common-math-cvs

common-math-cvs@common-lisp.net

5 discussions

[common-math-cvs] Is Israel a Democracy? -- The problem with intellectually insecure whites -- Should Christians Support Israeli Terrorism in Gaza?
by Lawrence Auster 23 Jan '09

23 Jan '09

The Jewish State of Israel has no constitution, nor does it name its borders. Israel's hidden constitution is Judaism. Israel's undeclared borders range from the Nile to the Euphrates rivers. Israel's desired jurisdiction extends over the entire Earth. It could not be more clear that the Jewish State follows a foreign policy which obeys Jewish Law as iterated in the Hebrew Bible, the Talmud, Maimonedes, the Cabalah, and the many commentaries and refinements of same. The Jews are genociding the native inhabitants of Palestine, just as their religion advises, and because their religion teaches them to do so. They treat non-Jews as if non-humans, just as their religion requires them to do. They make perpetual war on every nation on Earth, just as their genocidal Jewish God has instructed. The Jews of Israel are simply being Jews. Jews are an existential threat to the human race. Israel contains one third of the Jews of the World. It is not some aberration of the Jewish spirit, but the condensation and concentration of the perverse Jewish mentality, which malady also pervades the remaining two thirds of Jewry, who almost unanimously support the Jewish State, and who certainly do unanimously support the Jewish People and its consistent and constant crimes against the human race. Israel is Jewry and the danger of Israel is the danger of the Jewish People to all others, as the Jews have demonstrated each and every day of their existence. The Jews, the entire Jewish People of 15 million, will not relent until they have wiped out all non-Jews in "Greater Israel". They will not stop destroying all other cultures, nations, religions, ethnicities, races, competition, etc. until they are either stopped, or succeed in their ancient quest to destroy the human race. What Israel is doing is not some reaction to outside forces, nor was the formation of Israel a response to the Holocaust. Israel is simply following the plan laid out in the Jews' religious texts. The Jews have openly planned to take Palestine and genocide the native population of Palestine for some 2,500 years before the Holocaust. The Jews have openly complained that "anti-Semitism" is a threat that gives them the right to genocide the Palestinians, not merely since the advent of Nazism, but for some 2,500 years. The Jewish religion is the Constitution of the Jewish State of Israel, and, to a greater or lesser extent, the constitution of the nature of every Jew alive. The borders of Israel are the range the Jew roams over the entire World. The perverse Jewish mentality is inbred by a Jew's exposure to his parents and to his community. Judaism passes in the spit and slobber of Jewish mother telling her Jewish child that he is a "Jew", as much as Judaism passes in the poison and pain of a Talmudic tractate. The secular Jews did not suddenly come to life after the Enlightenment and the Jewish Reformation a body of vampires that appeared ex nihilo, in vacuo, mostly atheistical and undetached from formally practiced Judaism. Judaism is the Jew. It is a mindset that transcends and supercedes religion. It is a belief set, a way of life, a perception of one's self and one's relation to the World that makes a Jew, a Jew, and a danger to all of humanity. In fact, the religious shell of Judaism is like the stretched and infected skin of a lycanthropic pustule. When you lance it to cure the infection, the virus only becomes more contagious and spills directly on the non-Jew. The secular Jew is a deliberate product of the hyper-religious Jew, a monster created out of the hewed corpses of the fanatically religious Jew, a Golem which is conjured up to enter the World of the non-Jew and poison its blood, and boil its brain with a rabid lunacy that bites and spreads, until the infected community feeds on itself and fills the fields with rotting bloating bodies, where once human beings tilled the soil and tended to their families. The religious Jew created the secular Jew as an army of Esthers who seduce with open thighs, broad smiles, and a Siren call that lures in the non-Jew to cast his skull upon the jagged rocks and color the seas with his blood, sickened and blinded by the venereal disease of Judaism in secular form. Israel is not a secular democracy. It is a religious mockery. It is a rabid bat flying to the ends of the Earth, to end the Earth. No one will be free nor safe until the disease is quarantined and dies out. Source: http://www.ziopedia.org/articles/israel/how_can_israel_claim_to_be_a_%27dem… -------------------- The problem with intellectually insecure whites By Kevin MacDonald January 19, 2009 America will soon have a white minority. This is a much desired state of affairs for the hostile elites who hold political power and shape public opinion. But it certainly creates some management issues � at least in the long run. After all, it�s difficult to come up with an historical example of a nation with a solid ethnic majority (90% white in 1950) that has voluntarily decided to cede political and cultural power. Such transformations are typically accomplished by military invasions, great battles, and untold suffering. And it�s not as if everyone is doing it. Only Western nations view their own demographic and cultural eclipse as a moral imperative. Indeed, as I have noted previously, it is striking that racial nationalism has triumphed in Israel at the same time that the Jewish intellectual and political movements and the organized Jewish community have been the most active and effective force for a non-white America. Indeed, a poll in 2008 found that Avigdor Lieberman was the second most popular politician in Israel. Lieberman has advocated expulsion of Arabs from Israel and has declared himself a follower of Vladimir Jabotinsky, the leading pioneer of racial Zionism. The most popular politician in the poll was Benjamin Netanyahu � another admirer of Jabotinsky. Prime Minister Ehud Olmert and Foreign Minister Tzipi Livni are also Jabotinskyists. The racial Zionists are now carrying out yet another orgy of mass murder after a starvation-inducing blockade and the usual triggering assault designed to provoke Palestinian retaliation � which then becomes the cover for claims that Israel is merely defending itself against terrorism. This monstrosity was approved by overwhelming majorities of both Houses of Congress. The craven Bush administration did its part by abstaining from a UN resolution designed by the US Secretary of State as a result of a personal appeal by the Israeli Prime Minister. This is yet another accomplishment of the Israel Lobby, but one they would rather not have discussed in public. People might get the impression that the Lobby really does dictate US foreign policy in the Mideast. Obviously, such thoughts are only entertained by anti-Semites. But I digress. In managing the eclipse of white America, one strategy of the mainstream media is to simply ignore the issue. Christopher Donovan (�For the media, the less whites think about their coming minority status, the better�) has noted that the media, and in particular, the New York Times, are quite uninterested in doing stories that discuss what white people think about this state of affairs. It�s not surprising that the New York Times � the Jewish-owned flagship of anti-white, pro-multicultural media � ignores the issue. The issue is also missing from so-called conservative media even though one would think that conservatives would find the eclipse of white America to be an important issue. Certainly, their audiences would find it interesting. Now we have an article �The End of White America� written by Hua Hsu, an Assistant Professor of English at Vassar College. The article is a rather depressing display of what passes for intellectual discourse on the most important question confronting white people in America. Hsu begins by quoting a passage in F. Scott Fitzgerald�s The Great Gatsby in which a character, Tom Buchanan, states: �Have you read The Rise of the Colored Empires by this man Goddard?� � Well, it�s a fine book, and everybody ought to read it. The idea is if we don�t look out the white race will be�will be utterly submerged. It�s all scientific stuff; it�s been proved.� Buchanan�s comment is a thinly veiled reference to Lothrop Stoddard�s The Rising Tide of Color which Hsu describes as �rationalized hatred� presented in a scholarly, gentlemanly, and scientific tone. (This wording that will certainly help him when he comes up for tenure.) As Hsu notes, Stoddard had a doctorate from Harvard and was a member of many academic associations. His book was published by a major publisher. It was therefore �precisely the kind of book that a 1920s man of Buchanan�s profile � wealthy, Ivy League�educated, at once pretentious and intellectually insecure � might have been expected to bring up in casual conversation.� Let�s ponder that a bit. The simple reality is that in the year 2009 an Ivy League-educated person, "at once pretentious and intellectually insecure," would just as glibly assert the same sort of nonsense as Hsu. To wit: The coming white minority does not mean that the racial hierarchy of American culture will suddenly become inverted, as in 1995�s White Man�s Burden, an awful thought experiment of a film, starring John Travolta, that envisions an upside-down world in which whites are subjugated to their high-class black oppressors. There will be dislocations and resentments along the way, but the demographic shifts of the next 40 years are likely to reduce the power of racial hierarchies over everyone�s lives, producing a culture that�s more likely than any before to treat its inhabitants as individuals, rather than members of a caste or identity group. The fact is that no one can say for certain what multicultural America without a white majority will be like. There is no scientific or historical basis for claims like �the demographic shifts of the next 40 years are likely to reduce the power of racial hierarchies over everyone�s lives, producing a culture that�s more likely than any before to treat its inhabitants as individuals, rather than members of a caste or identity group.� Indeed, there is no evidence at all that we are proceeding to a color blind future. The election results continue to show that white people are coalescing in the Republican Party, while the Democrats are increasingly the party of a non-white soon-to-be majority. Is it so hard to believe that when this coalition achieves a majority that it will further compromise the interests of whites far beyond contemporary concerns such as immigration policy and affirmative action? Hsu anticipates a colorblind world, but affirmative action means that blacks and other minorities are certainly not treated as individuals. And it means that whites � especially white males � are losing out on opportunities they would have had without these policies and without the massive non-white immigration of the last few decades. Given the intractability of changing intelligence and other traits required for success in the contemporary economy, it is unlikely that 40 more years of affirmative action will attain the outcomes desired by the minority lobbies. Indeed, in Obama's America, blacks are rioting in Oakland over perceived racial injustices, and from 2002 �2007, black juvenile homicide victims increased 31%, while black juvenile homicide perpetrators increased 43%. Hence, the reasonable outlook is for a continuing need for affirmative action and for racial activism in these groups, even after whites become a minority. Whites will also lose out because of large-scale importation of relatively talented immigrants from East Asia. Indeed, as I noted over a decade ago, "The United States is well on the road to being dominated by an Asian technocratic elite and a Jewish business, professional, and media elite." Hsu shows that there already is considerable anxiety among whites about the future. An advertizing executive says, �I think white people feel like they�re under siege right now � like it�s not okay to be white right now, especially if you�re a white male. ... People are stressed out about it. �We used to be in control! We�re losing control�� Another says, "There�s a lot of fear and a lot of resentment." It's hard to see why these feelings won't increase in the future. A huge problem for white people is lack of intellectual and cultural confidence. Hsu quotes Christian (Stuff White People Like) Lander saying, "I get it: as a straight white male, I�m the worst thing on Earth." A professor comments that for his students "to be white is to be culturally broke. The classic thing white students say when you ask them to talk about who they are is, �I don�t have a culture.� They might be privileged, they might be loaded socioeconomically, but they feel bankrupt when it comes to culture � They feel disadvantaged, and they feel marginalized." This lack of cultural confidence is no accident. For nearly 100 years whites have been subjected to a culture of critique emanating from the most prestigious academic and media institutions. And, as Hsu points out, the most vibrant and influential aspect of American popular culture is hip-hop�a product of the African American urban culture. The only significant group of white people with any cultural confidence centers itself around country music, NASCAR, and the small town values of traditional white America. For this group of whites � and only this group � there is "a racial pride that dares not speak its name, and that defines itself through cultural cues instead�a suspicion of intellectual elites and city dwellers, a preference for folksiness and plainness of speech (whether real or feigned), and the association of a working-class white minority with 'the real America.'� This is what I term implicit whiteness � implicit because explicit assertions of white identity have been banned by the anti-white elites that dominate our politics and culture. It is a culture that, as Hsu notes, "cannot speak its name." But that implies that the submerged white identity of the white working class and the lack of cultural confidence exhibited by the rest of white America are imposed from outside. Although there may well be characteristics of whites that facilitate this process, this suppression of white identity and interests is certainly not the natural outcome of modernization or any other force internal to whites as a people. In my opinion, it is the result of the successful erection of a culture of critique in the West dominated by Jewish intellectual and political movements. The result is that educated, intellectually insecure white people these days are far more likely to believe in the utopian future described by Hsu than in hard and cautious thinking about what the future might have in store for them. It's worth dwelling a bit on the intellectual insecurity of the whites who mindlessly utter the mantras of multiculturalism that they have soaked up from the school system and from the media. Most people do not have much confidence in their intellectual ability and look to elite opinion to shape their beliefs. As I noted elsewhere, A critical component of the success of the culture of critique is that it achieved control of the most prestigious and influential institutions of the West, and it became a consensus among the elites, Jewish and non-Jewish alike. Once this happened, it is not surprising that this culture became widely accepted among people of very different levels of education and among people of different social classes. Most people are quite insecure about their intellectual ability. But they know that the professors at Harvard, and the editorial page of the New York Times and the Washington Post, and even conservative commentators like Rush Limbaugh and Sean Hannity are all on page when it comes to racial and ethnic issues. This is a formidable array, to the point that you almost have to be a crank to dissent from this consensus. I think one of the greatest triumphs of the left has been to get people to believe that people who assert white identity and interests or who make unflattering portrayals of organized Jewish movements are morally degenerate, stupid, and perhaps psychiatrically disturbed. Obviously, all of these adjectives designate low status. The reality is that the multicultural emperor has no clothes and, because of its support for racial Zionism and the racialism of ethnic minorities in America, it is massively hypocritical to boot. The New York Times, the academic left, and the faux conservatives that dominate elite discourse on race and ethnicity are intellectually bankrupt and can only remain in power by ruthlessly suppressing or ignoring the scientific findings. This is particularly a problem for college-educated whites. Like Fitzgerald's Tom Buchanan, such people have a strong need to feel that their ideas are respectable and part of the mainstream. But the respectable mainstream gives them absolutely nothing with which to validate themselves except perhaps the idea that the world will be a better place when people like them no longer have power. Hsu quotes the pathetic Christian Lander: "�Like, I�m aware of all the horrible crimes that my demographic has done in the world. ... And there�s a bunch of white people who are desperate � desperate � to say, �You know what? My skin�s white, but I�m not one of the white people who�s destroying the world.�� As a zombie leftist during the 1960s and 1970s, I know what that feeling of desperation is like � what it's like to be a self-hating white. We must get to the point where college-educated whites proudly and confidently say they are white and that they do not want to become a minority in America. This reminds me of the recent docudrama Milk, which depicts the life of gay activist Harvey Milk. Milk is sure be nominated for an Oscar as Best Picture because it lovingly illustrates a triumph of the cultural left. But is has an important message that should resonate with the millions of whites who have been deprived of their confidence and their culture: Be explicit. Just as Harvey Milk advocated being openly gay even in the face of dire consequences, whites need to tell their family and their friends that they have an identity as a white person and believe that whites have legitimate interests as white people. They must accept the consequences when they are harassed, fired from their jobs, or put in prison for such beliefs. They must run for political office as openly pro-white. Milk shows that homosexuals were fired from their jobs and arrested for congregating in public. Now it's the Southern Poverty Law Center and the rest of the leftist intellectual and political establishment that harasses and attempts to get people fired. But it's the same situation with the roles reversed. No revolution was ever accomplished without some martyrs. The revolution that restores the legitimacy of white identity and the legitimacy of white interests will be no exception. But it is a revolution that is absolutely necessary. The white majority is foolish indeed to entrust its future to a utopian hope that racial and ethnic identifications will disappear and that they won�t continue to influence public policy in ways that compromise the interests of whites. It does not take an overactive imagination to see that coalitions of minority groups could compromise the interests of formerly dominant whites. We already see numerous examples in which coalitions of minority groups attempt to influence public policy, including immigration policy, against the interests of the whites. Placing ourselves in a position of vulnerability would be extremely risky, given the deep sense of historical grievance fostered by many ethnic activists and organized ethnic lobbies. This is especially the case with Jews. Jewish organisations have been unanimous in condemning Western societies, Western traditions, and Christianity, for past crimes against Jews. Similar sentiments are typical of a great many African Americans and Latinos, and especially among the ethnic activists from these groups. The �God damn America� sermon by President Obama's pastor comes to mind as a recent notorious example. The precedent of the early decades of the Soviet Union should give pause to anyone who believes that surrendering ethnic hegemony does not carry risks. The Bolshevik revolution had a pronounced ethnic angle: To a very great extent, Jews and other non-Russians ruled over the Russian people, with disastrous consequences for the Russians and other ethnic groups that were not able to become part of the power structure. Jews formed a hostile elite within this power structure � as they will in the future white-minority America; Jews were �Stalin�s willing executioners.� Two passages from my review of Yuri Slezkine's The Jewish Century seem particularly appropriate here. The first passage reminds me of the many American Jews who adopt a veneer of support for leftist versions of social justice and racial tolerance while nevertheless managing to support racial Zionism and the mass murder, torture, and incarceration of the Palestinian people in one of the largest prison systems the world has ever seen. Such people may be very different when they become a hostile elite in a white-minority America. Many of the commentators on Jewish Bolsheviks noted the �transformation� of Jews [after the Bolshevik Revolution]. In the words of [a] Jewish commentator, G. A. Landau, �cruelty, sadism, and violence had seemed alien to a nation so far removed from physical activity.� And another Jewish commentator, Ia. A. Bromberg, noted that: the formerly oppressed lover of liberty had turned into a tyrant of �unheard-of-despotic arbitrariness��. The convinced and unconditional opponent of the death penalty not just for political crimes but for the most heinous offenses, who could not, as it were, watch a chicken being killed, has been transformed outwardly into a leather-clad person with a revolver and, in fact, lost all human likeness. ... After the Revolution, ... there was active suppression of any remnants of the older order and their descendants. ... The mass murder of peasants and nationalists was combined with the systematic exclusion of the previously existing non-Jewish middle class. The wife of a Leningrad University professor noted, �in all the institutions, only workers and Israelites are admitted; the life of the intelligentsia is very hard� (p. 243). Even at the end of the 1930s, prior to the Russification that accompanied World War II, �the Russian Federation�was still doing penance for its imperial past while also serving as an example of an ethnicity-free society� (p. 276). While all other nationalities, including Jews, were allowed and encouraged to keep their ethnic identities, the revolution remained an anti-majoritarian movement. The difference from the Soviet Union may well be that in white-minority America it will not be workers and Israelites who are favored, but non-whites and Israelites. Whites may dream that they are entering the post-racial utopia imagined by their erstwhile intellectual superiors. But it is quite possible that they are entering into a racial dystopia of unimaginable cruelty in which whites will be systematically excluded in favor of the new elites recruited from the soon-to-be majority. It's happened before. Kevin MacDonald is a professor of psychology at California State University�Long Beach. Permanent URL with hyperlinks: http://www.theoccidentalobserver.net/articles/MacDonald-Hsu.html ----------- Should Christians Support Israeli Terrorism in Gaza? A timely discussion between Rev. Ted Pike and Dr. David Duke, one especially important for the Christians in our audience http://www.davidduke.com/mp3/dukeradio090122DukeandPikeonGaza.mp3 In this vital discussion, Rev. Pike and Dr. Duke explore the Pro-Israel attitude of some Christian evangelical organizations, and why their position not only goes directly against Christian morality and decency, but actually is directly opposite of that expressed by Christian Scriptures. Today, Many Christians are instructed that Jews and today�s Israel has a special covenant� with God. In fact, the New Testament in the clearest of language states that the Jews �continued not in my covenant, and I considered them not, saith the Lord.� Here�s the quote that Christians aren�t supposed to notice.: 8:10 Not according to the covenant that I made with their fathers, in the day when I took them by the hand out of the land of Egypt; because they continued not in my covenant, and I regarded them not, saith the Lord. (Hebrews 8:10) They also don�t seem to notice that a 2000 year old Judaic war against Christianity that has been waged since time of Jesus Christ and still goes on today with the most powerful Jewish organizations attempting to destroy European and American traditions, that has even become a war on our Christmas traditions. Dr. Duke and Ted Pike also speak about how over a hundred thousand Christian Palestinians have suffered with their families from anti-Christian Israel! Christian support of Israel has resulted in the very birthplace of Jesus Christ, go from 90 percent Palestinian Christians to 35 percent today because of Israeli terror and occupation. They ask, �How could any Christian in good conscience support the anti-Christian state of Israel, bombing the homes, killing and maiming, torturing and oppressing fellow Christian men, women and children?� This is a vital show for every Christian reader and listener of DavidDuke.com. Next time, you hear someone say, �God tells us that we must support Israel� you will have the clear Christian answer that just the opposite is true! For documentation on this be sure to read some of the well-footnoted, sample chapters of Jewish Supremacism and My Awakening. Source : http://www.davidduke.com/general/should-christians-support-israeli-terroris… ------------------------------------- You or someone using your email adress is currently subscribed to the Lawrence Auster Newletter. If you wish to unsubscribe from our mailing list, please let us know by calling to 1 212 865 1284 Thanks, Lawrence Auster, 238 W 101 St Apt. 3B New York, NY 10025 Contact: lawrence.auster(a)att.net -------------------------------------

