('t') |
|
2484 |
inputFunctionVariable = inputFunction.variables()[0] |
|
2485 |
function = inputFunction.subs({inputFunctionVariable:i}) |
|
2486 |
# Polynomial Rings over the integers, for root finding. |
|
2487 |
Zi = ZZ[i] |
|
2488 |
Zt = ZZ[t] |
|
2489 |
Zit = ZZ[i,t] |
|
2490 |
## Number of iterations limit. |
|
2491 |
maxIter = 100000 |
|
2492 |
# |
|
2493 |
## Compute the scaled function and the degree, in their Sollya version |
|
2494 |
# once for all. |
|
2495 |
(scaledf, sdlb, sdub, silb, siub) = \ |
|
2496 |
slz_compute_scaled_function(function, lowerBound, upperBound, precision) |
|
2497 |
print "Scaled function:", scaledf._assume_str().replace('_SAGE_VAR_', '') |
|
2498 |
scaledfSo = sollya_lib_parse_string(scaledf._assume_str().replace('_SAGE_VAR_', '')) |
|
2499 |
degreeSo = pobyso_constant_from_int_sa_so(degree) |
|
2500 |
# |
|
2501 |
## Compute the scaling. boundsIntervalRifSa defined out of the loops. |
|
2502 |
domainBoundsInterval = RRIF(lowerBound, upperBound) |
|
2503 |
(unscalingFunction, scalingFunction) = \ |
|
2504 |
slz_interval_scaling_expression(domainBoundsInterval, i) |
|
2505 |
#print scalingFunction, unscalingFunction |
|
2506 |
## Set the Sollya internal precision (with an arbitrary minimum of 192). |
|
2507 |
internalSollyaPrec = ceil((RR('1.5') * targetHardnessToRound) / 64) * 64 |
|
2508 |
if internalSollyaPrec < 192: |
|
2509 |
internalSollyaPrec = 192 |
|
2510 |
pobyso_set_prec_sa_so(internalSollyaPrec) |
|
2511 |
print "Sollya internal precision:", internalSollyaPrec |
|
2512 |
## Some variables. |
|
2513 |
### General variables |
|
2514 |
lb = sdlb |
|
2515 |
ub = sdub |
|
2516 |
nbw = 0 |
|
2517 |
intervalUlp = ub.ulp() |
|
2518 |
#### Will be set by slz_interval_and_polynomila_to_sage. |
|
2519 |
ic = 0 |
|
2520 |
icAsInt = 0 # Set from ic. |
|
2521 |
solutionsSet = set() |
|
2522 |
tsErrorWidth = [] |
|
2523 |
csErrorVectors = [] |
|
2524 |
csVectorsResultants = [] |
|
2525 |
floatP = 0 # Taylor polynomial. |
|
2526 |
floatPcv = 0 # Ditto with variable change. |
|
2527 |
intvl = "" # Taylor interval |
|
2528 |
terr = 0 # Taylor error. |
|
2529 |
iterCount = 0 |
|
2530 |
htrnSet = set() |
|
2531 |
### Timers and counters. |
|
2532 |
wallTimeStart = 0 |
|
2533 |
cpuTimeStart = 0 |
|
2534 |
taylCondFailedCount = 0 |
|
2535 |
coppCondFailedCount = 0 |
|
2536 |
resultCondFailedCount = 0 |
|
2537 |
coppCondFailed = False |
|
2538 |
resultCondFailed = False |
|
2539 |
globalResultsList = [] |
|
2540 |
basisConstructionsCount = 0 |
|
2541 |
basisConstructionsFullTime = 0 |
|
2542 |
basisConstructionTime = 0 |
|
2543 |
reductionsCount = 0 |
|
2544 |
reductionsFullTime = 0 |
|
2545 |
reductionTime = 0 |
|
2546 |
resultantsComputationsCount = 0 |
|
2547 |
resultantsComputationsFullTime = 0 |
|
2548 |
resultantsComputationTime = 0 |
|
2549 |
rootsComputationsCount = 0 |
|
2550 |
rootsComputationsFullTime = 0 |
|
2551 |
rootsComputationTime = 0 |
|
2552 |
|
|
2553 |
## Global times are started here. |
|
2554 |
wallTimeStart = walltime() |
|
2555 |
cpuTimeStart = cputime() |
|
2556 |
## Main loop. |
|
2557 |
while True: |
|
2558 |
if lb >= sdub: |
|
2559 |
print "Lower bound reached upper bound." |
|
2560 |
break |
|
2561 |
if iterCount == maxIter: |
|
2562 |
print "Reached maxIter. Aborting" |
|
2563 |
break |
|
2564 |
iterCount += 1 |
|
2565 |
print "[", lb, ",", ub, "]", ((ub - lb) / intervalUlp).log2().n(), \ |
|
2566 |