1 0

[common-math-cvs] r4 - trunk/common-math/numerics/linear-algebra
by mantoniotti＠common-lisp.net 23 Sep '06

23 Sep '06

Author: mantoniotti Date: Sat Sep 23 18:23:10 2006 New Revision: 4 Modified: trunk/common-math/numerics/linear-algebra/inversion.lisp trunk/common-math/numerics/linear-algebra/lu-decomposition.lisp Log: Minor changes. Modified: trunk/common-math/numerics/linear-algebra/inversion.lisp ============================================================================== --- trunk/common-math/numerics/linear-algebra/inversion.lisp (original) +++ trunk/common-math/numerics/linear-algebra/inversion.lisp Sat Sep 23 18:23:10 2006 @@ -20,10 +20,15 @@ (inverse (matrix-data a))) +(defmethod inverse ((a array)) (error "Unimplemented ~S" a)) (defmethod inverse/double-float ((a matrix)) (inverse/double-float (matrix-data a))) +(defmethod inverse/double-float ((a array)) (error "Unimplemented ~S" a)) + + + ;;; end of file -- inversion.lisp -- Modified: trunk/common-math/numerics/linear-algebra/lu-decomposition.lisp ============================================================================== --- trunk/common-math/numerics/linear-algebra/lu-decomposition.lisp (original) +++ trunk/common-math/numerics/linear-algebra/lu-decomposition.lisp Sat Sep 23 18:23:10 2006 @@ -243,9 +243,8 @@ do (cond ((not (cl:zerop ii)) (loop for j of-type fixnum from ii below (1- i) do (decf sum (* (aref lu i j) (aref b i))))) - ((not (cl:zerop sum)) - (setf (aref b i) sum))) - ) + ((not (cl:zerop sum)) (setf ii i))) + do (setf (aref b i) sum)) (loop for i of-type fixnum from (1- n) downto 0 for sum of-type number = (aref b i) do (loop for j of-type fixnum from (1+ i) below n

1 0

[common-math-cvs] r3 - trunk/common-math/numerics/linear-algebra
by mantoniotti＠common-lisp.net 18 Aug '06

18 Aug '06

Author: mantoniotti Date: Fri Aug 18 11:54:06 2006 New Revision: 3 Added: trunk/common-math/numerics/linear-algebra/determinants.lisp trunk/common-math/numerics/linear-algebra/inversion.lisp Modified: trunk/common-math/numerics/linear-algebra/linear-algebra-pkg.lisp trunk/common-math/numerics/linear-algebra/linear-algebra.system trunk/common-math/numerics/linear-algebra/lu-decomposition.lisp trunk/common-math/numerics/linear-algebra/matrix.lisp Log: Rearranged a few files, added two more and fixed a small number of issues here and there. Added: trunk/common-math/numerics/linear-algebra/determinants.lisp ============================================================================== --- (empty file) +++ trunk/common-math/numerics/linear-algebra/determinants.lisp Fri Aug 18 11:54:06 2006 @@ -0,0 +1,44 @@ +;;;; -*- Mode: Lisp -*- + +;;;; determinants.lisp -- + + +(in-package "CL.MATH.LINEAR-ALGEBRA") + +;;;=========================================================================== +;;; Determinants + +;;; det + +(defgeneric det (m)) + +(defgeneric det/double-float (m)) + + +(defmethod det ((a array)) + (assert (square-matrix-p a)) + (multiple-value-bind (lu permutations parity) + (lu-decompose a) + (declare (ignore permutations)) + (dotimes (i (array-dimension lu 0) parity) + (setf parity (* parity (aref lu i i)))))) + +(defmethod det ((a matrix)) + (det (matrix-data a))) + + +(defmethod det/double-float ((a array)) + (assert (square-matrix-p a)) + (multiple-value-bind (lu permutations parity) + (lu-decompose a) + (declare (type (array double-float (cl:* cl:*)) lu) + (ignore permutations) + (type double-float parity)) + (dotimes (i (array-dimension lu 0) parity) + (setf parity (* parity (aref lu i i)))))) + +(defmethod det/double-float ((a matrix)) + (det/double-float (matrix-data a))) + + +;;; end of file -- determinant.lisp -- Added: trunk/common-math/numerics/linear-algebra/inversion.lisp ============================================================================== --- (empty file) +++ trunk/common-math/numerics/linear-algebra/inversion.lisp Fri Aug 18 11:54:06 2006 @@ -0,0 +1,29 @@ +;;;; -*- Mode: Lisp -*- + +;;;; inversion.lisp -- + + +(in-package "CL.MATH.LINEAR-ALGEBRA") + +;;;=========================================================================== +;;; Inversion + +;;; inverse + +(defgeneric inverse (m)) + +(defgeneric inverse/double-float (m)) + + + +(defmethod inverse ((a matrix)) + (inverse (matrix-data a))) + + + + +(defmethod inverse/double-float ((a matrix)) + (inverse/double-float (matrix-data a))) + + +;;; end of file -- inversion.lisp -- Modified: trunk/common-math/numerics/linear-algebra/linear-algebra-pkg.lisp ============================================================================== --- trunk/common-math/numerics/linear-algebra/linear-algebra-pkg.lisp (original) +++ trunk/common-math/numerics/linear-algebra/linear-algebra-pkg.lisp Fri Aug 18 11:54:06 2006 @@ -13,6 +13,8 @@ (:export ".*" "./" "EXPT*") (:export ".*%2" ".*%1" "./%2" "./%1") + + (:export "DET" "SOLVE") ) Modified: trunk/common-math/numerics/linear-algebra/linear-algebra.system ============================================================================== --- trunk/common-math/numerics/linear-algebra/linear-algebra.system (original) +++ trunk/common-math/numerics/linear-algebra/linear-algebra.system Fri Aug 18 11:54:06 2006 @@ -8,6 +8,9 @@ "common-math-extra" "vector" "matrix" - "lu-decomposition")) + (:file "lu-decomposition" :depends-on ("matrix")) + (:file "determinants" :depends-on ("lu-decomposition")) + (:file "inversion" :depends-on ("lu-decomposition")) + )) ;;;; end of file -- linear-algebra.system -- Modified: trunk/common-math/numerics/linear-algebra/lu-decomposition.lisp ============================================================================== --- trunk/common-math/numerics/linear-algebra/lu-decomposition.lisp (original) +++ trunk/common-math/numerics/linear-algebra/lu-decomposition.lisp Fri Aug 18 11:54:06 2006 @@ -14,13 +14,6 @@ (defparameter *epsilon* 1.0e-20) ; Change this to CL constant. -(defgeneric square-matrix-p (m) - (:method ((m array)) - (and (matrix-array-p m) - (cl:= (array-dimension m 0) (array-dimension m 1)))) - (:method ((m matrix)) - (cl:= (matrix-row-number m) (matrix-column-number m)))) - ;;; Different base type. @@ -283,33 +276,11 @@ (solve/double-float (matrix-data a) b result)) -;;;--------------------------------------------------------------------------- -;;; det - -(defmethod det ((a array)) - (assert (square-matrix-p a)) - (multiple-value-bind (lu permutations parity) - (lu-decompose a) - (declare (ignore permutations)) - (dotimes (i (array-dimension lu 0) parity) - (setf parity (* parity (aref lu i i)))))) - -(defmethod det ((a matrix)) - (det (matrix-data a))) - - -(defmethod det/double-float ((a array)) - (assert (square-matrix-p a)) - (multiple-value-bind (lu permutations parity) - (lu-decompose a) - (declare (type (array double-float (cl:* cl:*)) lu) - (ignore permutations) - (type double-float parity)) - (dotimes (i (array-dimension lu 0) parity) - (setf parity (* parity (aref lu i i)))))) +;;;=========================================================================== +;;; Determinants -(defmethod det/double-float ((a matrix)) - (det/double-float (matrix-data a))) +;;;=========================================================================== +;;; Inverse ;;;=========================================================================== Modified: trunk/common-math/numerics/linear-algebra/matrix.lisp ============================================================================== --- trunk/common-math/numerics/linear-algebra/matrix.lisp (original) +++ trunk/common-math/numerics/linear-algebra/matrix.lisp Fri Aug 18 11:54:06 2006 @@ -146,6 +146,14 @@ (shape-equal-p m1 m2)) + +(defgeneric square-matrix-p (m) + (:method ((m array)) + (cl:= (matrix-row-number m) (matrix-column-number m))) + (:method ((m matrix)) + (cl:= (matrix-row-number m) (matrix-column-number m)))) + + ;;;--------------------------------------------------------------------------- ;;; Creation.

1 0

[common-math-cvs] r2 - in trunk/common-math: . numerics/linear-algebra
by mantoniotti＠common-lisp.net 17 Aug '06

17 Aug '06

Author: mantoniotti Date: Thu Aug 17 12:39:36 2006 New Revision: 2 Modified: trunk/common-math/common-math.lisp trunk/common-math/numerics/linear-algebra/matrix.lisp Log: Added a few missing operations (.*%2 ./%2 .%\\2) and fixed a couple of loose things. Modified: trunk/common-math/common-math.lisp ============================================================================== --- trunk/common-math/common-math.lisp (original) +++ trunk/common-math/common-math.lisp Thu Aug 17 12:39:36 2006 @@ -52,6 +52,8 @@ ;;; purposes and setting up a separate readtable seems a little bit ;;; too much for the time being. +;;; <%2 + (defgeneric <%2 (x y) (:documentation "The <%2 generic function. The binary LESS generic function which is specialized for various @@ -90,6 +92,12 @@ ) +(defmethod <%2 ((x symbol) (y symbol)) + (string< x y)) + + +;;; >%2 + (defgeneric >%2 (x y) (:method ((x number) (y number)) (cl:> x y)) @@ -122,14 +130,13 @@ ) -(defmethod <%2 ((x symbol) (y symbol)) - (string< x y)) - (defmethod >%2 ((x symbol) (y symbol)) (string> y x)) +;;; <=%2 + (defgeneric <=%2 (x y) (:method ((x number) (y number)) (cl:<= x y)) @@ -162,6 +169,8 @@ ) +;;; >=%2 + (defgeneric >=%2 (x y) (:method ((x number) (y number)) (cl:>= x y)) @@ -194,6 +203,8 @@ ) +;;; +%2 + (defgeneric +%2 (x y &optional r) (:method ((x number) (y number) &optional r) (declare (ignore r)) @@ -237,6 +248,8 @@ ) +;;; *%2 + (defgeneric *%2 (x y &optional r) (:method ((x number) (y number) &optional r) (declare (ignore r)) @@ -289,12 +302,16 @@ ) +;;; -%2 + (defgeneric -%2 (x y &optional r) (:method ((x t) (y t) &optional r) (declare (ignore r)) (+%2 x (-%1 y)))) +;;; /%2 + (defgeneric /%2 (x y &optional r) (:method ((x number) (y number) &optional r) (declare (ignore r)) @@ -305,6 +322,8 @@ (*%2 x (/%1 y)))) +;;; +%1 + (defgeneric +%1 (x &optional r) (:method ((x number) &optional r) (declare (ignore r)) @@ -320,6 +339,9 @@ ) + +;;; *%1 + (defgeneric *%1 (x &optional r) (:method ((x number) &optional r) (declare (ignore r)) @@ -335,6 +357,9 @@ ) + +;;; -%1 + (defgeneric -%1 (x &optional r) (:method ((x number) &optional r) (declare (ignore r)) @@ -350,6 +375,9 @@ ) + +;;; /%1 + (defgeneric /%1 (x &optional r) (:method ((x number) &optional r) (declare (ignore r)) @@ -365,6 +393,8 @@ ) +;;; =2% + (defgeneric =%2 (x y) (:method ((x number) (y number)) (cl:= x y)) @@ -396,11 +426,17 @@ ) +;;; =%1 + (defgeneric =%1 (x) (:method ((x t)) T) ) +;;;--------------------------------------------------------------------------- +;;; Other operations + + (defgeneric gcd%2 (x y) (:method ((x integer) (y integer)) (cl:gcd x y))) Modified: trunk/common-math/numerics/linear-algebra/matrix.lisp ============================================================================== --- trunk/common-math/numerics/linear-algebra/matrix.lisp (original) +++ trunk/common-math/numerics/linear-algebra/matrix.lisp Thu Aug 17 12:39:36 2006 @@ -773,20 +773,10 @@ (defmethod *%2 ((y array) (x number) &optional (r (copy-matrix y) r-supplied-p)) + (declare (ignore r-supplied-p)) (*%2 x y r)) -;;; The next one breaks the return type convention. -#| Defined in 'vector.lisp'. -(defmethod *%2 ((x vector) (y vector) &optional r) - (declare (ignore r)) - (assert (conforming-*-dimensions-p x y nil)) - (let ((result 0)) - (dotimes (i (length x) result) - (setf result (+%2 result (* (aref x i) (aref y i))))))) -|# - - ;;;--------------------------------------------------------------------------- ;;; Division. ;;; Only the simple form of division is implemented here. @@ -810,6 +800,8 @@ ;;;--------------------------------------------------------------------------- ;;; Element-wise operations. +;;; .*%2 + (defmethod .*%2 ((x number) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) (when r-supplied-p (assert (shape-equal-p y r))) @@ -850,6 +842,187 @@ (*%2 (row-major-aref x i) (row-major-aref y i))))) +(defmethod .*%2 ((x matrix) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (assert (shape-equal-p x y)) + + (with-slots ((x-data data)) x + (with-slots ((y-data data)) y + (with-slots ((result data)) r + + (dotimes (i (array-total-size result) r) + (setf (row-major-aref result i) + (*%2 (row-major-aref x-data i) (row-major-aref y-data i))))) + ))) + + +(defmethod .*%2 ((x matrix) (y array) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (assert (shape-equal-p x y)) + (.*%2 (matrix-data x) y r)) + + +(defmethod .*%2 ((x array) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (assert (shape-equal-p x y)) + + (with-slots ((y-data data)) y + (with-slots ((result data)) r + (dotimes (i (array-total-size result) r) + (setf (row-major-aref result i) + (*%2 (row-major-aref x i) (row-major-aref y-data i)))) + ))) + + + +;;; ./%2 + +(defmethod ./%2 ((x number) (y matrix) + &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (with-slots (data) y + (with-slots ((result data)) r + (dotimes (i (array-total-size data) r) + (setf (row-major-aref result i) (/%2 x (row-major-aref data i)))) + ))) + + +(defmethod ./%2 ((y matrix) (x number) &optional (r (copy-matrix y))) + (.*%2 y (cl:/ x) r)) + + +(defmethod ./%2 ((x number) (y array) + &optional (r (copy-matrix y) r-supplied-p)) + (assert (matrix-array-p y)) + (when r-supplied-p (assert (shape-equal-p y r))) + (let* ((data y) + (result r) + ) + (dotimes (i (array-total-size data) r) + (setf (row-major-aref result i) (/%2 x (row-major-aref data i)))) + )) + + +(defmethod ./%2 ((y array) (x number) &optional (r (copy-matrix y))) + (.*%2 y (cl:/ x) r)) + + + +(defmethod ./%2 ((x array) (y array) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (assert (shape-equal-p x y)) + + (dotimes (i (array-total-size y) r) + (setf (row-major-aref r i) + (/%2 (row-major-aref x i) (row-major-aref y i))))) + + +(defmethod ./%2 ((x matrix) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (assert (shape-equal-p x y)) + + (with-slots ((x-data data)) x + (with-slots ((y-data data)) y + (with-slots ((result data)) r + + (dotimes (i (array-total-size result) r) + (setf (row-major-aref result i) + (/%2 (row-major-aref x-data i) (row-major-aref y-data i))))) + ))) + + +(defmethod ./%2 ((x matrix) (y array) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (assert (shape-equal-p x y)) + (./%2 (matrix-data x) y r)) + + +(defmethod ./%2 ((x array) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (assert (shape-equal-p x y)) + + (with-slots ((y-data data)) y + (with-slots ((result data)) r + (dotimes (i (array-total-size result) r) + (setf (row-major-aref result i) + (/%2 (row-major-aref x i) (row-major-aref y-data i)))) + ))) + + +;;; .\\%2 + +(defmethod .\\%2 ((x number) (y matrix) + &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (with-slots (data) y + (with-slots ((result data)) r + (dotimes (i (array-total-size data) r) + (setf (row-major-aref result i) (/%2 (row-major-aref data i) x))) + ))) + + +(defmethod .\\%2 ((y matrix) (x number) &optional (r (copy-matrix y))) + (./%2 x y r)) + + +(defmethod .\\%2 ((x number) (y array) + &optional (r (copy-matrix y) r-supplied-p)) + (assert (matrix-array-p y)) + (when r-supplied-p (assert (shape-equal-p y r))) + (let* ((data y) + (result r) + ) + (dotimes (i (array-total-size data) r) + (setf (row-major-aref result i) (/%2 (row-major-aref data i) x))) + )) + + +(defmethod .\\%2 ((y array) (x number) &optional (r (copy-matrix y))) + (./%2 x y r)) + + + +(defmethod .\\%2 ((x array) (y array) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (assert (shape-equal-p x y)) + + (dotimes (i (array-total-size y) r) + (setf (row-major-aref r i) + (/%2 (row-major-aref y i) (row-major-aref x i))))) + + +(defmethod .\\%2 ((x matrix) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (assert (shape-equal-p x y)) + + (with-slots ((x-data data)) x + (with-slots ((y-data data)) y + (with-slots ((result data)) r + + (dotimes (i (array-total-size result) r) + (setf (row-major-aref result i) + (/%2 (row-major-aref y-data i) (row-major-aref x-data i))))) + ))) + + +(defmethod .\\%2 ((x matrix) (y array) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (assert (shape-equal-p x y)) + (.\\%2 (matrix-data x) y r)) + + +(defmethod .\\%2 ((x array) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (assert (shape-equal-p x y)) + + (with-slots ((y-data data)) y + (with-slots ((result data)) r + (dotimes (i (array-total-size result) r) + (setf (row-major-aref result i) + (/%2 (row-major-aref y-data i) (row-major-aref x i)))) + ))) + + ;;;--------------------------------------------------------------------------- ;;; Transpose

1 0

[common-math-cvs] r1 - in trunk: . common-math common-math/numerics common-math/numerics/linear-algebra
by mantoniotti＠common-lisp.net 16 Aug '06

16 Aug '06

Author: mantoniotti Date: Wed Aug 16 17:07:41 2006 New Revision: 1 Added: trunk/ trunk/common-math/ trunk/common-math/common-math-pkg.lisp trunk/common-math/common-math-user-pkg.lisp trunk/common-math/common-math.lisp trunk/common-math/common-math.system trunk/common-math/numerics/ trunk/common-math/numerics/linear-algebra/ trunk/common-math/numerics/linear-algebra/common-math-extra.lisp trunk/common-math/numerics/linear-algebra/linear-algebra-pkg.lisp trunk/common-math/numerics/linear-algebra/linear-algebra.system trunk/common-math/numerics/linear-algebra/lu-decomposition.lisp trunk/common-math/numerics/linear-algebra/matrix-pkg.lisp trunk/common-math/numerics/linear-algebra/matrix.lisp trunk/common-math/numerics/linear-algebra/vector-matrix-conditions.lisp trunk/common-math/numerics/linear-algebra/vector-matrix.lisp trunk/common-math/numerics/linear-algebra/vector.lisp trunk/common-math/numerics/numerics.system Log: Initial import. Added: trunk/common-math/common-math-pkg.lisp ============================================================================== --- (empty file) +++ trunk/common-math/common-math-pkg.lisp Wed Aug 16 17:07:41 2006 @@ -0,0 +1,47 @@ +;;;; -*- Mode: Lisp -*- + +;;;; common-math-pkg.lisp -- + +(defpackage "COMMON-MATH" (:use "COMMON-LISP") + (:documentation "The COMMON MATH Package. +This package contains `top level' definitions of a `generic math' +package which can be used as a drop in replacement for the standard +CL mathematics routines. Note however, that this will of course +come at a price. + +The COMMON-MATH package addresses the problem of having a generic and +as complete as possible mathematics interface layer over a number of +different algorithms and representations. In this sense it aspires +to bring to CL the flexibility of generic math operations, while +preserving the low level hooks that could be used for (relative) efficiency.") + + (:nicknames "CL.MATH" "CL.MATHEMATICS") + + (:shadow "=" "+" "-" "*" "/" ">" "<" ">=" "<=") + (:shadow "ZEROP" "EXPT" "GCD") + + (:export "=" "+" "-" "*" "/" ">" "<" ">=" "<=") + (:export "ZEROP" "EXPT" "GCD") + + (:export "+POSITIVE-INFINITY+" "+NEGATIVE-INFINITY+") + + (:export "INCREASING" "DECREASING") + + (:export + "+%1" "+%2" + "*%1" "*%2" + "-%1" "-%2" + "/%1" "/%2" + "=%1" "=%2" + "<%2" + ">%2" + "<=%2" + ">=%2" + ) + + ;; Conditions + (:export "UNDEFINED-OPERATION") + + ) + +;;;; end of file -- common-math-pkg.lisp -- Added: trunk/common-math/common-math-user-pkg.lisp ============================================================================== --- (empty file) +++ trunk/common-math/common-math-user-pkg.lisp Wed Aug 16 17:07:41 2006 @@ -0,0 +1,12 @@ +;;;; -*- Mode: Lisp -*- + +;;;; common-math-user-pkg.lisp -- +;;;; The COMMON-MATH-USER package. + +(defpackage "COMMON-MATH-USER" (:use "COMMON-MATH" "CL" #| "LINALG" |# ) + (:shadowing-import-from "COMMON-MATH" "+" "-" "*" "/") + (:shadowing-import-from "COMMON-MATH" "<" ">" "<=" ">=" "=") + (:shadowing-import-from "COMMON-MATH" "ZEROP" "EXPT" "GCD") + ) + +;;;; end of file -- common-math-user-pkg.lisp -- \ No newline at end of file Added: trunk/common-math/common-math.lisp ============================================================================== --- (empty file) +++ trunk/common-math/common-math.lisp Wed Aug 16 17:07:41 2006 @@ -0,0 +1,603 @@ +;;;; -*- Mode: Lisp -*- + +;;;; common-math.lisp -- + +(in-package "CL.MATHEMATICS") + +;;; undefined-operation -- + +(define-condition undefined-operation (error) + ((operator :reader undefined-operation-operator + :initarg :operator) + (arguments :reader undefined-operation-arguments + :initarg :arguments) + ) + (:documentation "The Undefined Operation Condition.") + (:report (lambda (uoc stream) + (format stream "Undefined operation ~S called with ~S." + (undefined-operation-operator uoc) + (undefined-operation-arguments uoc)))) + (:default-initargs :arguments () :operator nil)) + + +;;;--------------------------------------------------------------------------- +;;; Constants and parameters. + +(defconstant +negative-infinity+ '+negative-infinity+ + "The symbolic +NEGATIVE-INFINITY+ constant. +A representation of negative infinity.") +(defconstant +positive-infinity+ '+positive-infinity+ + "The symbolic +POSITIVE-INFINITY+ constant. +A representation of positive infinity.") + +(defparameter *ignore-comparison-errors-p* nil) + + +;;; (defconstant +negative-infinity+ most-negative-fixnum) +;;; (defconstant +positive-infinity+ most-positive-fixnum) + + +;;;--------------------------------------------------------------------------- +;;; Generic functions. +;;; Note that the generic functions corresponding to traditional math +;;; operators are named using a special notation to indicate their +;;; arity. Such 'operator' generic function are named as +;;; +;;; <op>%<arity> +;;; +;;; The #\% character in the name is used to suggest "low level" as +;;; usual. The Prolog-like #\/ could have been used instead but the +;;; #\% has more tradition in CL circles. The #\@ would have made +;;; more sense, but some CL implementations already use it for other +;;; purposes and setting up a separate readtable seems a little bit +;;; too much for the time being. + +(defgeneric <%2 (x y) + (:documentation "The <%2 generic function. +The binary LESS generic function which is specialized for various +combinations of argument types. + +At a minimum there is a guarantee that the method on two NUMBER +arguments dispatching on CL:< is defined.") + (:method ((x number) (y number)) + (cl:< x y)) + + (:method ((x number) (y (eql +negative-infinity+))) + nil) + + (:method ((y (eql +negative-infinity+)) (x number)) + T) + + (:method ((y (eql +negative-infinity+)) (x (eql +negative-infinity+))) + (unless *ignore-comparison-errors-p* + (error 'undefined-operation :operator 'lessp%2))) + + (:method ((y (eql +negative-infinity+)) (x (eql +positive-infinity+))) + T) + + (:method ((y (eql +positive-infinity+)) (x (eql +negative-infinity+))) + nil) + + (:method ((x number) (y (eql +positive-infinity+))) + T) + + (:method ((y (eql +positive-infinity+)) (x number)) + nil) + + (:method ((y (eql +positive-infinity+)) (x (eql +positive-infinity+))) + (unless *ignore-comparison-errors-p* + (error 'undefined-operation :operator 'lessp%2))) + ) + + +(defgeneric >%2 (x y) + (:method ((x number) (y number)) + (cl:> x y)) + + (:method ((x number) (y (eql +negative-infinity+))) + T) + + (:method ((y (eql +negative-infinity+)) (x number)) + nil) + + (:method ((y (eql +negative-infinity+)) (x (eql +negative-infinity+))) + (unless *ignore-comparison-errors-p* + (error 'undefined-operation :operator 'lessp%2))) + + (:method ((y (eql +negative-infinity+)) (x (eql +positive-infinity+))) + nil) + + (:method ((y (eql +positive-infinity+)) (x (eql +negative-infinity+))) + T) + + (:method ((x number) (y (eql +positive-infinity+))) + nil) + + (:method ((y (eql +positive-infinity+)) (x number)) + T) + + (:method ((y (eql +positive-infinity+)) (x (eql +positive-infinity+))) + (unless *ignore-comparison-errors-p* + (error 'undefined-operation :operator 'lessp%2))) + ) + + +(defmethod <%2 ((x symbol) (y symbol)) + (string< x y)) + + +(defmethod >%2 ((x symbol) (y symbol)) + (string> y x)) + + +(defgeneric <=%2 (x y) + (:method ((x number) (y number)) + (cl:<= x y)) + + (:method ((x number) (y (eql +negative-infinity+))) + nil) + + (:method ((y (eql +negative-infinity+)) (x number)) + T) + + (:method ((y (eql +negative-infinity+)) (x (eql +negative-infinity+))) + (unless *ignore-comparison-errors-p* + (error 'undefined-operation :operator 'lessp%2))) + + (:method ((y (eql +negative-infinity+)) (x (eql +positive-infinity+))) + T) + + (:method ((y (eql +positive-infinity+)) (x (eql +negative-infinity+))) + nil) + + (:method ((x number) (y (eql +positive-infinity+))) + T) + + (:method ((y (eql +positive-infinity+)) (x number)) + nil) + + (:method ((y (eql +positive-infinity+)) (x (eql +positive-infinity+))) + (unless *ignore-comparison-errors-p* + (error 'undefined-operation :operator 'lessp%2))) + ) + + +(defgeneric >=%2 (x y) + (:method ((x number) (y number)) + (cl:>= x y)) + + (:method ((x number) (y (eql +negative-infinity+))) + T) + + (:method ((y (eql +negative-infinity+)) (x number)) + nil) + + (:method ((y (eql +negative-infinity+)) (x (eql +negative-infinity+))) + (unless *ignore-comparison-errors-p* + (error 'undefined-operation :operator 'lessp%2))) + + (:method ((y (eql +negative-infinity+)) (x (eql +positive-infinity+))) + nil) + + (:method ((y (eql +positive-infinity+)) (x (eql +negative-infinity+))) + T) + + (:method ((x number) (y (eql +positive-infinity+))) + nil) + + (:method ((y (eql +positive-infinity+)) (x number)) + T) + + (:method ((y (eql +positive-infinity+)) (x (eql +positive-infinity+))) + (unless *ignore-comparison-errors-p* + (error 'undefined-operation :operator 'lessp%2))) + ) + + +(defgeneric +%2 (x y &optional r) + (:method ((x number) (y number) &optional r) + (declare (ignore r)) + (cl:+ x y)) + + (:method ((x number) (y (eql +negative-infinity+)) &optional r) + (declare (ignore r)) + +negative-infinity+) + + (:method ((y (eql +negative-infinity+)) (x number) &optional r) + (declare (ignore r)) + +negative-infinity+) + + (:method ((y (eql +negative-infinity+)) (x (eql +negative-infinity+)) + &optional r) + (declare (ignore r)) + +negative-infinity+) + + (:method ((y (eql +negative-infinity+)) (x (eql +positive-infinity+)) + &optional r) + (declare (ignore r)) + (error 'undefined-operation :operator '+%2)) + + (:method ((x number) (y (eql +positive-infinity+)) &optional r) + (declare (ignore r)) + +positive-infinity+) + + (:method ((y (eql +positive-infinity+)) (x number) &optional r) + (declare (ignore r)) + +positive-infinity+) + + (:method ((y (eql +positive-infinity+)) (x (eql +positive-infinity+)) + &optional r) + (declare (ignore r)) + +negative-infinity+) + + (:method ((y (eql +positive-infinity+)) (x (eql +negative-infinity+)) + &optional r) + (declare (ignore r)) + (error 'undefined-operation :operator '+%2)) + ) + + +(defgeneric *%2 (x y &optional r) + (:method ((x number) (y number) &optional r) + (declare (ignore r)) + (cl:* x y)) + + (:method ((x number) (y (eql +negative-infinity+)) &optional r) + (declare (ignore r)) + (cond ((zerop x) (error 'undefined-operation :operator '*%2)) + ((plusp x) +negative-infinity+) + (t +positive-infinity+))) + + (:method ((y (eql +negative-infinity+)) (x number) &optional r) + (declare (ignore r)) + (cond ((zerop x) (error 'undefined-operation :operator '*%2)) + ((plusp x) +negative-infinity+) + (t +positive-infinity+))) + + + (:method ((y (eql +negative-infinity+)) (x (eql +negative-infinity+)) + &optional r) + (declare (ignore r)) + +positive-infinity+) + + (:method ((y (eql +negative-infinity+)) (x (eql +positive-infinity+)) + &optional r) + (declare (ignore r)) + +negative-infinity+) + + (:method ((x number) (y (eql +positive-infinity+)) &optional r) + (declare (ignore r)) + (cond ((zerop x) (error 'undefined-operation :operator '*%2)) + ((plusp x) +positive-infinity+) + (t +negative-infinity+))) + + (:method ((y (eql +positive-infinity+)) (x number) &optional r) + (declare (ignore r)) + (cond ((zerop x) (error 'undefined-operation :operator '*%2)) + ((plusp x) +positive-infinity+) + (t +negative-infinity+))) + + (:method ((y (eql +positive-infinity+)) (x (eql +positive-infinity+)) + &optional r) + (declare (ignore r)) + +positive-infinity+) + + (:method ((y (eql +positive-infinity+)) (x (eql +negative-infinity+)) + &optional r) + (declare (ignore r)) + +negative-infinity+) + ) + + +(defgeneric -%2 (x y &optional r) + (:method ((x t) (y t) &optional r) + (declare (ignore r)) + (+%2 x (-%1 y)))) + + +(defgeneric /%2 (x y &optional r) + (:method ((x number) (y number) &optional r) + (declare (ignore r)) + (cl:/ x y)) + + (:method ((x t) (y t) &optional r) + (declare (ignore r)) + (*%2 x (/%1 y)))) + + +(defgeneric +%1 (x &optional r) + (:method ((x number) &optional r) + (declare (ignore r)) + (cl:+ x)) + + (:method ((x (eql +positive-infinity+)) &optional r) + (declare (ignore r)) + +positive-infinity+) + + (:method ((x (eql +negative-infinity+)) &optional r) + (declare (ignore r)) + +negative-infinity+) + + ) + +(defgeneric *%1 (x &optional r) + (:method ((x number) &optional r) + (declare (ignore r)) + (cl:* x)) + + (:method ((x (eql +positive-infinity+)) &optional r) + (declare (ignore r)) + +positive-infinity+) + + (:method ((x (eql +negative-infinity+)) &optional r) + (declare (ignore r)) + +negative-infinity+) + + ) + +(defgeneric -%1 (x &optional r) + (:method ((x number) &optional r) + (declare (ignore r)) + (cl:- x)) + + (:method ((x (eql +positive-infinity+)) &optional r) + (declare (ignore r)) + +negative-infinity+) + + (:method ((x (eql +negative-infinity+)) &optional r) + (declare (ignore r)) + +positive-infinity+) + + ) + +(defgeneric /%1 (x &optional r) + (:method ((x number) &optional r) + (declare (ignore r)) + (cl:/ x)) + + (:method ((x (eql +positive-infinity+)) &optional r) + (declare (ignore r)) + 0) + + (:method ((x (eql +negative-infinity+)) &optional r) + (declare (ignore r)) + 0) + ) + + +(defgeneric =%2 (x y) + (:method ((x number) (y number)) (cl:= x y)) + + (:method ((x number) (y (eql +negative-infinity+))) + nil) + + (:method ((y (eql +negative-infinity+)) (x number)) + nil) + + (:method ((y (eql +negative-infinity+)) (x (eql +negative-infinity+))) + (unless *ignore-comparison-errors-p* + (error 'undefined-operation :operator '=%2))) + + (:method ((y (eql +negative-infinity+)) (x (eql +positive-infinity+))) + nil) + + (:method ((y (eql +positive-infinity+)) (x (eql +negative-infinity+))) + nil) + + (:method ((x number) (y (eql +positive-infinity+))) + nil) + + (:method ((y (eql +positive-infinity+)) (x number)) + nil) + + (:method ((y (eql +positive-infinity+)) (x (eql +positive-infinity+))) + (unless *ignore-comparison-errors-p* + (error 'undefined-operation :operator '=%2))) + ) + + +(defgeneric =%1 (x) + (:method ((x t)) T) + ) + + +(defgeneric gcd%2 (x y) + (:method ((x integer) (y integer)) + (cl:gcd x y))) + + +;;;--------------------------------------------------------------------------- +;;; Redefined Common Lisp operators. + +(defun + (&rest args) + (if (null args) + (cl:+) + (let ((n-args (list-length args))) + (cond ((cl:= n-args 1) + (+%1 (first args))) + ((cl:= n-args 2) + (+%2 (first args) (second args))) + (t (+%2 (first args) (apply #'+ (rest args)))) + )))) + + +(defun - (arg1 &rest args) + (if (null args) + (-%1 arg1) + (let ((n-args (list-length args))) + (if (cl:= n-args 1) + (-%2 arg1 (first args)) + ;; The loop is done in order not to assume anything about + ;; the nature of summation. + (loop for arg in args + for minuend =(-%2 arg1 arg) then (-%2 minuend