"log2(numbers)." |
|
2567 |
### Compute a Sollya polynomial that will honor the Taylor condition. |
|
2568 |
prceSo = slz_compute_polynomial_and_interval_01(scaledfSo, |
|
2569 |
degreeSo, |
|
2570 |
lb, |
|
2571 |
ub, |
|
2572 |
polyApproxAccur) |
|
2573 |
### Convert back the data into Sage space. |
|
2574 |
(floatP, floatPcv, intvl, ic, terr) = \ |
|
2575 |
slz_interval_and_polynomial_to_sage((prceSo[0], prceSo[0], |
|
2576 |
prceSo[1], prceSo[2], |
|
2577 |
prceSo[3])) |
|
2578 |
intvl = RRIF(intvl) |
|
2579 |
## Clean-up Sollya stuff. |
|
2580 |
for elem in prceSo: |
|
2581 |
sollya_lib_clear_obj(elem) |
|
2582 |
#print floatP, floatPcv, intvl, ic, terr |
|
2583 |
#print floatP |
|
2584 |
#print intvl.endpoints()[0].n(), \ |
|
2585 |
# ic.n(), |
|
2586 |
#intvl.endpoints()[1].n() |
|
2587 |
### Check returned data. |
|
2588 |
#### Is approximation error OK? |
|
2589 |
if terr > polyApproxAccur: |
|
2590 |
exceptionErrorMess = \ |
|
2591 |
"Approximation failed - computed error:" + \ |
|
2592 |
str(terr) + " - target error: " |
|
2593 |
exceptionErrorMess += \ |
|
2594 |
str(polyApproxAccur) + ". Aborting!" |
|
2595 |
raise Exception(exceptionErrorMess) |
|
2596 |
#### Is lower bound OK? |
|
2597 |
if lb != intvl.endpoints()[0]: |
|
2598 |
exceptionErrorMess = "Wrong lower bound:" + \ |
|
2599 |
str(lb) + ". Aborting!" |
|
2600 |
raise Exception(exceptionErrorMess) |
|
2601 |
#### Set upper bound. |
|
2602 |
if ub > intvl.endpoints()[1]: |
|
2603 |
ub = intvl.endpoints()[1] |
|
2604 |
print "[", lb, ",", ub, "]", ((ub - lb) / intervalUlp).log2().n(), \ |
|
2605 |
"log2(numbers)." |
|
2606 |
taylCondFailedCount += 1 |
|
2607 |
#### Is interval not degenerate? |
|
2608 |
if lb >= ub: |
|
2609 |
exceptionErrorMess = "Degenerate interval: " + \ |
|
2610 |
"lowerBound(" + str(lb) +\ |
|
2611 |
")>= upperBound(" + str(ub) + \ |
|
2612 |
"). Aborting!" |
|
2613 |
raise Exception(exceptionErrorMess) |
|
2614 |
#### Is interval center ok? |
|
2615 |
if ic <= lb or ic >= ub: |
|
2616 |
exceptionErrorMess = "Invalid interval center for " + \ |
|
2617 |
str(lb) + ',' + str(ic) + ',' + \ |
|
2618 |
str(ub) + ". Aborting!" |
|
2619 |
raise Exception(exceptionErrorMess) |
|
2620 |
##### Current interval width and reset future interval width. |
|
2621 |
bw = ub - lb |
|
2622 |
nbw = 0 |
|
2623 |
icAsInt = int(ic * toIntegerFactor) |
|
2624 |
#### The following ratio is always >= 1. In case we may want to |
|
2625 |
# enlarge the interval |
|
2626 |
curTaylErrRat = polyApproxAccur / terr |
|
2627 |
### Make the integral transformations. |
|
2628 |
#### Bounds and interval center. |
|
2629 |
intIc = int(ic * toIntegerFactor) |
|
2630 |
intLb = int(lb * toIntegerFactor) - intIc |
|
2631 |
intUb = int(ub * toIntegerFactor) - intIc |
|
2632 |
# |
|
2633 |
#### Polynomials |
|
2634 |
basisConstructionTime = cputime() |
|
2635 |
##### To a polynomial with rational coefficients with rational arguments |
|
2636 |
ratRatP = slz_float_poly_of_float_to_rat_poly_of_rat_pow_two(floatP) |
|
2637 |
##### To a polynomial with rational coefficients with integer arguments |
|
2638 |
ratIntP = \ |
|
2639 |
slz_rat_poly_of_rat_to_rat_poly_of_int(ratRatP, precision) |
|
2640 |
##### Ultimately a multivariate polynomial with integer coefficients |
|
2641 |
# with integer arguments. |
|
2642 |
coppersmithTuple = \ |
|
2643 |
slz_rat_poly_of_int_to_poly_for_coppersmith(ratIntP, |
|
2644 |
precision, |
|
2645 |
targetHardnessToRound, |
|
2646 |
i, t) |
|
2647 |
#### Recover Coppersmith information. |
|
2648 |
intIntP = coppersmithTuple[0] |
|
2649 |
N = coppersmithTuple[1] |
|
2650 |
nAtAlpha = N^alpha |
|
2651 |
tBound = coppersmithTuple[2] |
|
2652 |
leastCommonMultiple = coppersmithTuple[3] |
|
2653 |
iBound = max(abs(intLb),abs(intUb)) |
|
2654 |
basisConstructionsFullTime += cputime(basisConstructionTime) |
|
2655 |
basisConstructionsCount += 1 |
|
2656 |
""" |
|
2657 |
#### Compute the matrix to reduce. |
|
2658 |
matrixToReduce = slz_compute_initial_lattice_matrix(intIntP, |
|
2659 |
alpha, |
|
2660 |
N, |
|
2661 |
iBound, |
|
2662 |
tBound) |
|
2663 |
matrixFile = file('/tmp/matrixToReduce.txt', 'w') |
|
2664 |
for row in matrixToReduce.rows(): |
|
2665 |
matrixFile.write(str(row) + "\n") |
|
2666 |
matrixFile.close() |
|
2667 |
raise Exception("Deliberate stop here.") |
|
2668 |
""" |
|
2669 |
reductionTime = cputime() |
|
2670 |
#### Compute the reduced polynomials. |
|
2671 |
ccReducedPolynomialsList = \ |
|
2672 |
slz_compute_coppersmith_reduced_polynomials(intIntP, |
|
2673 |
alpha, |
|
2674 |
N, |
|
2675 |
iBound, |
|
2676 |
tBound) |
|
2677 |
if ccReducedPolynomialsList is None: |
|
2678 |
raise Exception("Reduction failed.") |
|
2679 |
reductionsFullTime += cputime(reductionTime) |
|
2680 |
reductionsCount += 1 |
|
2681 |
if len(ccReducedPolynomialsList) < 2: |
|
2682 |
print "Nothing to form resultants with." |
|
2683 |
|
|
2684 |
coppCondFailedCount += 1 |
|
2685 |
coppCondFailed = True |
|
2686 |
##### Apply a different shrink factor according to |
|
2687 |
# the number of compliant polynomials. |
|
2688 |
if len(ccReducedPolynomialsList) == 0: |
|
2689 |
ub = lb + bw * noCoppersmithIntervalShrink |
|
2690 |
else: # At least one compliant polynomial. |
|
2691 |
ub = lb + bw * oneCoppersmithIntervalShrink |
|
2692 |
if ub > sdub: |
|
2693 |
ub = sdub |
|
2694 |
if lb == ub: |
|
2695 |
raise Exception("Cant shrink interval \ |
|
2696 |
anymore to get Coppersmith condition.") |
|
2697 |
nbw = 0 |
|
2698 |
continue |
|
2699 |
#### We have at least two polynomials. |
|
2700 |
# Let us try to compute resultants. |
|
2701 |
# For each resultant computed, go for the solutions. |
|
2702 |
##### Build the pairs list. |
|
2703 |
polyPairsList = [] |
|
2704 |
for polyOuterIndex in xrange(0, len(ccReducedPolynomialsList) - 1): |
|
2705 |
for polyInnerIndex in xrange(polyOuterIndex+1, |
|
2706 |
len(ccReducedPolynomialsList)): |
|
2707 |
polyPairsList.append((ccReducedPolynomialsList[polyOuterIndex], |
|
2708 |
ccReducedPolynomialsList[polyInnerIndex])) |
|
2709 |
#### Actual root search. |
|
2710 |
iRootsSet = set() |
|
2711 |
hasNonNullResultant = False |
|
2712 |
for polyPair in polyPairsList: |
|
2713 |
resultantsComputationTime = cputime() |
|
2714 |
currentResultantI = \ |
|
2715 |
slz_resultant(polyPair[0], |
|
2716 |
polyPair[1], |
|
2717 |
t) |
|
2718 |
resultantsComputationsCount += 1 |
|
2719 |
resultantsComputationsFullTime += \ |
|
2720 |
cputime(resultantsComputationTime) |
|
2721 |
#### Function slz_resultant returns None both for None and O |
|
2722 |
# resultants. |
|
2723 |
if currentResultantI is None: |
|
2724 |
print "Nul resultant" |
|
2725 |
continue # Next polyPair. |
|
2726 |
## We deleted the currentResultantI computation. |
|
2727 |
#### We have a non null resultant. From now on, whatever this |
|
2728 |