arg) + finally (return minuend)))) + )) + + +(defun * (&rest args) + (if (null args) + (cl:*) + (let ((n-args (list-length args))) + (cond ((cl:= n-args 1) + (*%1 (first args))) + ((cl:= n-args 2) + (*%2 (first args) (second args))) + (t (*%2 (first args) (apply #'* (rest args)))) + )))) + + +(defun / (arg1 &rest args) + (if (null args) + (/%1 arg1) + (let ((n-args (list-length args))) + (if (cl:= n-args 1) + (/%2 arg1 (first args)) + ;; The loop is done in order not to assume anything about + ;; the nature of divisions. + (loop for arg in args + for dividend = (/%2 arg1 arg) then (/%2 dividend arg) + finally (return dividend)))) + )) + + +(defun = (arg1 &rest args) + (if (null args) + (=%1 arg1) + (let ((n-args (list-length args))) + (if (cl:= n-args 1) + (=%2 arg1 (first args)) + ;; The loop is done in order not to assume anything about + ;; the nature of the equality test. + (loop for arg in args + always (=%2 arg1 arg)))) + )) + + +(defun < (arg1 &rest args) + (if (null args) + ;; (<%1 arg1) + t + (let ((n-args (list-length args))) + (if (cl:= n-args 1) + (<%2 arg1 (first args)) + ;; The loop is done in order not to assume anything about + ;; the nature of the equality test. + (loop for a1 = arg1 then a2 + for a2 in args + always (<%2 a1 a2)))) + )) + + +(defun > (arg1 &rest args) + (if (null args) + t + (let ((n-args (list-length args))) + (if (cl:= n-args 1) + (>%2 arg1 (first args)) + ;; The loop is done in order not to assume anything about + ;; the nature of the equality test. + (loop for a1 = arg1 then a2 + for a2 in args + always (>%2 a1 a2)))) + )) + + +(defun <= (arg1 &rest args) + (if (null args) + t + (let ((n-args (list-length args))) + (if (cl:= n-args 1) + (<=%2 arg1 (first args)) + ;; The loop is done in order not to assume anything about + ;; the nature of the equality test. + (loop for a1 = arg1 then a2 + for a2 in args + always (<=%2 a1 a2)))) + )) + + +(defun >= (arg1 &rest args) + (if (null args) + t + (let ((n-args (list-length args))) + (if (cl:= n-args 1) + (>=%2 arg1 (first args)) + ;; The loop is done in order not to assume anything about + ;; the nature of the equality test. + (loop for a1 = arg1 then a2 + for a2 in args + always (>=%2 a1 a2)))) + )) + + +(defun gcd (&rest args) + (if (null args) + (cl:gcd) + (let ((n-args (list-length args))) + (cond ((cl:= n-args 1) + (gcd%2 (first args) (first args))) + ((cl:= n-args 1) + (gcd%2 (first args) (second args))) + (t + (apply #'gcd + (gcd%2 (first args) (second args)) + (cddr args))) + )))) + + +(defgeneric expt (base power) + (:method ((base number) (power number)) + (cl:expt base power))) + + +(defgeneric zerop (n) + (:method ((n number)) + (cl:zerop n)) + + (:method ((x (eql +positive-infinity+))) + nil) + + (:method ((x (eql +negative-infinity+))) + nil) + ) + + + +;;;--------------------------------------------------------------------------- +;;; Other functions. + + +(defun increasing (x) + (declare (type sequence x)) + (loop for i from 0 below (1- (length x)) + for a1 = (elt x 0) then a2 + for a2 = (elt x (1+ i)) + always (<=%2 a1 a2))) + + +(defun decreasing (x) + (declare (type sequence x)) + (loop for i from 0 below (1- (length x)) + for a1 = (elt x 0) then a2 + for a2 = (elt x (1+ i)) + always (>=%2 a1 a2))) + + + +(defun strictly-increasing (x) + (declare (type sequence x)) + (loop for i from 0 below (1- (length x)) + for a1 = (elt x 0) then a2 + for a2 = (elt x (1+ i)) + always (<%2 a1 a2))) + + +(defun strictly-decreasing (x) + (declare (type sequence x)) + (loop for i from 0 below (1- (length x)) + for a1 = (elt x 0) then a2 + for a2 = (elt x (1+ i)) + always (>%2 a1 a2))) + +;;; end of file -- common-math.lisp -- Added: trunk/common-math/common-math.system ============================================================================== --- (empty file) +++ trunk/common-math/common-math.system Wed Aug 16 17:07:41 2006 @@ -0,0 +1,24 @@ +;;;; -*- Mode: Lisp -*- + +;;;; common-math.system -- + +(eval-when (:load-toplevel :execute) + (mk:add-registry-location (make-pathname :name nil + :type nil + :defaults *load-truename*)) + ) + + +(mk:defsystem "common-math" + :components ((:file "common-math-pkg") + (:file "common-math" + :depends-on ("common-math-pkg")) + (:system "numerics") + ;; (:system "polynomials") + ;; (:system "formulas") + (:file "common-math-user-pkg" + :depends-on ("numerics")) + ) + ) + +;;;; end of file -- common-math.system -- Added: trunk/common-math/numerics/linear-algebra/common-math-extra.lisp ============================================================================== --- (empty file) +++ trunk/common-math/numerics/linear-algebra/common-math-extra.lisp Wed Aug 16 17:07:41 2006 @@ -0,0 +1,98 @@ +;;;; -*- Mode: Lisp -*- + +;;;; common-math-extra.lisp -- + +(in-package "CL.MATH.LINEAR-ALGEBRA") + +;;;--------------------------------------------------------------------------- +;;; More functions. Essentially the top level element-wise array operations. + +(defgeneric .*%2 (x y &optional r) + (:method ((x number) (y number) &optional r) + (declare (ignore r)) + (*%2 x y)) + + (:method ((x (eql +negative-infinity+)) (y number) &optional r) + (declare (ignore r)) + (*%2 x y)) + + (:method ((x (eql +positive-infinity+)) (y number) &optional r) + (declare (ignore r)) + (*%2 x y)) + + + (:method ((y number) (x (eql +negative-infinity+)) &optional r) + (declare (ignore r)) + (*%2 x y)) + + + (:method ((y number) (x (eql +positive-infinity+)) &optional r) + (declare (ignore r)) + (*%2 x y)) + + + (:method ((y (eql +positive-infinity+)) (x (eql +positive-infinity+)) &optional r) + (declare (ignore r)) + +positive-infinity+) + + (:method ((y (eql +negative-infinity+)) (x (eql +negative-infinity+)) &optional r) + (declare (ignore r)) + +positive-infinity+) + + (:method ((y (eql +positive-infinity+)) (x (eql +negative-infinity+)) &optional r) + (declare (ignore r)) + +negative-infinity+) + + (:method ((x (eql +negative-infinity+)) (y (eql +positive-infinity+)) &optional r) + (declare (ignore r)) + +negative-infinity+) + ) + + +(defgeneric .*%1 (x &optional r) + (:method ((x number) &optional r) + (declare (ignore r)) + (*%1 x)) + + (:method ((x (eql +positive-infinity+)) &optional r) + (declare (ignore r)) + +positive-infinity+) + + (:method ((x (eql +negative-infinity+)) &optional r) + (declare (ignore r)) + +negative-infinity+) + ) + + +(defgeneric ./%2 (x y &optional r)) + +(defgeneric ./%1 (x &optional r)) + + +;;;--------------------------------------------------------------------------- +;;; The generic top-level functions. + +(defun .* (&rest args) + (if (null args) + (cl:*) + (let ((n-args (list-length args))) + (cond ((cl:= n-args 1) (.*%1 (first args))) + ((cl:= n-args 2) (.*%2 (first args) (second args))) + (t (.*%2 (first args) (apply #'.* (rest args)))) + )))) + + +(defun ./ (arg1 &rest args) + (if (null args) + (./%1 arg1) + (let ((n-args (list-length args))) + (if (cl:= n-args 1) + (./%2 arg1 (first args)) + ;; The loop is done in order not to assume anything about + ;; the nature of divisions. + (loop for arg in args + for dividend = (./%2 arg1 arg) then (./%2 dividend arg) + finally (return dividend)))) + )) + +;;;; end of file -- common-math-extra.lisp -- Added: trunk/common-math/numerics/linear-algebra/linear-algebra-pkg.lisp ============================================================================== --- (empty file) +++ trunk/common-math/numerics/linear-algebra/linear-algebra-pkg.lisp Wed Aug 16 17:07:41 2006 @@ -0,0 +1,19 @@ +;;; -*- Mode: Lisp -*- + +(defpackage "CL.MATH.LINEAR-ALGEBRA" (:use "CL.MATH" "COMMON-LISP") + (:nicknames "CL-MATH-LINALG" "LINALG") + + (:shadow cl:zerop cl:expt cl:gcd) + + (:shadowing-import-from "CL.MATH" "+" "-" "*" "/" "=" ">" "<" ">=" "<=") + + (:export "=" "+" "-" "*" "/" ">" "<" ">=" "<=") + (:export "ZEROP" "EXPT" "GCD") + + (:export ".*" "./" "EXPT*") + + (:export ".*%2" ".*%1" "./%2" "./%1") + + ) + +;;; end of file -- linear-algebra-pkg.lisp -- Added: trunk/common-math/numerics/linear-algebra/linear-algebra.system ============================================================================== --- (empty file) +++ trunk/common-math/numerics/linear-algebra/linear-algebra.system Wed Aug 16 17:07:41 2006 @@ -0,0 +1,13 @@ +;;;; -*- Mode: Lisp -*- + +;;;; linear-algebra.system -- + +(mk:defsystem "linear-algebra" + :components ("linear-algebra-pkg" + "vector-matrix-conditions" + "common-math-extra" + "vector" + "matrix" + "lu-decomposition")) + +;;;; end of file -- linear-algebra.system -- Added: trunk/common-math/numerics/linear-algebra/lu-decomposition.lisp ============================================================================== --- (empty file) +++ trunk/common-math/numerics/linear-algebra/lu-decomposition.lisp Wed Aug 16 17:07:41 2006 @@ -0,0 +1,326 @@ +;;;; -*- Mode: Lisp -*- + +;;;; lu-decomposition.lisp -- + +(in-package "CL.MATH.LINEAR-ALGEBRA") + +(define-condition singular-matrix-in-operation (arithmetic-error) + () + (:report (lambda (cnd stream) + (format stream "Singular matrix passed to operation ~S." + (arithmetic-error-operation cnd))))) + + +(defparameter *epsilon* 1.0e-20) ; Change this to CL constant. + + +(defgeneric square-matrix-p (m) + (:method ((m array)) + (and (matrix-array-p m) + (cl:= (array-dimension m 0) (array-dimension m 1)))) + (:method ((m matrix)) + (cl:= (matrix-row-number m) (matrix-column-number m)))) + + + +;;; Different base type. +;;; Probably a macrolet is in order. + +(defun lu-decompose (m &optional + (lu (make-array (array-dimensions m) + :element-type (array-element-type m) + :initial-element (coerce 0 (array-element-type m)))) + (permutations (make-array (array-dimension m 0) + :initial-element 0 + :element-type 'fixnum))) + (declare (type (array number (cl:* cl:*)) m lu) + (type (vector fixnum) permutations)) + + (assert (square-matrix-p m)) + (assert (square-matrix-p lu)) + (assert (cl:= (length permutations) (array-dimension m 0))) + + (let* ((n (array-dimension m 0)) + (row-scaling-vector (make-array n + :initial-element (coerce 0 (array-element-type m)) + :element-type (array-element-type m))) + (parity 1) + (imax 0) + ) + (declare (dynamic-extent row-scaling-vector) + (type fixnum imax n) + (type (member -1 1) parity) + (type (simple-array number (cl:*)) row-scaling-vector)) + + ;; Copy M into LU and then just operate the in-place algorithm on LU. + ;; If M is EQ to LU, that means that the argumen was supplied and + ;; that the user actually wants a destructive operation. Copying + ;; can be avoided in this case. + (unless (eq m lu) + (dotimes (i n) + (dotimes (j n) + (setf (aref lu i j) (aref m i j))))) + + + ;; Loop over rows to get the implicit scaling information. + ;; Meanwhile check that the matrix is non-singular. + (dotimes (i n) + (loop for k of-type fixnum from 0 below n + maximize (abs (aref lu i k)) into big ;; of-type (number) ; not CL. + finally + (when (cl:zerop big) + (error 'singular-matrix-in-operation + :operation 'lu-decompose + :operands (list m))) + (setf (aref row-scaling-vector i) (/ big)))) + + ;; Loop over the columns of Crout's method. + (dotimes (j n) + (block operate-on-column + ;; (format t "Column ~D IMAX = ~D~%" j imax) + (dotimes (i j) + (let ((sum (aref lu i j))) + (declare (type number sum)) + (dotimes (k i) + (decf sum (* (aref lu i k) (aref lu k j)))) + (setf (aref lu i j) sum))) + + ;; Initialize the search for the largest pivot. + (loop with big of-type number = 0.0 + for i from j below n + do (let ((sum (aref lu i j))) + (declare (type number sum)) + (dotimes (k j) + (decf sum (* (aref lu i k) (aref lu k j)))) + (setf (aref lu i j) sum) + (let ((temp (* (aref row-scaling-vector i) (abs sum)))) + (declare (type number temp)) + (when (>= temp big) + (setf big temp + imax i))) + )) + + ;; (format t "Now IMAX = ~D~%" imax) + (when (/= imax j) + ;; We need to interchange rows. + (dotimes (k n) + (rotatef (aref lu imax k) (aref lu j k))) + (setf parity (- parity)) + (setf (aref row-scaling-vector imax) + (aref row-scaling-vector j))) + (setf (aref permutations j) imax) + ;; (format t "Now J = ~D IMAX = ~D~2%" j imax) + (when (cl:zerop (aref lu j j)) + (setf (aref lu j j) *epsilon*)) + + (when (/= j (1- n)) + (loop with pivot of-type number = (/ (aref lu j j)) + for i from (1+ j) below n + do (setf (aref lu i j) (* pivot (aref lu i j))))) + + )) + (values lu permutations parity) + )) + + +(defun lu-decompose/double-float (m &optional + (lu (make-array (array-dimensions m) + :element-type 'double-float + :initial-element 0.0d0)) + (permutations (make-array (array-dimension m 0) + :initial-element 0 + :element-type 'fixnum))) + (declare (type (array double-float (cl:* cl:*)) m lu) + (type (vector fixnum) permutations)) + + (assert (square-matrix-p m)) + (assert (square-matrix-p lu)) + (assert (cl:= (length permutations) (array-dimension m 0))) + + (let* ((n (array-dimension m 0)) + (row-scaling-vector (make-array n + :initial-element (coerce 0 (array-element-type m)) + :element-type (array-element-type m))) + (parity 1.0d0) + (imax 0) + ) + (declare (dynamic-extent row-scaling-vector) + (type fixnum n imax) + (type (double-float -1.01d0 1.01d0) parity) + (type (simple-array double-float (cl:*)) row-scaling-vector)) + + ;; Copy M into LU and then just operate the in-place algorithm on LU. + ;; If M is EQ to LU, that means that the argumen was supplied and + ;; that the user actually wants a destructive operation. Copying + ;; can be avoided in this case. + (unless (eq m lu) + (dotimes (i n) + (dotimes (j n) + (setf (aref lu i j) (aref m i j))))) + + + ;; Loop over rows to get the implicit scaling information. + ;; Meanwhile check that the matrix is non-singular. + (dotimes (i n) + (loop for k of-type fixnum from 0 below n + maximize (abs (aref lu i k)) into big of-type double-float + finally + (when (cl:zerop big) + (error 'singular-matrix-in-operation + :operation 'lu-decompose + :operands (list m))) + (setf (aref row-scaling-vector i) (/ big)))) + + ;; Loop over the columns of Crout's method. + (dotimes (j n) + (block operate-on-column + ;; (format t "Column ~D IMAX = ~D~%" j imax) + (dotimes (i j) + (let ((sum (aref lu i j))) + (declare (type double-float sum)) + (dotimes (k i) + (decf sum (* (aref lu i k) (aref lu k j)))) + (setf (aref lu i j) sum))) + + ;; Initialize the search for the largest pivot. + (loop with big of-type double-float = 0.0d0 + for i of-type fixnum from j below n + do (let ((sum (aref lu i j))) + (declare (type double-float sum)) + (dotimes (k j) + (decf sum (* (aref lu i k) (aref lu k j)))) + (setf (aref lu i j) sum) + (let ((temp (* (aref row-scaling-vector i) (abs sum)))) + (declare (type double-float temp)) + (when (>= temp big) + (setf big temp + imax i))) + )) + + ;; (format t "Now IMAX = ~D~%" imax) + (when (/= imax j) + ;; We need to interchange rows. + (dotimes (k n) + (rotatef (aref lu imax k) (aref lu j k))) + (setf parity (- parity)) + (setf (aref row-scaling-vector imax) + (aref row-scaling-vector j))) + (setf (aref permutations j) imax) + ;; (format t "Now J = ~D IMAX = ~D~2%" j imax) + (when (cl:zerop (aref lu j j)) + (setf (aref lu j j) *epsilon*)) + + (when (/= j (1- n)) + (loop with pivot of-type double-float = (/ (aref lu j j)) + for i from (1+ j) below n + do (setf (aref lu i j) (* pivot (aref lu i j))))) + + )) + (values lu permutations parity) + )) + + +(defun split-lu-matrix (lu) + (let ((l (make-array (array-dimensions lu) :initial-element 0)) + (u (make-array (array-dimensions lu) :initial-element 0)) + (n (array-dimension lu 0)) + ) + (dotimes (i n) + (dotimes (j i) + (setf (aref l i j) (aref lu i j))) + (setf (aref l i i) 1)) + + (dotimes (i n) + (loop for j from i below n + do (setf (aref u i j) (aref lu i j)))) + + (values l u))) + + +;;;--------------------------------------------------------------------------- +;;; lu-back-substitution + +(defun lu-back-substitution (lu permutations b) + (let ((n (array-dimension lu 0))) + (loop with ii of-type fixnum = 0 + for i of-type fixnum from 0 below n + for ip of-type number = (aref permutations i) + for sum of-type number = (aref b ip) + do (setf (aref b ip) (aref b i)) + do (cond ((not (cl:zerop ii)) + (loop for j of-type fixnum from ii below (1- i) + do (decf sum (* (aref lu i j) (aref b i))))) + ((not (cl:zerop sum)) + (setf (aref b i) sum))) + ) + (loop for i of-type fixnum from (1- n) downto 0 + for sum of-type number = (aref b i) + do (loop for j of-type fixnum from (1+ i) below n + do (decf sum (* (aref lu i j) (aref b i)))) + do (setf (aref b i) (/ sum (aref lu i i)))) + ) + b) + + +;;;--------------------------------------------------------------------------- +;;; solve + +(defmethod solve ((a array) (b vector) &optional (result (copy-seq b))) + (multiple-value-bind (lu permutations) + (lu-decompose a) + (lu-back-substitution lu permutations result))) + +(defmethod solve ((a matrix) (b vector) &optional (result (copy-seq b))) + (solve (matrix-data a) b result)) + + +(defmethod solve/double-float ((a array) (b vector) &optional (result (copy-seq b))) + (multiple-value-bind (lu permutations) + (lu-decompose/double-float a) + (lu-back-substitution lu permutations result))) + +(defmethod solve/double-float ((a matrix) (b vector) &optional (result (copy-seq b))) + (solve/double-float (matrix-data a) b result)) + + +;;;--------------------------------------------------------------------------- +;;; det + +(defmethod det ((a array)) + (assert (square-matrix-p a)) + (multiple-value-bind (lu permutations parity) + (lu-decompose a) + (declare (ignore permutations)) + (dotimes (i (array-dimension lu 0) parity) + (setf parity (* parity (aref lu i i)))))) + +(defmethod det ((a matrix)) + (det (matrix-data a))) + + +(defmethod det/double-float ((a array)) + (assert (square-matrix-p a)) + (multiple-value-bind (lu permutations parity) + (lu-decompose a) + (declare (type (array double-float (cl:* cl:*)) lu) + (ignore permutations) + (type double-float parity)) + (dotimes (i (array-dimension lu 0) parity) + (setf parity (* parity (aref lu i i)))))) + +(defmethod det/double-float ((a matrix)) + (det/double-float (matrix-data a))) + + +;;;=========================================================================== +;;; Test + +(defvar A #2A((1 2 3) + (2 -1 1) + (3 4 -1))) + +(defvar LU-A #2A((1 2 3) + (2 -5 -5) + (3 0.4 -8))) + +;;; end of file -- lu-decomposition.lisp -- Added: trunk/common-math/numerics/linear-algebra/matrix-pkg.lisp ============================================================================== --- (empty file) +++ trunk/common-math/numerics/linear-algebra/matrix-pkg.lisp Wed Aug 16 17:07:41 2006 @@ -0,0 +1,19 @@ +;;; -*- Mode: Lisp -*- + +(defpackage "CL.MATH.POLYNOMIALS" (:use "CL.MATH" "COMMON-LISP") + (:nicknames "POLY") + + (:shadow cl:zerop cl:expt cl:gcd) + + (:shadowing-import-from "CL.MATH" "+" "-" "*" "/" "=" ">" "<" ">=" "<=") + + (:export "=" "+" "-" "*" "/" ">" "<" ">=" "<=") + (:export "ZEROP" "EXPT" "GCD") + + (:export "MAKE-POLYNOMIAL" + "POLYNOMIAL" + "UNIVARIATE-POLYNOMIAL" + ) + ) + +;;; end of file -- polymonials-pkg.lisp -- Added: trunk/common-math/numerics/linear-algebra/matrix.lisp ============================================================================== --- (empty file) +++ trunk/common-math/numerics/linear-algebra/matrix.lisp Wed Aug 16 17:07:41 2006 @@ -0,0 +1,950 @@ +;;;; -*- Mode: Lisp -*- + +;;;; matrix.lisp -- + +(in-package "CL.MATH.LINEAR-ALGEBRA") + +(deftype matrix-array (&optional (type 'number) (m 'cl:*) (n 'cl:*)) + `(array ,type (,m ,n))) + + +(defparameter *default-matrix-element-type* 'double-float) + +(defclass matrix () + ((data :reader matrix-data :initarg :data)) + (:default-initargs :data (make-array '(0 0) :element-type 'number))) + +(defgeneric matrix-p (x + #|&optional + (element-type *default-matrix-element-type*) + |# + ) + (:method ((x matrix)) t) + (:method ((x t)) nil) + (:method ((x vector)) t) ; Note that this is technically correct, but + ; it may cause problems. + (:method ((x array)) (cl:<= (array-rank x) 2)) ; Same here. + ) + + +(defvar *check-element-type* nil) + + +;;;--------------------------------------------------------------------------- +;;; Protocol. + +(defgeneric matrix-array-p (x &key check-element-type) + (:method ((x array) &key (check-element-type *check-element-type*)) + (and (cl:<= (array-rank x) 2) + (or (not check-element-type) + (nth-value 0 (subtypep (array-element-type x) 'number))))) + + (:method ((x t) &key check-element-type) + (declare (ignore check-element-type)) + nil)) + + + +(defgeneric matrix-element-type (m)) + +(defgeneric matrix-row-number (m) + (:method ((x matrix)) (array-dimension (matrix-data x) 0)) + (:method ((x vector)) 1) + (:method ((x array)) + (assert (matrix-array-p x)) + (array-dimension x 0))) + + +(defgeneric matrix-column-number (m) + (:method ((x matrix)) (array-dimension (matrix-data x) 1)) + (:method ((x vector)) (length x)) + (:method ((x array)) + (assert (matrix-array-p x)) + (array-dimension x 1))) + + +(defgeneric matrix-dimensions (m) + (:method ((x matrix)) (array-dimensions (matrix-data x))) + (:method ((x array)) + (assert (matrix-array-p x)) + (array-dimensions x))) + + +(defgeneric matrix-dimension (m axis) + (:method ((m matrix) axis) + (declare (type fixnum axis)) + (array-dimension (matrix-data m) axis)) + (:method ((m array) axis) + (declare (type fixnum axis)) + (array-dimension m axis))) + + +(defmethod matrix-dimension :before ((m array) axis) + (assert (matrix-array-p m))) + + + +(defmethod matrix-element-type ((m matrix)) (array-element-type (matrix-data m))) + +(defmethod matrix-element-type ((m array)) (array-element-type m)) + +(defmethod matrix-element-type :before ((m array)) + (assert (matrix-array-p m))) + + +(defgeneric shape-equal-p (m1 m2) + (:documentation "Returns non-NIL when M1 and M2 have the same `shape'.")) + + +(defmethod shape-equal-p ((m1 matrix) (m2 matrix)) + (with-slots ((m1d data)) + m1 + (with-slots ((m2d data)) + m2 + (and (cl:= (array-dimension m1d 0) + (array-dimension m2d 0)) + (cl:= (array-dimension m1d 1) + (array-dimension m2d 1)))))) + + +(defmethod shape-equal-p ((m1 array) (m2 matrix)) + (assert (matrix-array-p m1 :check-element-type nil)) + (let ((m1d m1)) + (with-slots ((m2d data)) + m2 + (and (cl:= (array-dimension m1d 0) + (array-dimension m2d 0)) + (cl:= (array-dimension m1d 1) + (array-dimension m2d 1)))))) + + +(defmethod shape-equal-p ((m2 matrix) (m1 array)) + (shape-equal-p m1 m2)) + + +(defmethod shape-equal-p ((m1 array) (m2 array)) + (assert (matrix-array-p m1 :check-element-type nil)) + (assert (matrix-array-p m2 :check-element-type nil)) + (let ((m1d m1) + (m2d m2) + ) + (and (cl:= (array-dimension m1d 0) + (array-dimension m2d 0)) + (cl:= (array-dimension m1d 1) + (array-dimension m2d 1))))) + + +(defmethod shape-equal-p ((m1 vector) (m2 vector)) + (cl:= (cl:length m1) (cl:length m2))) + + +(defmethod shape-equal-p ((m1 column) (m2 vector)) + (cl:= 1 (vector-length m1) (cl:length m2))) + + +(defmethod shape-equal-p ((m2 vector) (m1 column)) + (shape-equal-p m1 m2)) + + +;;;--------------------------------------------------------------------------- +;;; Creation. + +(defun make-matrix (n m + &rest array-keys + &key + (element-type *default-matrix-element-type*) + (initial-element (coerce 0 element-type) ie-supplied-p) + (initial-contents () ic-supplied-p) + &allow-other-keys) + (when (and ie-supplied-p ic-supplied-p) + (error "The MAKE-MATRIX :INITIAL-ELEMENT and :INITIAL-CONTENTS keywords may not be simultaneously supplied.")) + + (let ((data (cond ((and ic-supplied-p + (matrix-array-p initial-contents)) + initial-contents) + (ic-supplied-p + (apply #'make-array (list n m) + (append array-keys + (list :element-type element-type)))) + (t + (apply #'make-array (list n m) + (append array-keys + (list :element-type element-type + :initial-element initial-element))) + ))) + ) + (make-instance 'matrix :data data))) + + +(defun make-identity-matrix (n &rest array-keys &key &allow-other-keys) + (when (getf array-keys :initial-contents) + (warn "MAKE-IDENTITY-MATRIX passed an :INITIAL-CONTENTS argument.") + (remf array-keys :initial-contents)) + (when (getf array-keys :initial-element) + (warn "MAKE-IDENTITY-MATRIX passed an :INITIAL-ELEMENT argument.") + (remf array-keys :initial-element)) + (let* ((id (apply #'make-matrix n n array-keys)) + (data (matrix-data id)) + (one (coerce 1 (array-element-type data))) + ) + (dotimes (i n id) + (setf (aref data i i) one)))) + + +(defun eye (n &rest array-keys &key &allow-other-keys) + (apply #'make-identity-matrix n array-keys)) + + +(defun make-zero-matrix (n m + &rest array-keys + &key (element-type *default-matrix-element-type*) + &allow-other-keys) + (when (getf array-keys :initial-element) + (warn "MAKE-ZERO-MATRIX passed and :INITIAL-ELEMENT argument.") + (remf array-keys :initial-element)) + (apply #'make-matrix n m (append array-keys + (list :initial-element (coerce 0 element-type) + :element-type element-type)))) + + +(defun zeros (n m &rest array-keys &key &allow-other-keys) + (apply #'make-zero-matrix n m array-keys)) + + +;;;--------------------------------------------------------------------------- +;;; copy-matrix -- +;;; Uses the code from K Pitman and B. Margolin appeared on CLL a long +;;; long time ago. +;;; Note that at least on LW this code has some of the usual problems +;;; with DOUBLE-FLOATS. +#| +(defun copy-array (array) + (adjust-array (make-array (array-dimensions array) + :displaced-to array + :element-type (array-element-type array)) + (array-dimensions array) + :displaced-to nil)) +|# + + +(defmethod copy-matrix ((m matrix)) + (let* ((array (matrix-data m)) + (result-data + (adjust-array (make-array (array-dimensions array) + :displaced-to array + :element-type (array-element-type array)) + (array-dimensions array) + :displaced-to nil)) + ) + (make-instance 'matrix :data result-data))) + + +(defmethod copy-matrix ((m array)) + (check-type m matrix-array) + (adjust-array (make-array (array-dimensions m) + :displaced-to m + :element-type (array-element-type m)) + (array-dimensions m) + :displaced-to nil)) + + +;;;--------------------------------------------------------------------------- +;;; Equality. + +(defmethod =%2 ((x array) (y array)) + (and (matrix-array-p x) + (matrix-array-p y) + ;; (cl:= (matrix-row-number x) (matrix-row-number y)) + ;; (cl:= (matrix-column-number x) (matrix-column-number y)) + (cl:= (array-dimension x 0) (array-dimension y 0)) + (cl:= (array-dimension x 1) (array-dimension y 1)) + (loop for i from 0 below (array-total-size x) + always (=%2 (row-major-aref x i) (row-major-aref y i))))) + + +(defmethod =%2 ((x matrix) (y matrix)) + (or (eq x y) + (=%2 (matrix-data x) (matrix-data y)))) + + +(defmethod =%2 ((x matrix) (y array)) + (=%2 (matrix-data x) y)) + + +(defmethod =%2 ((y array) (x matrix)) + (=%2 (matrix-data x) y)) + + +;;;--------------------------------------------------------------------------- +;;; zerop + +(defmethod zerop ((x array)) + (and (matrix-array-p x) + (loop for i from 0 below (array-total-size x) + always (zerop (row-major-aref x i))))) + + +(defmethod zerop ((x matrix)) + (zerop (matrix-data x))) + + +;;;--------------------------------------------------------------------------- +;;; Addition. + +(defmethod +%2 :before ((x matrix) (y matrix) + &optional + (r nil r-supplied-p)) + (assert (shape-equal-p x y)) + (when r-supplied-p + (assert (matrix-p x)) + (assert (shape-equal-p x r)))) + + +(defmethod +%2 :before ((x matrix) (y array) + &optional + (r nil r-supplied-p)) + (assert (shape-equal-p x y)) + (when r-supplied-p + (assert (matrix-p x)) + (assert (shape-equal-p x r)))) + + +(defmethod +%2 :before ((y array) (x matrix) + &optional + (r nil r-supplied-p)) + (assert (shape-equal-p x y)) + (when r-supplied-p + (assert (matrix-p x)) + (assert (shape-equal-p x r)))) + + +(defmethod +%2 ((x number) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (with-slots (data) y + (with-slots ((result data)) r + (dotimes (i (array-total-size data) r) + (setf (row-major-aref result i) (+%2 x (row-major-aref data i)))) + ))) + + +(defmethod +%2 ((y matrix) (x number) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (with-slots (data) y + (with-slots ((result data)) r + (dotimes (i (array-total-size data) r) + (setf (row-major-aref result i) (+%2 x (row-major-aref data i)))) + ))) + + + +#+with-slots-slow +(defmethod +%2 ((x matrix) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (with-slots ((m1 data)) x + (with-slots ((m2 data)) y + (with-slots ((result data)) r + (dotimes (i (array-total-size m1)) + (setf (row-major-aref result i) + (cl:+ (row-major-aref m1 i) (row-major-aref m2 i)))) + ))) + r + ) + + +(defmethod +%2 ((x matrix) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (declare (ignore r-supplied-p)) + (let ((m1 (matrix-data x)) + (m2 (matrix-data y)) + (result (matrix-data r)) + ) + (dotimes (i (array-total-size m1) r) + (setf (row-major-aref result i) + (cl:+ (row-major-aref m1 i) (row-major-aref m2 i)))) + )) + + +#+with-slots-slow +(defmethod +%2 ((x array) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (let ((m1 x)) + (with-slots ((m2 data)) y + (with-slots ((result data)) r + (dotimes (i (array-total-size m1)) + (setf (row-major-aref result i) + (cl:+ (row-major-aref m1 i) (row-major-aref m2 i)))) + ))) + r) + + + +(defmethod +%2 ((x array) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (declare (ignore r-supplied-p)) + (let ((m1 x) + (m2 (matrix-data y)) + (result (matrix-data r)) + ) + (dotimes (i (array-total-size m1) r) + (setf (row-major-aref result i) + (cl:+ (row-major-aref m1 i) (row-major-aref m2 i)))) + )) + + +#+with-slots-slow +(defmethod +%2 ((y matrix) (x array) &optional (r (copy-matrix y) r-supplied-p)) + (let ((m1 x)) + (with-slots ((m2 data)) y + (with-slots ((result data)) r + (dotimes (i (array-total-size m1)) + (setf (row-major-aref result i) + (cl:+ (row-major-aref m1 i) (row-major-aref m2 i)))) + ))) + r) + + + +(defmethod +%2 ((y matrix) (x array) &optional (r (copy-matrix y) r-supplied-p)) + (declare (ignore r-supplied-p)) + (let ((m1 x) + (m2 (matrix-data y)) + (result (matrix-data r)) + ) + (dotimes (i (array-total-size m1) r) + (setf (row-major-aref result i) + (cl:+ (row-major-aref m1 i) (row-major-aref m2 i)))) + )) + + +;;;--------------------------------------------------------------------------- +;;; Subtraction. + + +(defmethod -%2 :before ((x matrix) (y matrix) + &optional + (r nil r-supplied-p)) + (assert (shape-equal-p x y)) + (when r-supplied-p + (assert (matrix-p x)) + (assert (shape-equal-p x r)))) + + +(defmethod -%2 :before ((x matrix) (y array) + &optional + (r nil r-supplied-p)) + (assert (shape-equal-p x y)) + (when r-supplied-p + (assert (matrix-p x)) + (assert (shape-equal-p x r)))) + + +(defmethod -%2 :before ((y array) (x matrix) + &optional + (r nil r-supplied-p)) + (assert (shape-equal-p x y)) + (when r-supplied-p + (assert (matrix-p x)) + (assert (shape-equal-p x r)))) + + +(defmethod -%2 ((x matrix) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (declare (ignore r-supplied-p)) + (with-slots ((m1 data)) x + (with-slots ((m2 data)) y + (with-slots ((result data)) r + (dotimes (i (array-total-size m1)) + (setf (row-major-aref result i) + (cl:- (row-major-aref m1 i) (row-major-aref m2 i)))) + ))) + r + ) + +(defmethod -%2 ((x array) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (declare (ignore r-supplied-p)) + (let ((m1 x)) + (with-slots ((m2 data)) y + (with-slots ((result data)) r + (dotimes (i (array-total-size m1)) + (setf (row-major-aref result i) + (cl:- (row-major-aref m1 i) (row-major-aref m2 i)))) + ))) + r) + +(defmethod -%2 ((y matrix) (x array) &optional (r (copy-matrix y) r-supplied-p)) + (declare (ignore r-supplied-p)) + (with-slots ((m1 data)) y + (let ((m2 x)) + (with-slots ((result data)) r + (dotimes (i (array-total-size m1)) + (setf (row-major-aref result i) + (cl:- (row-major-aref m1 i) (row-major-aref m2 i)))) + ))) + r) + + +;;;--------------------------------------------------------------------------- +;;; Multiplication. + +(defgeneric make-*-result-matrix (x y &optional element-type) + (:documentation + "Creates a matrix serve as result of a multiplication.") + + (:method ((x matrix) (y matrix) + &optional (element-type *default-matrix-element-type*)) + (make-matrix (matrix-row-number x) + (matrix-column-number y) + :element-type element-type)) + + (:method ((x array) (y matrix) + &optional (element-type *default-matrix-element-type*)) + (make-matrix (matrix-row-number x) + (matrix-column-number y) + :element-type element-type)) + + (:method ((x matrix) (y array) + &optional (element-type *default-matrix-element-type*)) + (make-matrix (matrix-row-number x) + (matrix-column-number y) + :element-type element-type)) + + (:method ((x vector) (y matrix) + &optional (element-type *default-matrix-element-type*)) + (make-matrix 1 + (matrix-column-number y) + :element-type element-type)) + + (:method ((x matrix) (y vector) + &optional (element-type *default-matrix-element-type*)) + (make-matrix (matrix-row-number x) + 1 + :element-type element-type))) + + + +(defgeneric conforming-*-dimensions-p (x y r) + (:documentation + "Checks whether the dimensions of R are correct to hold XxY.") + + (:method ((x t) (y t) (r t)) + (error "Result object is not a matrix: ~S." r)) + + ;; 1. Matrix results. + + ;; 1.1. Number methods. + + (:method ((x number) (y matrix) (r matrix)) + (shape-equal-p y r)) + + (:method ((x number) (y array) (r matrix)) + (shape-equal-p y r)) + + (:method ((y matrix) (x number) (r matrix)) + (conforming-*-dimensions-p x y r)) + + (:method ((y array) (x number) (r matrix)) + (conforming-*-dimensions-p x y r)) + + ;; 1.2. Matrix and Array arguments. + + (:method ((x array) (y array) (r matrix)) + (and (matrix-array-p x) + (matrix-array-p y) + (cl:= (matrix-row-number x) (matrix-column-number y)) + (cl:= (matrix-row-number x) (matrix-row-number r)) + (cl:= (matrix-column-number y) (matrix-column-number r)))) + + (:method ((x vector) (y array) (r matrix)) + (and (matrix-array-p y) + (cl:= (matrix-row-number x) (matrix-column-number y)) + (cl:= (matrix-row-number x) (matrix-row-number r)) + (cl:= (matrix-column-number y) (matrix-column-number r)))) + + + (:method ((x array) (y vector) (r matrix)) + (and (matrix-array-p x) + (cl:= (matrix-row-number x) (matrix-column-number y)) + (cl:= (matrix-row-number x) (matrix-row-number r)) + (cl:= (matrix-column-number y) (matrix-column-number r)))) + + (:method ((x matrix) (y matrix) (r matrix)) + (conforming-*-dimensions-p (matrix-data x) (matrix-data y) r)) + + (:method ((x array) (y matrix) (r matrix)) + (conforming-*-dimensions-p x (matrix-data y) r)) + + (:method ((x matrix) (y array) (r matrix)) + (conforming-*-dimensions-p (matrix-data x) y r)) + + + ;; 2. Array results. + + ;; 2.1. Number methods. + + (:method ((x number) (y matrix) (r array)) + (shape-equal-p y r)) + + (:method ((x number) (y array) (r array)) + (shape-equal-p y r)) + + (:method ((y matrix) (x number) (r array)) + (conforming-*-dimensions-p x y r)) + + (:method ((y array) (x number) (r array)) + (conforming-*-dimensions-p x y r)) + + ;; 2.2. Matrix and Array arguments. + + (:method ((x array) (y array) (r array)) + (and (matrix-array-p x) + (matrix-array-p y) + (matrix-array-p r) + (cl:= (matrix-row-number x) (matrix-column-number y)) + (cl:= (matrix-row-number x) (matrix-row-number r)) + (cl:= (matrix-column-number y) (matrix-column-number r)))) + + (:method ((x vector) (y array) (r array)) + (and (matrix-array-p y) + (cl:= (matrix-row-number x) (matrix-column-number y)) + (cl:= (matrix-row-number x) (matrix-row-number r)) + (cl:= (matrix-column-number y) (matrix-column-number r)))) + + (:method ((x array) (y vector) (r array)) + (and (matrix-array-p x) + (cl:= (matrix-row-number x) (matrix-column-number y)) + (cl:= (matrix-row-number x) (matrix-row-number r)) + (cl:= (matrix-column-number y) (matrix-column-number r)))) + + (:method ((x matrix) (y matrix) (r array)) + (conforming-*-dimensions-p (matrix-data x) (matrix-data y) r)) + + (:method ((x array) (y matrix) (r array)) + (conforming-*-dimensions-p x (matrix-data y) r)) + + (:method ((x matrix) (y array) (r array)) + (conforming-*-dimensions-p (matrix-data x) y r)) + + ;; 3. Other methods. + + (:method ((x vector) (y vector) (r t)) + ;; Maybe a warning here is needed. + (cl:= (length x) (length y))) + ) + + +;;; *%2 Methods. + +(defmethod *%2 ((x number) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (conforming-*-dimensions-p x y r))) + (with-slots ((m data)) y + (with-slots ((result data)) r + (dotimes (i (array-total-size m) r) + (setf (row-major-aref result i) (*%2 x (row-major-aref m i)))) + ))) + + +(defmethod *%2 ((y matrix) (x number) &optional (r (copy-matrix y) r-supplied-p)) + (declare (ignore r-supplied-p)) + (*%2 x y r)) + + +(defmethod *%2 ((x matrix) (y matrix) + &optional (r (make-*-result-matrix x y) r-supplied-p)) + (when r-supplied-p (assert (conforming-*-dimensions-p x y r))) + ;; X is a NxK matrix + ;; Y is a KxM matrix + ;; R is a NxM matrix + (let* ((m1 (matrix-data x)) + (m2 (matrix-data y)) + (result (matrix-data r)) + (n (array-dimension m1 0)) + (k (array-dimension m1 1)) + (m (array-dimension m2 0)) + (z (coerce 0 (array-element-type result))) + ) + (declare (type fixnum n k m)) + (dotimes (i n r) + (dotimes (j m) + (let ((temp z)) + (dotimes (l k) + (setf (aref result i j) + (setf temp (+%2 temp (*%2 (aref m1 i l) (aref m2 l j))))) + )) + )))) + + +(defmethod *%2 ((x array) (y matrix) + &optional (r (make-*-result-matrix x y) r-supplied-p)) + (when r-supplied-p (assert (conforming-*-dimensions-p x y r))) + ;; X is a NxK matrix + ;; Y is a KxM matrix + ;; R is a NxM matrix + (let* ((m1 x) + (m2 (matrix-data y)) + (result (matrix-data r)) + (n (array-dimension m1 0)) + (k (array-dimension m1 1)) + (m (array-dimension m2 0)) + (z (coerce 0 (array-element-type result))) + ) + (declare (type fixnum n k m)) + (dotimes (i n r) + (dotimes (j m) + (let ((temp z)) + (dotimes (l k) + (setf (aref result i j) + (incf temp (* (aref m1 i l) (aref m2 l j)))) + )) + )))) + + +(defmethod *%2 ((x matrix) (y array) + &optional (r (make-*-result-matrix x y) r-supplied-p)) + (when r-supplied-p (assert (conforming-*-dimensions-p x y r))) + ;; X is a NxK matrix + ;; Y is a KxM matrix + ;; R is a NxM matrix + (let* ((m1 (matrix-data x)) + (m2 y) + (result (matrix-data r)) + (n (array-dimension m1 0)) + (k (array-dimension m1 1)) + (m (array-dimension m2 1)) + (z (coerce 0 (array-element-type result))) + ) + (declare (type fixnum n k m)) + (dotimes (i n r) + (dotimes (j m) + (let ((temp z)) + (dotimes (l k) + (setf (aref result i j) + (incf temp (* (aref m1 i l) (aref m2 l j)))) + )) + )))) + + +(defmethod *%2 ((x vector) (y matrix) + &optional (r (make-*-result-matrix x y) r-supplied-p)) + (when r-supplied-p (assert (conforming-*-dimensions-p x y r))) + (let* ((m1 x) + (m2 (matrix-data y)) + (result (matrix-data r)) + (n 1) + (k (length m1)) + (m (array-dimension m2 1)) + (z (coerce 0 (array-element-type result))) + ) + (declare (type fixnum k m) (type (integer 1 1) n)) + (dotimes (i n r) + (dotimes (j m) + (let ((temp z)) + (dotimes (l k) + (setf (aref result i j) + (incf temp (* (aref m1 l) (aref m2 l j)))) + )) + )))) + + +(defmethod *%2 ((x matrix) (y vector) + &optional (r (make-*-result-matrix x y) r-supplied-p)) + (when r-supplied-p (assert (conforming-*-dimensions-p x y r))) + (let* ((m1 y ) + (m2 (matrix-data x)) + (result (matrix-data r)) + (n (array-dimension m1 0)) + (k (length y)) + (m k) + (z (coerce 0 (array-element-type result))) + ) + (declare (type fixnum n k m)) + (dotimes (i n r) + (dotimes (j m) + (let ((temp z)) + (dotimes (l k) + (setf (aref result i j) + (incf temp (* (aref m1 l) (aref m2 l j)))) + )) + )))) + + +(defmethod *%2 ((x number) (y array) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (conforming-*-dimensions-p x y r))) + (let ((m y) + (result r) + ) + (dotimes (i (array-total-size m) r) + (setf (row-major-aref result i) (*%2 x (row-major-aref m i)))) + )) + + +(defmethod *%2 ((y array) (x number) &optional (r (copy-matrix y) r-supplied-p)) + (*%2 x y r)) + + +;;; The next one breaks the return type convention. +#| Defined in 'vector.lisp'. +(defmethod *%2 ((x vector) (y vector) &optional r) + (declare (ignore r)) + (assert (conforming-*-dimensions-p x y nil)) + (let ((result 0)) + (dotimes (i (length x) result) + (setf result (+%2 result (* (aref x i) (aref y i))))))) +|# + + +;;;--------------------------------------------------------------------------- +;;; Division. +;;; Only the simple form of division is implemented here. +;;; The general form requires the inverse, which requires extra machinery + +(defmethod /%2 ((x number) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (declare (ignore r-supplied-p)) + (*%2 (/ x) y r)) + + +(defmethod /%2 ((y matrix) (x number) &optional (r (copy-matrix y) r-supplied-p)) + (declare (ignore r-supplied-p)) + (*%2 (/ x) y r)) + + +;;;--------------------------------------------------------------------------- +;;; Exponentiation. +;;; ... missing ... + + +;;;--------------------------------------------------------------------------- +;;; Element-wise operations. + +(defmethod .*%2 ((x number) (y matrix) + &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (with-slots (data) y + (with-slots ((result data)) r + (dotimes (i (array-total-size data) r) + (setf (row-major-aref result i) (*%2 x (row-major-aref data i)))) + ))) + + +(defmethod .*%2 ((y matrix) (x number) &optional (r (copy-matrix y))) + (.*%2 x y r)) + + +(defmethod .*%2 ((x number) (y array) + &optional (r (copy-matrix y) r-supplied-p)) + (assert (matrix-array-p y)) + (when r-supplied-p (assert (shape-equal-p y r))) + (let* ((data y) + (result r) + ) + (dotimes (i (array-total-size data) r) + (setf (row-major-aref result i) (*%2 x (row-major-aref data i)))) + )) + + +(defmethod .*%2 ((y array) (x number) &optional (r (copy-matrix y))) + (.*%2 x y r)) + + + +(defmethod .*%2 ((x array) (y array) &optional (r (copy-matrix y) r-supplied-p)) + (when r-supplied-p (assert (shape-equal-p y r))) + (assert (shape-equal-p x y)) + + (dotimes (i (array-total-size y) r) + (setf (row-major-aref r i) + (*%2 (row-major-aref x i) (row-major-aref y i))))) + + +;;;--------------------------------------------------------------------------- +;;; Transpose + +(defgeneric conforming-transpose-dimensions-p (x r) + (:documentation + "Checks whether the dimensions of R are correct to hold X'.") + + (:method ((x t) (r t)) + (error "Result object is not a matrix: ~S." r)) + + ;; 1. Matrix results. + + (:method ((x array) (r matrix)) + (and (matrix-array-p x) + (cl:= (matrix-row-number x) (matrix-column-number r)) + (cl:= (matrix-column-number x) (matrix-row-number r)))) + + (:method ((x matrix) (r matrix)) + (and (cl:= (matrix-row-number x) (matrix-column-number r)) + (cl:= (matrix-column-number x) (matrix-row-number r)))) + + + ;; 2. Array results. + + (:method ((x array) (r array)) + (and (matrix-array-p x) + (matrix-array-p r) + (cl:= (matrix-row-number x) (matrix-column-number r)) + (cl:= (matrix-column-number x) (matrix-row-number r)))) + + (:method ((x matrix) (r array)) + (and (matrix-array-p r) + (cl:= (matrix-row-number x) (matrix-column-number r)) + (cl:= (matrix-column-number x) (matrix-row-number r)))) + ) + + + +(defmethod transpose ((m matrix) + &optional + (r (make-matrix (matrix-column-number m) + (matrix-row-number m) + :element-type + (matrix-element-type m)) + r-supplied-p)) + (when r-supplied-p (assert (conforming-transpose-dimensions-p m r))) + (let ((rows (matrix-row-number m)) + (cols (matrix-column-number m)) + (m (matrix-data m)) + (result (matrix-data r)) + ) + (dotimes (i rows r) + (dotimes (j cols) + (setf (aref result j i) (aref m i j)))))) + + +(defmethod transpose ((m array) + &optional + (r (make-array (array-dimensions m) + :element-type + (array-element-type m) + ) + r-supplied-p)) + (when r-supplied-p (assert (conforming-transpose-dimensions-p m r))) + (let ((rows (matrix-row-number m)) + (cols (matrix-column-number m)) + ) + (dotimes (i rows r) + (dotimes (j cols) + (setf (aref r j i) (aref m i j)))))) + + +(defmethod transpose :before ((m array) &optional r) + (declare (ignore r)) + (assert (matrix-array-p m))) + + +;;;--------------------------------------------------------------------------- +;;; Utilities. + +#+simple +(defmethod print-object ((m matrix) stream) + (print-unreadable-object (m stream :identity t :type t) + (format stream "~S" (matrix-data m)))) + + +(defmethod print-object ((m matrix) stream) + (let ((d (matrix-data m))) + (format stream "#M(~S ~S ~S)" + (array-element-type d) + (array-dimensions d) + d))) + +;;; Missing: +;;; Define #M read macro. + + +;;;; end of file -- matrix.lisp -- Added: trunk/common-math/numerics/linear-algebra/vector-matrix-conditions.lisp ============================================================================== --- (empty file) +++ trunk/common-math/numerics/linear-algebra/vector-matrix-conditions.lisp Wed Aug 16 17:07:41 2006 @@ -0,0 +1,27 @@ +;;;; -*- Mode: Lisp -*- + +;;;; vector-matrix-conditions.lisp -- + +(in-package "CL.MATH.LINEAR-ALGEBRA") + +;;;--------------------------------------------------------------------------- +;;; Conditions relative to the vectors and matrix subsystem.. + +(define-condition incompatible-shapes (error) + ((a :accessor incompatible-shape-a + :initarg :a) + (b :accessor incompatible-shape-b + :initarg :b) + ) + (:documentation "The Incompatible Shapes Error. +This error is signalled by functions that get arguments -- mainly +matrices and/or vectors -- whose 'shape' is not compatible with +respect to the semantics of the operation.") + + (:report (lambda (isc stream) + (declare (ignore isc)) + (format stream "The shapes passed have incompatible dimensions."))) + ) + + +;;;; end of file -- vector-matrix-conditions.lisp -- Added: trunk/common-math/numerics/linear-algebra/vector-matrix.lisp ============================================================================== --- (empty file) +++ trunk/common-math/numerics/linear-algebra/vector-matrix.lisp Wed Aug 16 17:07:41 2006 @@ -0,0 +1,613 @@ +;;;; -*- Mode: Lisp -*- + +;;;; vector-matrix.lisp -- +;;;; Protocols and functions common to and dependent on vector and matrix +;;;; definitions. + +(in-package "CL.MATH.LINEAR-ALGEBRA") + +;;;;=========================================================================== +;;;; Protocol. + +;;; join and stack +;;; Note: no 'result' machinery in place yet. + +(defgeneric join%2 (d1 d2)) + +(defgeneric stack%2 (d1 d2)) + + +;;; slice + +(defgeneric slice (e shape-designator)) + + +;;; range + +(defstruct (range (:constructor %range (start &optional end step))) + (start 0 :type (or fixnum (member * cl:*))) + (end nil :type (or null fixnum (member * cl:*))) + (step 1 :type fixnum)) + + +(defstruct (decreasing-range (:include range))) + +(defstruct (increasing-range (:include range))) + + +(defun range (start &optional end (step 1)) + (if end + (if (<= start end) + (make-increasing-range :start start :end end :step step) + (make-decreasing-range :start start :end end :step step)) + (if (plusp step) + (make-increasing-range :start start :end end :step step) + (make-decreasing-range :start start :end end :step step)) + )) + + + +;;; ref + +(defgeneric ref (e index &rest indexes)) + +(defgeneric ref%2 (e i j)) + + +;;;--------------------------------------------------------------------------- +;;; Implementation. + +;;; join%2 -- + +(defmethod join%2 ((d1 vector) (d2 vector)) + (concatenate 'vector d1 d2)) + + +(defmethod join%2 ((d1 string) (d2 string)) + (concatenate 'string d1 d2)) + + +(defmethod join%2 ((d1 list) (d2 list)) + (concatenate 'list d1 d2)) + + +(defmethod join%2 ((d1 vector) (d2 sequence)) + (concatenate 'vector d1 d2)) + + +(defmethod join%2 ((d1 sequence) (d2 vector)) + (concatenate (type-of d1) d1 d2)) + + +(defmethod join%2 ((d1 column) (d2 column)) + (let ((initial-contents + (map 'list #'vector (column-vector d1) (column-vector d2)))) + (declare (dynamic-extent initial-contents)) + (make-array (list (length (column-vector d1)) 2) + :initial-contents initial-contents))) + + +(defmethod join%2 ((d1 vector) (d2 column)) + (unless (cl:= (length (column-vector d2)) 1) + (error "Trying to JOIN a vector (1x~D) to a column (~Dx1)." + (length d1) + (length (column-vector d2)))) + (concatenate 'vector d1 (column-vector d2))) + + +(defmethod join%2 ((d1 column) (d2 vector)) + (unless (cl:= (length (column-vector d1)) 1) + (error "Trying to JOIN a column (~Dx1) to a vector (1x~D)." + (length (column-vector d1)) + (length d2))) + (concatenate 'vector (column-vector d1) d2)) + + +(defmethod join%2 ((d1 column) (d2 array)) + (let ((nr (array-dimension d2 0))) + (assert (cl:= (length (column-vector d1)) nr)) + (let* ((nc (1+ (array-dimension d2 1))) + (result (make-array (list nr nc) + :initial-element (aref d2 0 0) + :element-type (array-element-type d2))) + (col (column-vector d1)) + ) + (dotimes (i nr result) + (setf (aref result i 0) (aref col i)) + (loop for j from 1 below nc + do (setf (aref result i j) (aref d2 i (1- j))))) + ))) + + +(defmethod join%2 ((d1 array) (d2 column)) + (let ((nr (array-dimension d1 0))) + (assert (cl:= (length (column-vector d2)) nr)) + (let* ((nc (1+ (array-dimension d1 1))) + (nc-1 (1- nc)) + (result (make-array (list nr nc) + :initial-element (aref d1 0 0) + :element-type (array-element-type d1))) + (col (column-vector d2)) + ) + + (dotimes (i nr result) + (setf (aref result i nc-1) (aref col i)) + (dotimes (j nc-1) + (setf (aref result i j) (aref d1 i j)))) + ))) + + +(defmethod join%2 ((d1 array) (d2 array)) + (let ((nr1 (array-dimension d1 0)) + (nr2 (array-dimension d2 0)) + ) + (assert (cl:= nr1 nr2)) + (let* ((nc1 (array-dimension d1 1)) + (nc2 (array-dimension d2 1)) + (result (make-array (list nr1 (+ nc1 nc2)) + :initial-element (aref d1 0 0) + :element-type (array-element-type d1))) + ) + + (dotimes (i nr1) + (dotimes (j nc1) + (setf (aref result i j) (aref d1 i j)))) + + (dotimes (i nr2 result) + (dotimes (j nc2) + (setf (aref result i (+ j nc1)) (aref d2 i j)))) + ))) + + +(defmethod join%2 ((d1 matrix) (d2 matrix)) + (let ((nr1 (matrix-row-number d1)) + (nr2 (matrix-row-number d2)) + ) + (assert (cl:= nr1 nr2)) + (let* ((nc1 (matrix-column-number d1)) + (nc2 (matrix-column-number d2)) + (data1 (matrix-data d1)) + (data2 (matrix-data d2)) + (result (make-array (list nr1 (+ nc1 nc2)) + :initial-element (aref data1 0 0) + :element-type (array-element-type data1))) + ) + + (dotimes (i nr1) + (dotimes (j nc1) + (setf (aref result i j) (aref data1 i j)))) + + (dotimes (i nr2 (make-instance 'matrix :data result)) + (dotimes (j nc2) + (setf (aref result i (+ j nc1)) (aref data2 i j)))) + ))) + + +(defmethod join%2 ((d1 array) (d2 matrix)) + (let ((nr1 (matrix-row-number d1)) + (nr2 (matrix-row-number d2)) + ) + (assert (cl:= nr1 nr2)) + (let* ((nc1 (matrix-column-number d1)) + (nc2 (matrix-column-number d2)) + (data1 d1) + (data2 (matrix-data d2)) + (result (make-array (list nr1 (+ nc1 nc2)) + :initial-element (aref data1 0 0) + :element-type (array-element-type data1))) + ) + + (dotimes (i nr1) + (dotimes (j nc1) + (setf (aref result i j) (aref data1 i j)))) + + (dotimes (i nr2 (make-instance 'matrix :data result)) + (dotimes (j nc2) + (setf (aref result i (+ j nc1)) (aref data2 i j)))) + ))) + + +(defmethod join%2 ((d1 matrix) (d2 array)) + (let ((nr1 (matrix-row-number d1)) + (nr2 (matrix-row-number d2)) + ) + (assert (cl:= nr1 nr2)) + (let* ((nc1 (matrix-column-number d1)) + (nc2 (matrix-column-number d2)) + (data1 (matrix-data d1)) + (data2 d2) + (result (make-array (list nr1 (+ nc1 nc2)) + :initial-element (aref data1 0 0) + :element-type (array-element-type data1))) + ) + + (dotimes (i nr1) + (dotimes (j nc1) + (setf (aref result i j) (aref data1 i j)))) + + (dotimes (i nr2 (make-instance 'matrix :data result)) + (dotimes (j nc2) + (setf (aref result i (+ j nc1)) (aref data2 i j)))) + ))) + + +(defmethod join%2 ((d1 vector) (d2 matrix)) + (let ((result (join%2 d1 (matrix-data d2)))) + (make-instance 'matrix :data result))) + + +(defmethod join%2 ((d2 matrix) (d1 vector)) + (let ((result (join%2 (matrix-data d2) d1))) + (make-instance 'matrix :data result))) + + +(defmethod join%2 ((d1 column) (d2 matrix)) + (let ((result (join%2 d1 (matrix-data d2)))) + (make-instance 'matrix :data result))) + + +(defmethod join%2 ((d2 matrix) (d1 column)) + (let ((result (join%2 (matrix-data d2) d1))) + (make-instance 'matrix :data result))) + + +;;; stack%2 -- + +(defmethod stack%2 ((d1 column) (d2 column)) + (column (concatenate 'vector (column-vector d1) (column-vector d2)))) + + +(defmethod stack%2 ((d1 vector) (d2 array)) + (let ((nc (array-dimension d2 1))) + (assert (cl:= (length d1) nc)) + (let* ((nr (1+ (array-dimension d2 0))) + (result (make-array (list nr nc) + :initial-element (aref d2 0 0) + :element-type (array-element-type d2))) + ) + (dotimes (i nc) + (setf (aref result 0 i) (aref d1 i))) + (loop for i from 1 below nr + do (dotimes (j nc) + (setf (aref result i j) (aref d2 (1- i) j)))) + result))) + + +(defmethod stack%2 ((d1 sequence) (d2 array)) + (let ((nc (array-dimension d2 1))) + (assert (cl:= (length d1) nc)) + (let* ((nr (1+ (array-dimension d2 0))) + (result (make-array (list nr nc) + :initial-element (aref d2 0 0) + :element-type (array-element-type d2))) + ) + (dotimes (i nc) + (setf (aref result 0 i) (elt d1 i))) + (loop for i from 1 below nr + do (dotimes (j nc) + (setf (aref result i j) (aref d2 (1- i) j)))) + result))) + + +(defmethod stack%2 ((d1 array) (d2 vector)) + (let ((nc (array-dimension d1 1))) + (assert (cl:= (length d2) nc)) + (let* ((nr (1+ (array-dimension d1 0))) + (nr-1 (1- nr)) + (result (make-array (list nr nc) + :initial-element (aref d1 0 0) + :element-type (array-element-type d1))) + ) + + (dotimes (i nr-1) + (dotimes (j nc) + (setf (aref result i j) (aref d1 i j)))) + (dotimes (i nc) + (setf (aref result nr-1 i) (aref d2 i))) + result))) + + +(defmethod stack%2 ((d1 array) (d2 sequence)) + (let ((nc (array-dimension d1 1))) + (assert (cl:= (length d2) nc)) + (let* ((nr (1+ (array-dimension d1 0))) + (nr-1 (1- nr)) + (result (make-array (list nr nc) + :initial-element (aref d1 0 0) + :element-type (array-element-type d1))) + ) + + (dotimes (i nr-1) + (dotimes (j nc) + (setf (aref result i j) (aref d1 i j)))) + (dotimes (i nc) + (setf (aref result nr-1 i) (elt d2 i))) + result))) + + +(defmethod stack%2 ((d1 array) (d2 array)) + (let ((nc1 (array-dimension d1 1)) + (nc2 (array-dimension d2 1)) + ) + (assert (cl:= nc1 nc2)) + (let* ((nr1 (array-dimension d1 0)) + (nr2 (array-dimension d2 0)) + (result (make-array (list (+ nr1 nr2) nc1) + :initial-element (aref d1 0 0) + :element-type (array-element-type d1))) + ) + + (dotimes (i nr1) + (dotimes (j nc1) + (setf (aref result i j) (aref d1 i j)))) + + (dotimes (i nr2 result) + (dotimes (j nc2) + (setf (aref result (+ i nr1) j) (aref d2 i j)))) + ))) + + +(defmethod stack%2 ((d1 matrix) (d2 matrix)) + (let ((nc1 (matrix-column-number d1)) + (nc2 (matrix-column-number d2)) + ) + (assert (cl:= nc1 nc2)) + (let* ((nr1 (matrix-row-number d1)) + (nr2 (matrix-row-number d2)) + (data1 (matrix-data d1)) + (data2 (matrix-data d2)) + (result (make-array (list (+ nr1 nr2) nc1) + :initial-element (aref data1 0 0) + :element-type (array-element-type data1))) + ) + + (dotimes (i nr1) + (dotimes (j nc1) + (setf (aref result i j) (aref data1 i j)))) + + (dotimes (i nr2 (make-instance 'matrix :data result)) + (dotimes (j nc2) + (setf (aref result (+ i nr1) j) (aref data2 i j)))) + ))) + + +(defmethod stack%2 ((d1 array) (d2 matrix)) + (let ((nc1 (matrix-column-number d1)) + (nc2 (matrix-column-number d2)) + ) + (assert (cl:= nc1 nc2)) + (let* ((nr1 (matrix-row-number d1)) + (nr2 (matrix-row-number d2)) + (data1 d1) + (data2 (matrix-data d2)) + (result (make-array (list (+ nr1 nr2) nc1) + :initial-element (aref data1 0 0) + :element-type (array-element-type data1))) + ) + + (dotimes (i nr1) + (dotimes (j nc1) + (setf (aref result i j) (aref data1 i j)))) + + (dotimes (i nr2 (make-instance 'matrix :data result)) + (dotimes (j nc2) + (setf (aref result (+ i nr1) j) (aref data2 i j)))) + ))) + + + +(defmethod stack%2 ((d1 matrix) (d2 array)) + (let ((nc1 (matrix-column-number d1)) + (nc2 (matrix-column-number d2)) + ) + (assert (cl:= nc1 nc2)) + (let* ((nr1 (matrix-row-number d1)) + (nr2 (matrix-row-number d2)) + (data1 (matrix-data d1)) + (data2 d2) + (result (make-array (list (+ nr1 nr2) nc1) + :initial-element (aref data1 0 0) + :element-type (array-element-type data1))) + ) + + (dotimes (i nr1) + (dotimes (j nc1) + (setf (aref result i j) (aref data1 i j)))) + + (dotimes (i nr2 (make-instance 'matrix :data result)) + (dotimes (j nc2) + (setf (aref result (+ i nr1) j) (aref data2 i j)))) + ))) + + +(defmethod stack%2 ((d1 vector) (d2 matrix)) + (let ((result (stack%2 d1 (matrix-data d2)))) + (make-instance 'matrix :data result))) + + +(defmethod stack%2 ((d2 matrix) (d1 vector)) + (let ((result (stack%2 (matrix-data d2) d1))) + (make-instance 'matrix :data result))) + + +(defmethod stack%2 ((d1 column) (d2 matrix)) + (let ((result (stack%2 d1 (matrix-data d2)))) + (make-instance 'matrix :data result))) + + +(defmethod stack%2 ((d2 matrix) (d1 column)) + (let ((result (stack%2 (matrix-data d2) d1))) + (make-instance 'matrix :data result))) + + +;;; slice -- +#| +(defmethod slice ((v vector) (x fixnum)) + (aref v x)) + + +(defmethod slice ((v vector) (sd list)) + (assert (<= (list-length sd) 2)) + (destructuring-bind (start &optional end) + sd + (let ((start (cond ((or (eq start '*) (eq start 'cl:*)) 0) + ((typep start `(mod ,array-total-size-limit)) start) + (t (error "Unrecognized start ~S in slice designator." + start)))) + (end (cond ((or (null end) (eq end '*) (eq end 'cl:*)) nil) + ((typep end `(mod ,array-total-size-limit)) end) + (t (error "Unrecognized end ~S in slice designator." + end)))) + ) + (subseq v start end)))) + + +(defmethod slice ((v column) (sd list)) + (column (slice (column-vector v) sd))) + + +(defmethod slice ((v vector) (sd (eql '*))) + (copy-vector v)) + + +(defmethod slice ((v vector) (sd (eql 'cl:*))) + (copy-vector v)) + + +(defmethod slice ((v column) (sd (eql '*))) + (copy-vector v)) + + +(defmethod slice ((v column) (sd (eql 'cl:*))) + (copy-vector v)) + + +(defmethod slice (())) + |# + + + +;;; ref -- + +(defmethod ref ((v vector) (i fixnum) &rest indexes) + (when indexes + (error "REF of a vector requires only one index but ~D were supplied." + (1+ (list-length indexes)))) + + + (aref v i)) + + +(defmethod ref ((v vector) (i increasing-range) &rest indexes) + (when indexes + (error "REF of a vector requires only one index or range but a range and ~D indexes were supplied." + (list-length indexes))) + + (loop for vi from (range-start i) + below (or (range-end i) (length v)) + by (range-step i) + collect (aref v vi) into vs + finally (return (coerce vs 'vector)))) + + +(defmethod ref ((v vector) (i decreasing-range) &rest indexes) + (when indexes + (error "REF of a vector requires only one index or range but a range and ~D indexes were supplied." + (list-length indexes))) + + (loop for vi from (or (range-end i) (1- (length v))) + downto (range-start i) + by (abs (range-step i)) + collect (aref v vi) into vs + finally (return (coerce vs 'vector)))) + + +(defmethod ref ((v vector) (i vector) &rest indexes) + (when indexes + (error "REF of a vector requires only one index or subset but a subset and ~D indexes were supplied." + (list-length indexes))) + + (loop for x across i + collect (aref v x) into vs + finally (return (coerce vs 'vector)))) + + +(defmethod ref ((v