# root search yields, no extra root search is necessary. |
|
2729 |
hasNonNullResultant = True |
|
2730 |
#### A constant resultant leads to no root. Root search is done. |
|
2731 |
if currentResultantI.degree() < 1: |
|
2732 |
print "Resultant is constant:", currentResultantI |
|
2733 |
break # There is no root. |
|
2734 |
#### Actual iroots computation. |
|
2735 |
rootsComputationTime = cputime() |
|
2736 |
iRootsList = Zi(currentResultantI).roots() |
|
2737 |
rootsComputationsCount += 1 |
|
2738 |
rootsComputationsFullTime = cputime(rootsComputationTime) |
|
2739 |
if len(iRootsList) == 0: |
|
2740 |
print "No roots in \"i\"." |
|
2741 |
break # No roots in i. |
|
2742 |
else: |
|
2743 |
for iRoot in iRootsList: |
|
2744 |
# A root is given as a (value, multiplicity) tuple. |
|
2745 |
iRootsSet.add(iRoot[0]) |
|
2746 |
# End loop for polyPair in polyParsList. We only loop again if a |
|
2747 |
# None or zero resultant is found. |
|
2748 |
#### Prepare for results for the current interval.. |
|
2749 |
intervalResultsList = [] |
|
2750 |
intervalResultsList.append((lb, ub)) |
|
2751 |
#### Check roots. |
|
2752 |
rootsResultsList = [] |
|
2753 |
for iRoot in iRootsSet: |
|
2754 |
specificRootResultsList = [] |
|
2755 |
failingBounds = [] |
|
2756 |
# Root qualifies for modular equation, test it for hardness to round. |
|
2757 |
hardToRoundCaseAsFloat = RRR((icAsInt + iRoot) / toIntegerFactor) |
|
2758 |
#print "Before unscaling:", hardToRoundCaseAsFloat.n(prec=precision) |
|
2759 |
#print scalingFunction |
|
2760 |
scaledHardToRoundCaseAsFloat = \ |
|
2761 |
scalingFunction(hardToRoundCaseAsFloat) |
|
2762 |
print "Candidate HTRNc at x =", \ |
|
2763 |
scaledHardToRoundCaseAsFloat.n().str(base=2), |
|
2764 |
if slz_is_htrn(scaledHardToRoundCaseAsFloat, |
|
2765 |
function, |
|
2766 |
2^-(targetHardnessToRound), |
|
2767 |
RRR): |
|
2768 |
print hardToRoundCaseAsFloat, "is HTRN case." |
|
2769 |
specificRootResultsList.append(hardToRoundCaseAsFloat.n().str(base=2)) |
|
2770 |
if lb <= hardToRoundCaseAsFloat and hardToRoundCaseAsFloat <= ub: |
|
2771 |
print "Found in interval." |
|
2772 |
else: |
|
2773 |
print "Found out of interval." |
|
2774 |
# Check the i root is within the i bound. |
|
2775 |
if abs(iRoot) > iBound: |
|
2776 |
print "IRoot", iRoot, "is out of bounds for modular equation." |
|
2777 |
print "i bound:", iBound |
|
2778 |
failingBounds.append('i') |
|
2779 |
failingBounds.append(iRoot) |
|
2780 |
failingBounds.append(iBound) |
|
2781 |
if len(failingBounds) > 0: |
|
2782 |
specificRootResultsList.append(failingBounds) |
|
2783 |
else: # From slz_is_htrn... |
|
2784 |
print "is not an HTRN case." |
|
2785 |
if len(specificRootResultsList) > 0: |
|
2786 |
rootsResultsList.append(specificRootResultsList) |
|
2787 |
if len(rootsResultsList) > 0: |
|
2788 |
intervalResultsList.append(rootsResultsList) |
|
2789 |
### Check if a non null resultant was found. If not shrink the interval. |
|
2790 |
if not hasNonNullResultant: |
|
2791 |
print "Only null resultants for this reduction, shrinking interval." |
|
2792 |
resultCondFailed = True |
|
2793 |
resultCondFailedCount += 1 |
|
2794 |
### Shrink interval for next iteration. |
|
2795 |
ub = lb + bw * onlyNullResultantsShrink |
|
2796 |
if ub > sdub: |
|
2797 |
ub = sdub |
|
2798 |
nbw = 0 |
|
2799 |
continue |
|
2800 |
#### An intervalResultsList has at least the bounds. |
|
2801 |
globalResultsList.append(intervalResultsList) |