vector) (i list) &rest indexes) + (when indexes + (error "REF of a vector requires only one index or subset but a subset and ~D indexes were supplied." + (list-length indexes))) + + (loop for x in i + collect (aref v x) into vs + finally (return (coerce vs 'vector)))) + + +(defmethod ref ((v column) (i fixnum) &rest indexes) + (apply #'ref (column-vector v) i indexes)) + + +(defmethod ref ((v column) (i range) &rest indexes) + (column (apply #'ref (column-vector v) i indexes))) + + +(defmethod ref ((v column) (i vector) &rest indexes) + (column (apply #'ref (column-vector v) i indexes))) + + +(defmethod ref ((v column) (i list) &rest indexes) + (column (apply #'ref (column-vector v) i indexes))) + + +(defmethod ref ((m matrix) (i t) &rest indexes) + (when (cl:/= 1 (list-length indexes)) + (error "........")) + + (ref%2 m i (first indexes))) + + + +(defmethod ref%2 ((m matrix) (i fixnum) (j fixnum)) + (aref (matrix-data m) i j)) + + +(defmethod ref%2 ((m matrix) (i fixnum) (j vector)) + (loop with md = (matrix-data m) + for x across j + collecting (aref md i x) into result + finally return (coerce result 'vector))) + + +(defmethod ref%2 ((m matrix) (i vector) (j fixnum)) + (loop with md = (matrix-data m) + for x across i + collecting (aref md x j) into result + finally return (coerce result 'vector))) + + +(defmethod ref%2 ((m matrix) (i vector) (j vector)) + (loop with md = (matrix-data m) + for x across i + collecting (loop for y across j + collecting (aref md x y)) + into result + finally + (return (make-instance 'matrix + :data (make-array (list (length i) + (length j)) + :initial-contents result))))) + + + + + +;;;;=========================================================================== +;;;; Generic Interface Functions. + +(defun join (&rest elements) + (cl:reduce #'join%2 elements)) + + +(defun stack (&rest elements) + (cl:reduce #'stack%2 elements)) + +;;;; end of file -- vector-matrix.lisp -- Added: trunk/common-math/numerics/linear-algebra/vector.lisp ============================================================================== --- (empty file) +++ trunk/common-math/numerics/linear-algebra/vector.lisp Wed Aug 16 17:07:41 2006 @@ -0,0 +1,461 @@ +;;;; -*- Mode: Lisp -*- + +;;;; vector.lisp -- +;;;; There is no wrapper for vectors, but we do have a class (struct) +;;;; for 'column' vectors. +;;;; Apart from that there are some specialized generic functions on +;;;; vectors. + +(in-package "CL.MATH.LINEAR-ALGEBRA") + +(defparameter *default-vector-element-type* 'double-float) + + +(deftype column-array (&optional (type 'cl:number) (length 'cl:*)) + `(array ,type (,length 1))) + + +(defstruct (column (:constructor column (vector))) + "The Column Structure. +A representation for `column' vectors." + (vector #() :type (or list vector))) + + +(defmethod print-object ((c column) (s stream)) + (format s "(~S ~S)" 'column (column-vector c))) + + +;;;--------------------------------------------------------------------------- +;;; Protocol. + +(defgeneric vector-p (v) + (:method ((v vector)) t) + (:method ((v column)) t) + (:method ((v t)) nil) + ) + + +(defgeneric vector-element-type (v) + (:method ((v vector)) (array-element-type v)) + (:method ((v column)) (array-element-type (column-vector v))) + ) + + +(defgeneric vector-length (m) + (:method ((x vector)) (length x)) + (:method ((x column)) (length (column-vector x))) + ) + + +;;;--------------------------------------------------------------------------- +;;; Creation. + +(defun make-vector (n + &rest array-keys + &key + (element-type *default-vector-element-type*) + (column-p nil) + &allow-other-keys + ) + (remf array-keys :column-p) + (let ((new-v (apply #'make-array n + (append array-keys + (list :initial-element (coerce 0 element-type) + :element-type + element-type))))) + (if column-p + (column new-v) + new-v))) + + +(defun make-zero-vector (n + &rest array-keys + &key + (element-type *default-matrix-element-type*) + &allow-other-keys) + (when (getf array-keys :initial-element) + (warn "MAKE-ZERO-MATRIX passed and :INITIAL-ELEMENT argument.") + (remf array-keys :initial-element)) + (apply #'make-vector + n + (append array-keys + (list :initial-element (coerce 0 element-type) + :element-type element-type)))) + + +;;;--------------------------------------------------------------------------- +;;; copy-vector +;;; Uses the code from K Pitman and B. Margolin appeared on CLL a long +;;; long time ago. +;;; Note that at least on LW this code has some of the usual problems +;;; with DOUBLE-FLOATS. +#| +(defun copy-array (array) + (adjust-array (make-array (array-dimensions array) + :displaced-to array + :element-type (array-element-type array)) + (array-dimensions array) + :displaced-to nil)) +|# + +(defgeneric copy-vector (v) + (:documentation "Returns a (shallow) copy of vector V. +The effect are the same as CL:COPY-SEQ for CL vectors while columns +are treated accordingly.")) + + +(defmethod copy-vector ((v vector)) + "COPY-VECTOR method for VECTORs." + (cl:copy-seq v)) + + +(defmethod copy-vector ((v column)) + "COPY-VECTOR method for COLUMN vectors." + (column (cl:copy-seq (column-vector v)))) + + +(defmethod copy-vector ((v array)) + (check-type v column-array) + (adjust-array (make-array (array-dimensions v) + :displaced-to v + :element-type (array-element-type v)) + (array-dimensions v) + :displaced-to nil)) + + + +#| +(defmethod copy-vector ((v vector) + &optional + (result + (adjust-array + (make-array (length v) + :displaced-to v + :element-type (array-element-type v)) + (array-dimensions v) + :displaced-to nil) + result-supplied-p)) + "COPY-VECTOR method for VECTORs." + (when result-supplied-p + (assert (vector-p result)) + (assert (cl:= (length result) (vector-length v))) + (map-into result #'identity v)) + result) + + +(defmethod copy-vector ((c column) + &optional + (result + (let ((v (column-vector c))) + (column + (adjust-array + (make-array (length v) + :displaced-to v + :element-type (array-element-type v)) + (array-dimensions v) + :displaced-to nil))) + result-supplied-p)) + "COPY-VECTOR method for COLUMN vectors." + (when result-supplied-p + (assert (column-p result)) + (assert (cl:= (vector-length result) (vector-length c))) + (map-into (column-vector result) #'identity (column-vector c))) + result) +|# + + + +;;;--------------------------------------------------------------------------- +;;; Equality. + +(defmethod =%2 ((x vector) (y vector)) + (and (cl:= (length x) (length y)) + (every #'=%2 x y))) + + +(defmethod =%2 ((x vector) (y column)) + (and (or (= 1 (length x) (vector-length y)) ; Only cases when this may happen. + (= 0 (length x) (vector-length y))) + (=%2 x (column-vector y)))) + + +(defmethod =%2 ((y column) (x vector)) + (=%2 x y)) + + +;;;--------------------------------------------------------------------------- +;;; zerop + +(defmethod zerop ((x vector)) + (cl:every #'zerop x)) + + +(defmethod zerop ((x column)) + (cl:every #'zerop (column-vector x))) + + +;;;--------------------------------------------------------------------------- +;;; Addition. + +(defmethod +%2 :before ((x vector) (y vector) + &optional + (r nil r-supplied-p)) + (unless (eq y r) + (let ((xl (length x))) + (assert (cl:= xl (length y))) + (when r-supplied-p + (assert (vector-p r)) + (assert (cl:= xl (vector-length r))) + )))) + + +(defmethod +%2 :before ((x vector) (y column) + &optional + (r nil r-supplied-p)) + (declare (ignore r r-supplied-p)) + (assert (or (cl:= 1 (length x) (vector-length y)) + (cl:= 0 (length x) (vector-length y)))) + ) + + +(defmethod +%2 :before ((y column) (x vector) + &optional + (r nil r-supplied-p)) + (declare (ignore r r-supplied-p)) + (assert (or (cl:= 1 (length x) (vector-length y)) + (cl:= 0 (length x) (vector-length y)))) + ) + + +(defmethod +%2 :before ((x number) (y vector) + &optional + (r nil r-supplied-p)) + (unless (eq y r) + (when r-supplied-p + (assert (vector-p r)) + (assert (cl:= (length y) (vector-length r))) + ))) + + +(defmethod +%2 ((x number) (y vector) &optional (r (copy-vector y))) + (dotimes (i (length y) r) + (setf (aref r i) (+%2 x (aref y i))))) + + +(defmethod +%2 ((y vector) (x number) &optional (r (copy-vector y))) + (+%2 x y r)) + + +(defmethod +%2 ((x vector) (y vector) &optional (r (copy-vector y))) + (dotimes (i (length x) r) + (setf (aref r i) (+%2 (aref x i) (aref y i))))) + + +(defmethod +%2 ((x vector) (y column) + &optional (r (copy-vector (column-vector y)))) + ;; This is valid only for vectors and columns of length 1 (or 0). + (+%2 x (column-vector y) r)) + + +(defmethod +%2 ((x column) (y vector) &optional (r (copy-vector y))) + ;; This is valid only for vectors and columns of length 1 (or 0). + (+%2 (column-vector x) y r)) + + +;;;--------------------------------------------------------------------------- +;;; Subtraction. + +(defmethod -%2 :before ((x vector) (y vector) + &optional + (r nil r-supplied-p)) + (let ((xl (length x))) + (assert (cl:= xl (length y))) + (when r-supplied-p + (assert (vector-p r)) + (assert (cl:= xl (vector-length r))) + ))) + + +(defmethod -%2 :before ((x vector) (y column) + &optional + (r nil r-supplied-p)) + (declare (ignore r r-supplied-p)) + (assert (or (cl:= 1 (length x) (vector-length y)) + (cl:= 0 (length x) (vector-length y)))) + ) + + +(defmethod -%2 :before ((y column) (x vector) + &optional + (r nil r-supplied-p)) + (declare (ignore r r-supplied-p)) + (assert (or (cl:= 1 (length x) (vector-length y)) + (cl:= 0 (length x) (vector-length y)))) + ) + + +(defmethod -%2 :before ((x number) (y vector) + &optional + (r nil r-supplied-p)) + (unless (eq y r) + (when r-supplied-p + (assert (vector-p r)) + (assert (cl:= (length y) (vector-length r))) + ))) + + +(defmethod -%2 ((x vector) (y vector) &optional (r (copy-vector y))) + (dotimes (i (length x) r) + (setf (aref r i) (-%2 (aref x i) (aref y i))))) + + + +(defmethod -%2 ((x vector) (y column) + &optional (r (copy-vector (column-vector y)))) + ;; This is valid only for vectors and columns of length 1 (or 0). + (-%2 x (column-vector y) r)) + + +(defmethod -%2 ((x column) (y vector) &optional (r (copy-vector y))) + ;; This is valid only for vectors and columns of length 1 (or 0). + (-%2 (column-vector x) r r)) + + +;;;--------------------------------------------------------------------------- +;;; Multiplication. +;;; Note that left and right multiplications are different! +;;; They follow the matrix multiplication rules. +;;; Here we treat only the vector x column multiplication as the +;;; column x vector has a matrix as a result. + +(defmethod *%2 :before ((x vector) (y column) &optional r) + (declare (ignore r)) + (assert (cl:= (cl:length x) (vector-length y)))) + + +(defmethod *%2 ((x vector) (y column) &optional r) + (declare (ignore r)) + (let ((result (coerce 0 (array-element-type x))) + (y (column-vector y)) + ) + (dotimes (i (cl:length x) result) + (setf result (+%2 result (*%2 (aref x i) (aref y i))))) + )) + +;;; The next one breaks the correct semantics passing of arguments, +;;; but it is provided as a convenience. + +(defmethod *%2 ((x vector) (y vector) &optional r) + (declare (ignore r)) + (assert (cl:= (cl:length x) (cl:length y))) + (let ((result (coerce 0 (array-element-type x)))) + (dotimes (i (cl:length x) result) + (setf result (+%2 result (*%2 (aref x i) (aref y i))))))) + + +;;; Scalar multiplication. + +(defmethod *%2 :before ((x number) (v vector) &optional (r nil r-supplied-p)) + (when r-supplied-p + (assert (cl:vectorp r)) + (assert (cl:= (cl:length v) (vector-length r))))) + + +(defmethod *%2 :before ((x number) (v column) &optional (r nil r-supplied-p)) + (when r-supplied-p + (assert (column-p r)) + (assert (cl:= (vector-length v) (vector-length r))))) + + +(defmethod *%2 ((x number) (v vector) &optional (r (copy-vector v))) + (map-into r (lambda (e) (*%2 x e)) v)) + + +(defmethod *%2 ((v vector) (x number) &optional (r (copy-vector v))) + (map-into r (lambda (e) (*%2 x e)) v)) + + +(defmethod *%2 ((x number) (v column) &optional (r (copy-vector v))) + (map-into (column-vector r) (lambda (e) (*%2 x e)) v) + r) + + +(defmethod *%2 ((v column) (x number) &optional (r (copy-vector v))) + (*%2 x v r)) + + +;;;--------------------------------------------------------------------------- +;;; Division. +;;; Only the simple form of division is implemented here. +;;; The general form requires the inverse, which requires extra machinery + +#| +(defmethod /%2 ((x number) (y matrix) &optional (r (copy-matrix y) r-supplied-p)) + (declare (ignore r-supplied-p)) + (*%2 (/ x) y r)) + + +(defmethod /%2 ((y matrix) (x number) &optional (r (copy-matrix y) r-supplied-p)) + (declare (ignore r-supplied-p)) + (*%2 (/ x) y r)) +|# + +;;;--------------------------------------------------------------------------- +;;; Exponentiation. +;;; All errors here ftb. + + +;;;--------------------------------------------------------------------------- +;;; Transpose + +(defmethod transpose ((v vector) &optional (r (column (copy-vector v)) r-supplied-p)) + (if r-supplied-p + (map-into (column-vector r) #'identity v)) + r) + + +(defmethod transpose ((v column) + &optional + (r (copy-vector (column-vector v)) r-supplied-p)) + (if r-supplied-p + (map-into r #'identity (column-vector v)) + r)) + + +(defmethod transpose :before ((v vector) &optional (r nil r-supplied-p)) + (when r-supplied-p + (assert (column-p r)) + (assert (cl:= (cl:length v) (vector-length r))))) + + +(defmethod transpose :before ((v column) &optional (r nil r-supplied-p)) + (when r-supplied-p + (assert (cl:vectorp r)) + (assert (cl:= (cl:length r) (vector-length v))))) + + +;;;--------------------------------------------------------------------------- +;;; Other functions. + +(defun iota (n &optional (start 0) (step 1 step-supplied-p) + &aux (initial-contents ())) + (cond ((< n start) + (when step-supplied-p (assert (minusp step))) + (loop for x from start above n by step + collect x into xs + finally (setf initial-contents xs))) + ((> n start) + (when step-supplied-p (assert (plusp step))) + (loop for x from start below n by step + collect x into xs + finally (setf initial-contents xs))) + (t ())) + (make-array n :initial-contents initial-contents)) + + + +;;;--------------------------------------------------------------------------- +;;; Utilities. + + +;;;; end of file -- matrix.lisp -- Added: trunk/common-math/numerics/numerics.system ============================================================================== --- (empty file) +++ trunk/common-math/numerics/numerics.system Wed Aug 16 17:07:41 2006 @@ -0,0 +1,18 @@ +;;;; -*- Mode: Lisp -*- + +;;;; numerics.system -- + +(mk:defsystem "numerics" + :components ((:system "linear-algebra") + ) + ) + + +(eval-when (:load-toplevel :execute) + (mk:add-registry-location (make-pathname :name nil + :type nil + :defaults *load-truename*)) + ) + +;;;; end of file -- numerics.system -- +

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

common-math-cvs