|
2802 |
#### Compute an incremented width for next upper bound, only |
|
2803 |
# if not Coppersmith condition nor resultant condition |
|
2804 |
# failed at the previous run. |
|
2805 |
if not coppCondFailed and not resultCondFailed: |
|
2806 |
nbw = noErrorIntervalStretch * bw |
|
2807 |
else: |
|
2808 |
nbw = bw |
|
2809 |
##### Reset the failure flags. They will be raised |
|
2810 |
# again if needed. |
|
2811 |
coppCondFailed = False |
|
2812 |
resultCondFailed = False |
|
2813 |
#### For next iteration (at end of loop) |
|
2814 |
#print "nbw:", nbw |
|
2815 |
lb = ub |
|
2816 |
ub += nbw |
|
2817 |
if ub > sdub: |
|
2818 |
ub = sdub |
|
2819 |
|
|
2820 |
# End while True |
|
2821 |
## Main loop just ended. |
|
2822 |
globalWallTime = walltime(wallTimeStart) |
|
2823 |
globalCpuTime = cputime(cpuTimeStart) |
|
2824 |
## Output results |
|
2825 |
print ; print "Intervals and HTRNs" ; print |
|
2826 |
for intervalResultsList in globalResultsList: |
|
2827 |
print "[", intervalResultsList[0][0], ",",intervalResultsList[0][1], "]", |
|
2828 |
if len(intervalResultsList) > 1: |
|
2829 |
rootsResultsList = intervalResultsList[1] |
|
2830 |
for specificRootResultsList in rootsResultsList: |
|
2831 |
print "\t", specificRootResultsList[0], |
|
2832 |
if len(specificRootResultsList) > 1: |
|
2833 |
print specificRootResultsList[1], |
|
2834 |
print ; print |
|
2835 |
#print globalResultsList |
|
2836 |
# |
|
2837 |
print "Timers and counters" |
|
2838 |
|
|
2839 |
print "Number of iterations:", iterCount |
|
2840 |
print "Taylor condition failures:", taylCondFailedCount |
|
2841 |
print "Coppersmith condition failures:", coppCondFailedCount |
|
2842 |
print "Resultant condition failures:", resultCondFailedCount |
|
2843 |
print "Iterations count: ", iterCount |
|
2844 |
print "Number of intervals:", len(globalResultsList) |
|
2845 |
print "Number of basis constructions:", basisConstructionsCount |
|
2846 |
print "Total CPU time spent in basis constructions:", \ |
|
2847 |
basisConstructionsFullTime |
|
2848 |
if basisConstructionsCount != 0: |
|
2849 |
print "Average basis construction CPU time:", \ |
|
2850 |
basisConstructionsFullTime/basisConstructionsCount |
|
2851 |
print "Number of reductions:", reductionsCount |
|
2852 |
print "Total CPU time spent in reductions:", reductionsFullTime |
|
2853 |
if reductionsCount != 0: |
|
2854 |
print "Average reduction CPU time:", \ |
|
2855 |
reductionsFullTime/reductionsCount |
|
2856 |
print "Number of resultants computation rounds:", \ |
|
2857 |
resultantsComputationsCount |
|
2858 |
print "Total CPU time spent in resultants computation rounds:", \ |
|
2859 |
resultantsComputationsFullTime |
|
2860 |
if resultantsComputationsCount != 0: |
|
2861 |
print "Average resultants computation round CPU time:", \ |
|
2862 |
resultantsComputationsFullTime/resultantsComputationsCount |
|
2863 |
print "Number of root finding rounds:", rootsComputationsCount |
|
2864 |
print "Total CPU time spent in roots finding rounds:", \ |
|
2865 |
rootsComputationsFullTime |
|
2866 |
if rootsComputationsCount != 0: |
|
2867 |
print "Average roots finding round CPU time:", \ |
|
2868 |
rootsComputationsFullTime/rootsComputationsCount |
|
2869 |
print "Global Wall time:", globalWallTime |
|
2870 |
print "Global CPU time:", globalCpuTime |
|
2871 |
## Output counters |
|
2872 |
# End srs_runSLZ-v05 |
Formats disponibles : Unified diff